[Music] cool I’ll get started I’m Diego I work
cool I’ll get started I’m Diego I work
cool I’ll get started I’m Diego I work at Google brain but that’s all I can
at Google brain but that’s all I can
at Google brain but that’s all I can tell you my normal rule is if I told you
tell you my normal rule is if I told you
tell you my normal rule is if I told you I’d have to kill you but there’s a whole
I’d have to kill you but there’s a whole
I’d have to kill you but there’s a whole lot of you guys which makes that kind of
lot of you guys which makes that kind of
lot of you guys which makes that kind of impractical right now a standard
impractical right now a standard
impractical right now a standard disclaimer everything I say reflects my
disclaimer everything I say reflects my
disclaimer everything I say reflects my own opinions it’s not representative of
own opinions it’s not representative of
own opinions it’s not representative of my employer I have not been at Google
my employer I have not been at Google
my employer I have not been at Google long enough to know any secrets though I
long enough to know any secrets though I
long enough to know any secrets though I will admit that I’ve pilfered some
will admit that I’ve pilfered some
will admit that I’ve pilfered some publicly available slides so if there is
publicly available slides so if there is
publicly available slides so if there is a Google bias it’s not because they made
a Google bias it’s not because they made
a Google bias it’s not because they made me do it it’s because I’m lazy as far as
me do it it’s because I’m lazy as far as
me do it it’s because I’m lazy as far as some background for myself
some background for myself
some background for myself I broke a 13-year losing streak for the
I broke a 13-year losing streak for the
I broke a 13-year losing streak for the Philippines in the International math
Philippines in the International math
Philippines in the International math Olympiad I got the top prize in the
Olympiad I got the top prize in the
Olympiad I got the top prize in the world in the inter display compass in
world in the inter display compass in
world in the inter display compass in modeling and if you’re familiar with
modeling and if you’re familiar with
modeling and if you’re familiar with kayo competitions I also want one of
kayo competitions I also want one of
kayo competitions I also want one of those so I like to compete and do all of
those so I like to compete and do all of
those so I like to compete and do all of these kinds of things and hopefully that
these kinds of things and hopefully that
these kinds of things and hopefully that convinces you that I know that I’m
convinces you that I know that I’m
convinces you that I know that I’m talking about but let’s get started the
talking about but let’s get started the
talking about but let’s get started the presentation is deep learning what is it
presentation is deep learning what is it
presentation is deep learning what is it and what can it do for you
and what can it do for you
and what can it do for you but I think the very first question is
but I think the very first question is
but I think the very first question is why should I care and that reminds me of
why should I care and that reminds me of
why should I care and that reminds me of a story a machine learning researcher a
a story a machine learning researcher a
a story a machine learning researcher a cryptocurrency expert and an Erlang
cryptocurrency expert and an Erlang
cryptocurrency expert and an Erlang programmer walk into a bar
programmer walk into a bar
programmer walk into a bar Facebook buys the bar for twenty seven
Facebook buys the bar for twenty seven
Facebook buys the bar for twenty seven billion dollars and also another
billion dollars and also another
billion dollars and also another disclaimer you may not know this but I’m
disclaimer you may not know this but I’m
disclaimer you may not know this but I’m both a machine learning researcher and
both a machine learning researcher and
both a machine learning researcher and from San Francisco that means all of my
from San Francisco that means all of my
from San Francisco that means all of my information comes from Twitter that’s
information comes from Twitter that’s
information comes from Twitter that’s not a joke so prepare for that for my
not a joke so prepare for that for my
not a joke so prepare for that for my slides but back to why you should care
slides but back to why you should care
slides but back to why you should care machine learning artificial intelligence
machine learning artificial intelligence
machine learning artificial intelligence deep learning they’re all getting a lot
deep learning they’re all getting a lot
deep learning they’re all getting a lot of press these days they’re all doing
of press these days they’re all doing
of press these days they’re all doing lots of stuff and there’s lots of hype
lots of stuff and there’s lots of hype
lots of stuff and there’s lots of hype lots of news articles about all sorts of
lots of news articles about all sorts of
lots of news articles about all sorts of things and everyone seems to want to get
things and everyone seems to want to get
things and everyone seems to want to get into it but people don’t really know
into it but people don’t really know
into it but people don’t really know what they’re talking about it seems the
what they’re talking about it seems the
what they’re talking about it seems the only thing everyone’s really sure of is
only thing everyone’s really sure of is
only thing everyone’s really sure of is that artificial intelligence and it
that artificial intelligence and it
that artificial intelligence and it seems like very recently in particular
seems like very recently in particular
seems like very recently in particular deep learning will be a catalyst for a
deep learning will be a catalyst for a
deep learning will be a catalyst for a lot of change that’s happening and
lot of change that’s happening and
lot of change that’s happening and people are asking any questions all the
people are asking any questions all the
people are asking any questions all the time about this kind of thing and some
time about this kind of thing and some
time about this kind of thing and some of these recurring themes are how will
of these recurring themes are how will
of these recurring themes are how will the world change
the world change
the world change what can a AI and in particular deep
what can a AI and in particular deep
what can a AI and in particular deep learning do and how do I take advantage
learning do and how do I take advantage
learning do and how do I take advantage of these trends
of these trends
of these trends even the legendary programmer Jeff Dean
even the legendary programmer Jeff Dean
even the legendary programmer Jeff Dean has said if you’re not considering how
has said if you’re not considering how
has said if you’re not considering how to use deep neural nets to solve your
to use deep neural nets to solve your
to use deep neural nets to solve your problems you almost certainly should be
problems you almost certainly should be
problems you almost certainly should be it almost sounds like a threat either
it almost sounds like a threat either
it almost sounds like a threat either way I hope you’re motivated to learn
way I hope you’re motivated to learn
way I hope you’re motivated to learn that’s all I have for motivation so
that’s all I have for motivation so
that’s all I have for motivation so let’s get started into what is deep
let’s get started into what is deep
let’s get started into what is deep learning and it’s quite easy this is
learning and it’s quite easy this is
learning and it’s quite easy this is deep learning you could memorize this
deep learning you could memorize this
deep learning you could memorize this diagram this will make the presentation
diagram this will make the presentation
diagram this will make the presentation a lot easier so that’s basically it just
a lot easier so that’s basically it just
a lot easier so that’s basically it just kidding this is neither complete there’s
kidding this is neither complete there’s
kidding this is neither complete there’s a lot more to it than that and it’s also
a lot more to it than that and it’s also
a lot more to it than that and it’s also pretty complicated we’re gonna start
pretty complicated we’re gonna start
pretty complicated we’re gonna start with something much more simple namely
with something much more simple namely
with something much more simple namely calculus may not sound right it’s even
calculus may not sound right it’s even
calculus may not sound right it’s even chapter 1 of the book calculus made easy
chapter 1 of the book calculus made easy
chapter 1 of the book calculus made easy it’s titled to deliver you from the
it’s titled to deliver you from the
it’s titled to deliver you from the preliminary terrors but we only need a
preliminary terrors but we only need a
preliminary terrors but we only need a little bit of calculus and particularly
little bit of calculus and particularly
little bit of calculus and particularly we need an algorithm called gradient
we need an algorithm called gradient
we need an algorithm called gradient descent the derivative of one thing with
descent the derivative of one thing with
descent the derivative of one thing with calculus tells you is how to take
calculus tells you is how to take
calculus tells you is how to take derivatives and a derivative loosely
derivatives and a derivative loosely
derivatives and a derivative loosely speaking if it tells you how a functions
speaking if it tells you how a functions
speaking if it tells you how a functions output changes when you change its input
output changes when you change its input
output changes when you change its input and gradient descent is just moving
and gradient descent is just moving
and gradient descent is just moving along the direction of the derivative in
along the direction of the derivative in
along the direction of the derivative in order to minimize a function that you
order to minimize a function that you
order to minimize a function that you can take the derivative of so the
can take the derivative of so the
can take the derivative of so the insight into all of machinery Sigma Xin
insight into all of machinery Sigma Xin
insight into all of machinery Sigma Xin learning is figure out how to frame your
learning is figure out how to frame your
learning is figure out how to frame your problem in such a way that what you care
problem in such a way that what you care
problem in such a way that what you care about is differentiable or a proxy of
about is differentiable or a proxy of
about is differentiable or a proxy of what you care about is differentiable
what you care about is differentiable
what you care about is differentiable and then minimize it and this like one
and then minimize it and this like one
and then minimize it and this like one extremely simple equation summarizes
extremely simple equation summarizes
extremely simple equation summarizes almost all of the recent work in machine
almost all of the recent work in machine
almost all of the recent work in machine learning that’s happened in the last
learning that’s happened in the last
learning that’s happened in the last half decade roughly there’s obviously
half decade roughly there’s obviously
half decade roughly there’s obviously exceptions to this rule but majority of
exceptions to this rule but majority of
exceptions to this rule but majority of what’s done either has been done with
what’s done either has been done with
what’s done either has been done with this exceptionally simple thing or can
this exceptionally simple thing or can
this exceptionally simple thing or can be done with this exceptionally simple
be done with this exceptionally simple
be done with this exceptionally simple simple thing and that’s basically it as
simple thing and that’s basically it as
simple thing and that’s basically it as far as what deep learning really is in
far as what deep learning really is in
far as what deep learning really is in its essence
its essence
its essence this might sound well good but this is
this might sound well good but this is
this might sound well good but this is just machine learning when’s it become
just machine learning when’s it become
just machine learning when’s it become deep learning also easy it’s when you
deep learning also easy it’s when you
deep learning also easy it’s when you make it really deep it might sound like
make it really deep it might sound like
make it really deep it might sound like a joke but acts
a joke but acts
a joke but acts really what happens when you have these
really what happens when you have these
really what happens when you have these you know these machine learning used to
you know these machine learning used to
you know these machine learning used to be composed these very very simple
be composed these very very simple
be composed these very very simple functions because this is all we knew
functions because this is all we knew
functions because this is all we knew how to optimize and what happens when
how to optimize and what happens when
how to optimize and what happens when you stack multiple of these simple
you stack multiple of these simple
you stack multiple of these simple functions together you get something
functions together you get something
functions together you get something that’s much much more powerful if we
that’s much much more powerful if we
that’s much much more powerful if we don’t really know how much more powerful
don’t really know how much more powerful
don’t really know how much more powerful it is some might claim it’s
it is some might claim it’s
it is some might claim it’s exponentially more powerful but either
exponentially more powerful but either
exponentially more powerful but either way we know it’s much more powerful and
way we know it’s much more powerful and
way we know it’s much more powerful and simply stacking these things and using
simply stacking these things and using
simply stacking these things and using the simple algorithm is what’s caused
the simple algorithm is what’s caused
the simple algorithm is what’s caused the deep learning revolution to hit and
the deep learning revolution to hit and
the deep learning revolution to hit and it’s just using same old simple
it’s just using same old simple
it’s just using same old simple algorithm even though as a caveat to
algorithm even though as a caveat to
algorithm even though as a caveat to that we part of the hard part of deep
that we part of the hard part of deep
that we part of the hard part of deep learning is knowing that the simple
learning is knowing that the simple
learning is knowing that the simple algorithm will work for these very
algorithm will work for these very
algorithm will work for these very complicated models that have like stacks
complicated models that have like stacks
complicated models that have like stacks of layers and making problems non convex
of layers and making problems non convex
of layers and making problems non convex is that it for deep learning yes this is
is that it for deep learning yes this is
is that it for deep learning yes this is it roughly I’m skipping all sorts of
it roughly I’m skipping all sorts of
it roughly I’m skipping all sorts of details that I’m sure will be covered
details that I’m sure will be covered
details that I’m sure will be covered later from software that makes things
later from software that makes things
later from software that makes things easier to write like tensorflow or pi
easier to write like tensorflow or pi
easier to write like tensorflow or pi torch to hardware that makes things
torch to hardware that makes things
torch to hardware that makes things faster to run like GPUs or multi-core
faster to run like GPUs or multi-core
faster to run like GPUs or multi-core CPUs or TP use to commonly use some
CPUs or TP use to commonly use some
CPUs or TP use to commonly use some functions or layers that people have
functions or layers that people have
functions or layers that people have done lots of trial and error and just
done lots of trial and error and just
done lots of trial and error and just found to work well for some of today’s
found to work well for some of today’s
found to work well for some of today’s problems without very good justification
problems without very good justification
problems without very good justification and also commonly use combinations of
and also commonly use combinations of
and also commonly use combinations of these layers or architectures that
these layers or architectures that
these layers or architectures that similarly people have used trial and
similarly people have used trial and
similarly people have used trial and error and found to work without very
error and found to work without very
error and found to work without very good justification but for the purpose
good justification but for the purpose
good justification but for the purpose of that talk the simpler algorithm is it
of that talk the simpler algorithm is it
of that talk the simpler algorithm is it next big question is what can it do for
next big question is what can it do for
next big question is what can it do for you
you
you this may there was a recent paper that
this may there was a recent paper that
this may there was a recent paper that came out that makes it a little bit
came out that makes it a little bit
came out that makes it a little bit easier to answer because this article
easier to answer because this article
easier to answer because this article came out that surveyed a I assume lot of
came out that surveyed a I assume lot of
came out that surveyed a I assume lot of machine learning researchers all of
machine learning researchers all of
machine learning researchers all of these Alliance are people so and there’s
these Alliance are people so and there’s
these Alliance are people so and there’s a survey on the future progress on AI so
a survey on the future progress on AI so
a survey on the future progress on AI so what you think will happen when will
what you think will happen when will
what you think will happen when will happen and there’s a lot of really
happen and there’s a lot of really
happen and there’s a lot of really interesting things here that are amusing
interesting things here that are amusing
interesting things here that are amusing and possibly informative may be mostly
and possibly informative may be mostly
and possibly informative may be mostly amusing
amusing
amusing and let’s break it down on the easy end
and let’s break it down on the easy end
and let’s break it down on the easy end you see Angry Birds at roughly the same
you see Angry Birds at roughly the same
you see Angry Birds at roughly the same difficulty as the World Series of Poker
difficulty as the World Series of Poker
difficulty as the World Series of Poker that’s very unusual to me I thought
that’s very unusual to me I thought
that’s very unusual to me I thought Angry Birds was solved I’m pretty sure
Angry Birds was solved I’m pretty sure
Angry Birds was solved I’m pretty sure Angry Birds is solved and the World
Angry Birds is solved and the World
Angry Birds is solved and the World Series of Poker is actually sounds
Series of Poker is actually sounds
Series of Poker is actually sounds really hard but it’s it’s down there
really hard but it’s it’s down there
really hard but it’s it’s down there near the easy end on the difficult and I
near the easy end on the difficult and I
near the easy end on the difficult and I find it really interesting that AI
find it really interesting that AI
find it really interesting that AI researchers think it’s it’s like the
researchers think it’s it’s like the
researchers think it’s it’s like the second top there a researcher is
second top there a researcher is
second top there a researcher is significantly harder than math
significantly harder than math
significantly harder than math researcher seems like a 50 year gap
researcher seems like a 50 year gap
researcher seems like a 50 year gap between math research being solved an AI
between math research being solved an AI
between math research being solved an AI research being solved not saying that I
research being solved not saying that I
research being solved not saying that I agree or disagree it’s just interesting
agree or disagree it’s just interesting
agree or disagree it’s just interesting to point out it actually looks like the
to point out it actually looks like the
to point out it actually looks like the gap between AI researcher and math
gap between AI researcher and math
gap between AI researcher and math researcher is larger than the gap
researcher is larger than the gap
researcher is larger than the gap between math researcher and playing
between math researcher and playing
between math researcher and playing Angry Birds at a human level so I yeah I
Angry Birds at a human level so I yeah I
Angry Birds at a human level so I yeah I I don’t know if this reflects something
I don’t know if this reflects something
I don’t know if this reflects something about the field maybe that’s why there’s
about the field maybe that’s why there’s
about the field maybe that’s why there’s no good deep learning theory right now
no good deep learning theory right now
no good deep learning theory right now but who knows what’s going on but
but who knows what’s going on but
but who knows what’s going on but importantly for knowing what deep
importantly for knowing what deep
importantly for knowing what deep learning can do for us is there’s a lot
learning can do for us is there’s a lot
learning can do for us is there’s a lot of differing opinions on what’s going to
of differing opinions on what’s going to
of differing opinions on what’s going to happen and when it’s going to happen
happen and when it’s going to happen
happen and when it’s going to happen there’s some people who think that we
there’s some people who think that we
there’s some people who think that we will be getting generally AI in roughly
will be getting generally AI in roughly
will be getting generally AI in roughly 10-15 years and there’s people who think
10-15 years and there’s people who think
10-15 years and there’s people who think it’s over a hundred years away I’m
it’s over a hundred years away I’m
it’s over a hundred years away I’m definitely in the latter camp to clarify
definitely in the latter camp to clarify
definitely in the latter camp to clarify but this makes it seem like answering
but this makes it seem like answering
but this makes it seem like answering the question what can I do for you
the question what can I do for you
the question what can I do for you correctly be really hard because there’s
correctly be really hard because there’s
correctly be really hard because there’s just so many different opinions how do
just so many different opinions how do
just so many different opinions how do you really know what the correct answer
you really know what the correct answer
you really know what the correct answer is it’s a tough problem but luckily I’m
is it’s a tough problem but luckily I’m
is it’s a tough problem but luckily I’m the only one on stage so I can just say
the only one on stage so I can just say
the only one on stage so I can just say whatever I want and no one can disagree
whatever I want and no one can disagree
whatever I want and no one can disagree with me I guess you can disagree with me
with me I guess you can disagree with me
with me I guess you can disagree with me in the Q&A and it’ll be a good debate
in the Q&A and it’ll be a good debate
in the Q&A and it’ll be a good debate but I’m gonna say what I think on this
but I’m gonna say what I think on this
but I’m gonna say what I think on this and take it with a grain of salt doesn’t
and take it with a grain of salt doesn’t
and take it with a grain of salt doesn’t reflect me it does reflect me it doesn’t
reflect me it does reflect me it doesn’t
reflect me it does reflect me it doesn’t reflect Google doesn’t reflect anything
reflect Google doesn’t reflect anything
reflect Google doesn’t reflect anything else I do like to stand on the shoulders
else I do like to stand on the shoulders
else I do like to stand on the shoulders of giants though and I think there’s
of giants though and I think there’s
of giants though and I think there’s been some people who have said things
been some people who have said things
been some people who have said things that resonates a lot with me and like
that resonates a lot with me and like
that resonates a lot with me and like really like when they say some things
really like when they say some things
really like when they say some things really precisely I feel like that helps
really precisely I feel like that helps
really precisely I feel like that helps refine my thinking on the problem this
refine my thinking on the problem this
refine my thinking on the problem this is one that I don’t know if I agree with
is one that I don’t know if I agree with
is one that I don’t know if I agree with but it’s a really strong statement and
but it’s a really strong statement and
but it’s a really strong statement and it seems like it could be a pretty good
it seems like it could be a pretty good
it seems like it could be a pretty good heuristic this is by and ruing who is no
heuristic this is by and ruing who is no
heuristic this is by and ruing who is no longer the chief scientist at Baidu
longer the chief scientist at Baidu
longer the chief scientist at Baidu but he says if a typical person can do a
but he says if a typical person can do a
but he says if a typical person can do a mental task with less than one second of
mental task with less than one second of
mental task with less than one second of thought we can probably automate it
thought we can probably automate it
thought we can probably automate it using AI either now or in the near
using AI either now or in the near
using AI either now or in the near future I can’t think of that many
future I can’t think of that many
future I can’t think of that many counter arguments that don’t require
counter arguments that don’t require
counter arguments that don’t require like a lot of like very specific domain
like a lot of like very specific domain
like a lot of like very specific domain knowledge like maybe people who play
knowledge like maybe people who play
knowledge like maybe people who play games a lot can play those games really
games a lot can play those games really
games a lot can play those games really fast cuz they’ve practiced it well but
fast cuz they’ve practiced it well but
fast cuz they’ve practiced it well but roughly it seems like a pretty good
roughly it seems like a pretty good
roughly it seems like a pretty good heuristic and it’s a very like strong
heuristic and it’s a very like strong
heuristic and it’s a very like strong statement so maybe this is something
statement so maybe this is something
statement so maybe this is something that could guide you answering that
that could guide you answering that
that could guide you answering that question another thing that isn’t as
question another thing that isn’t as
question another thing that isn’t as specific but I think is very important
specific but I think is very important
specific but I think is very important to consider is that a lot of the deep
to consider is that a lot of the deep
to consider is that a lot of the deep learning successes today have been I
learning successes today have been I
learning successes today have been I haven’t used the word simple but I feel
haven’t used the word simple but I feel
haven’t used the word simple but I feel like they’re more simple memorization
like they’re more simple memorization
like they’re more simple memorization problems and not really thinking
problems and not really thinking
problems and not really thinking problems it’s always hard to really say
problems it’s always hard to really say
problems it’s always hard to really say what this thinking really means because
what this thinking really means because
what this thinking really means because that might be a moving goalpost of like
that might be a moving goalpost of like
that might be a moving goalpost of like of course if algorithm is not thinking
of course if algorithm is not thinking
of course if algorithm is not thinking it’s using an a-star algorithm or
it’s using an a-star algorithm or
it’s using an a-star algorithm or something while we might think that’s
something while we might think that’s
something while we might think that’s kind of like thinking but even in this
kind of like thinking but even in this
kind of like thinking but even in this case it seems like when you’re when a
case it seems like when you’re when a
case it seems like when you’re when a task requires multiple steps of
task requires multiple steps of
task requires multiple steps of reasoning where you can’t like use
reasoning where you can’t like use
reasoning where you can’t like use heuristics to jump all the way from
heuristics to jump all the way from
heuristics to jump all the way from input to output it seems the deep
input to output it seems the deep
input to output it seems the deep learning has not been very good at that
learning has not been very good at that
learning has not been very good at that especially not without a lot of help
especially not without a lot of help
especially not without a lot of help which leads me to my general rule which
which leads me to my general rule which
which leads me to my general rule which is deep learning is an appropriate tool
is deep learning is an appropriate tool
is deep learning is an appropriate tool for supervised direct pattern matching
for supervised direct pattern matching
for supervised direct pattern matching tasks bonus points if you can design
tasks bonus points if you can design
tasks bonus points if you can design priors that are particularly suited to
priors that are particularly suited to
priors that are particularly suited to your problem the the priors in this case
your problem the the priors in this case
your problem the the priors in this case are specific layers that are popular for
are specific layers that are popular for
are specific layers that are popular for certain tasks but we don’t have to get
certain tasks but we don’t have to get
certain tasks but we don’t have to get into that right now
into that right now
into that right now even though feel free to ask me in the
even though feel free to ask me in the
even though feel free to ask me in the Q&A but here when I say supervised I
Q&A but here when I say supervised I
Q&A but here when I say supervised I mean that we tell the model directly
mean that we tell the model directly
mean that we tell the model directly what the correct answer is so roughly a
what the correct answer is so roughly a
what the correct answer is so roughly a human or some other process figures out
human or some other process figures out
human or some other process figures out what the right answer is via some means
what the right answer is via some means
what the right answer is via some means and tells the model this is what you
and tells the model this is what you
and tells the model this is what you should be outputting next time there are
should be outputting next time there are
should be outputting next time there are there have been incredible successes
there have been incredible successes
there have been incredible successes using reinforcement learning which is
using reinforcement learning which is
using reinforcement learning which is not supervised especially in the
not supervised especially in the
not supervised especially in the game-playing domain so if you’ve seen
game-playing domain so if you’ve seen
game-playing domain so if you’ve seen deep Minds deep queue networks playing
deep Minds deep queue networks playing
deep Minds deep queue networks playing Atari or deep Minds alphago playing go
Atari or deep Minds alphago playing go
Atari or deep Minds alphago playing go those use a lot of reinforcement
those use a lot of reinforcement
those use a lot of reinforcement learning and if they definitely have in
learning and if they definitely have in
learning and if they definitely have in some successes but how I feel about
some successes but how I feel about
some successes but how I feel about reinforcement learning is that it can
reinforcement learning is that it can
reinforcement learning is that it can work but you don’t want to rely on it
work but you don’t want to rely on it
work but you don’t want to rely on it working and in the bay everyone hasn’t
working and in the bay everyone hasn’t
working and in the bay everyone hasn’t started up on everything and there’s
started up on everything and there’s
started up on everything and there’s been a lot of people who kind of have
been a lot of people who kind of have
been a lot of people who kind of have bet their companies on this is a
bet their companies on this is a
bet their companies on this is a reinforcement learning problem let us
reinforcement learning problem let us
reinforcement learning problem let us sell people on using deep reinforcement
sell people on using deep reinforcement
sell people on using deep reinforcement learning to get this working and ending
learning to get this working and ending
learning to get this working and ending up with vaporware and kind of a sad end
up with vaporware and kind of a sad end
up with vaporware and kind of a sad end to that supervise for it but as far as
to that supervise for it but as far as
to that supervise for it but as far as direct pattern matching goes this goes
direct pattern matching goes this goes
direct pattern matching goes this goes back to what I was saying earlier where
back to what I was saying earlier where
back to what I was saying earlier where you want simple relationships between
you want simple relationships between
you want simple relationships between the input and the output
the input and the output
the input and the output almost almost like that a fraction of
almost almost like that a fraction of
almost almost like that a fraction of the input directly maps to some fraction
the input directly maps to some fraction
the input directly maps to some fraction of the output in some sort of additive
of the output in some sort of additive
of the output in some sort of additive ish way it doesn’t have to be completely
ish way it doesn’t have to be completely
ish way it doesn’t have to be completely additive but usually having some easy
additive but usually having some easy
additive but usually having some easy mapping allows you to bootstrap the more
mapping allows you to bootstrap the more
mapping allows you to bootstrap the more complicated mappings and a lot of the
complicated mappings and a lot of the
complicated mappings and a lot of the more complicated mappings turn out to be
more complicated mappings turn out to be
more complicated mappings turn out to be like lots of little simple mappings
like lots of little simple mappings
like lots of little simple mappings composed together and this kind of thing
composed together and this kind of thing
composed together and this kind of thing seems to be how deep learning tel tends
seems to be how deep learning tel tends
seems to be how deep learning tel tends to work this is all very vague but I am
to work this is all very vague but I am
to work this is all very vague but I am about to talk about some specifics about
about to talk about some specifics about
about to talk about some specifics about where deep learning has succeeded and
where deep learning has succeeded and
where deep learning has succeeded and where it seems have not succeeded yet
where it seems have not succeeded yet
where it seems have not succeeded yet another disclaimer this is only a subset
another disclaimer this is only a subset
another disclaimer this is only a subset of the potential cool things to talk
of the potential cool things to talk
of the potential cool things to talk about and I’m only talking about the
about and I’m only talking about the
about and I’m only talking about the intersection of things I find
intersection of things I find
intersection of things I find interesting because I want the slides to
interesting because I want the slides to
interesting because I want the slides to be interesting there’s lots of like
be interesting there’s lots of like
be interesting there’s lots of like little things that are cool but maybe
little things that are cool but maybe
little things that are cool but maybe wouldn’t be that interesting to people
wouldn’t be that interesting to people
wouldn’t be that interesting to people and visual and things that I know how to
and visual and things that I know how to
and visual and things that I know how to put in a presentation so I I have some
put in a presentation so I I have some
put in a presentation so I I have some attempts at videos but they’re optional
attempts at videos but they’re optional
attempts at videos but they’re optional but I there’s some cool lots of cool
but I there’s some cool lots of cool
but I there’s some cool lots of cool working audio but I just have no idea
working audio but I just have no idea
working audio but I just have no idea how to put that in a presentation and
how to put that in a presentation and
how to put that in a presentation and [Music]
[Music]
[Music] yeah maybe that’s my bad but we can
yeah maybe that’s my bad but we can
yeah maybe that’s my bad but we can solve go and build robots but technology
solve go and build robots but technology
solve go and build robots but technology isn’t there yet for reliable audio and
isn’t there yet for reliable audio and
isn’t there yet for reliable audio and video this is also what happens when I
video this is also what happens when I
video this is also what happens when I make the graphics myself so all of the
make the graphics myself so all of the
make the graphics myself so all of the pretty animated graphics have been
pretty animated graphics have been
pretty animated graphics have been stolen from other Googlers who know how
stolen from other Googlers who know how
stolen from other Googlers who know how to do ours
to do ours
to do ours back on topic let’s start with the
back on topic let’s start with the
back on topic let’s start with the easiest thing whenever you have a metric
easiest thing whenever you have a metric
easiest thing whenever you have a metric that when that metric goes up money goes
that when that metric goes up money goes
that when that metric goes up money goes up you probably want to use machine
up you probably want to use machine
up you probably want to use machine learning possibly deep learning but
learning possibly deep learning but
learning possibly deep learning but definitely machine learning this is
definitely machine learning this is
definitely machine learning this is actually I would describe the main use
actually I would describe the main use
actually I would describe the main use case of machine learning
case of machine learning
case of machine learning I have knobs to turn some combinations
I have knobs to turn some combinations
I have knobs to turn some combinations of these knobs are better than others
of these knobs are better than others
of these knobs are better than others how do I turn them basic stuff but we’re
how do I turn them basic stuff but we’re
how do I turn them basic stuff but we’re saying a big thing that people get
saying a big thing that people get
saying a big thing that people get caught on is unsupervised learning it’s
caught on is unsupervised learning it’s
caught on is unsupervised learning it’s a very interesting research problem but
a very interesting research problem but
a very interesting research problem but if you want to do anything practical I
if you want to do anything practical I
if you want to do anything practical I would probably advise you to not do that
would probably advise you to not do that
would probably advise you to not do that I think this is actually absolutely
I think this is actually absolutely
I think this is actually absolutely excellent advice rather than trying to
excellent advice rather than trying to
excellent advice rather than trying to if you can spend a month figuring it out
if you can spend a month figuring it out
if you can spend a month figuring it out on supervised learning please do it that
on supervised learning please do it that
on supervised learning please do it that that will solve a lot of people a lot of
that will solve a lot of people a lot of
that will solve a lot of people a lot of time if you could spend a year that
time if you could spend a year that
time if you could spend a year that would probably save a lot of people out
would probably save a lot of people out
would probably save a lot of people out of time if you could do ten years you’re
of time if you could do ten years you’re
of time if you could do ten years you’re probably on track with the rest of the
probably on track with the rest of the
probably on track with the rest of the field so if you have a problem that you
field so if you have a problem that you
field so if you have a problem that you care about don’t try to do some magic
care about don’t try to do some magic
care about don’t try to do some magic where you don’t know if it’s gonna work
where you don’t know if it’s gonna work
where you don’t know if it’s gonna work label some data usually these things are
label some data usually these things are
label some data usually these things are a lot more data efficient and people say
a lot more data efficient and people say
a lot more data efficient and people say they are and sticking to supervised
they are and sticking to supervised
they are and sticking to supervised learning will be much easier for your
learning will be much easier for your
learning will be much easier for your sanity as well as your eventual impact
sanity as well as your eventual impact
sanity as well as your eventual impact speech recognition has done really well
speech recognition has done really well
speech recognition has done really well really really well people think that
really really well people think that
really really well people think that this is probably going to be one of the
this is probably going to be one of the
this is probably going to be one of the biggest changes to interfaces in not
biggest changes to interfaces in not
biggest changes to interfaces in not just our lifetimes but in the next
just our lifetimes but in the next
just our lifetimes but in the next decade right now people don’t like to
decade right now people don’t like to
decade right now people don’t like to talk in phones because they could be
talk in phones because they could be
talk in phones because they could be kind of sucks but people can talk much
kind of sucks but people can talk much
kind of sucks but people can talk much faster they can type and a lot of people
faster they can type and a lot of people
faster they can type and a lot of people don’t know how to type very well so this
don’t know how to type very well so this
don’t know how to type very well so this could completely change the way people
could completely change the way people
could completely change the way people interact with electronics things like
interact with electronics things like
interact with electronics things like Google glass or I hear it’s really big
Google glass or I hear it’s really big
Google glass or I hear it’s really big in China speech recognition there’s all
in China speech recognition there’s all
in China speech recognition there’s all sorts of things that this could enable
sorts of things that this could enable
sorts of things that this could enable and this is only a fraction of the cool
and this is only a fraction of the cool
and this is only a fraction of the cool things happening in audio but I not
things happening in audio but I not
things happening in audio but I not think to talk about that much but
think to talk about that much but
think to talk about that much but there’s things with generating
there’s things with generating
there’s things with generating generating audio generating music lots
generating audio generating music lots
generating audio generating music lots of cool stuff there translation this
of cool stuff there translation this
of cool stuff there translation this animation is really cool and this
animation is really cool and this
animation is really cool and this problem is really cool this is showing
problem is really cool this is showing
problem is really cool this is showing that not only can deep networks improve
that not only can deep networks improve
that not only can deep networks improve on like the traditional statistical mess
on like the traditional statistical mess
on like the traditional statistical mess that things like Google Translate used
that things like Google Translate used
that things like Google Translate used to do but where you just have matching
to do but where you just have matching
to do but where you just have matching corpuses or Corp I I don’t know what the
corpuses or Corp I I don’t know what the
corpuses or Corp I I don’t know what the plural is but you can also translate
plural is but you can also translate
plural is but you can also translate between language pairs that you’ve never
between language pairs that you’ve never
between language pairs that you’ve never you don’t even have matching corpuses on
you don’t even have matching corpuses on
you don’t even have matching corpuses on so in this example you have English to
so in this example you have English to
so in this example you have English to Japanese pairs as well as Japanese so
Japanese pairs as well as Japanese so
Japanese pairs as well as Japanese so sorry English to Japanese in English to
sorry English to Japanese in English to
sorry English to Japanese in English to Korean and using these networks you can
Korean and using these networks you can
Korean and using these networks you can actually translate directly between
actually translate directly between
actually translate directly between Korean and Japanese without ever seeing
Korean and Japanese without ever seeing
Korean and Japanese without ever seeing paired data between Korean and Japanese
paired data between Korean and Japanese
paired data between Korean and Japanese which is actually huge
which is actually huge
which is actually huge it could enable a lot of translation on
it could enable a lot of translation on
it could enable a lot of translation on languages that between languages that
languages that between languages that
languages that between languages that there’s just no data on and you can do
there’s just no data on and you can do
there’s just no data on and you can do it in a much more accurate way because
it in a much more accurate way because
it in a much more accurate way because you don’t need to translate into an
you don’t need to translate into an
you don’t need to translate into an intermediate language where you lose
intermediate language where you lose
intermediate language where you lose some information if you ever like do
some information if you ever like do
some information if you ever like do Google like what’s the game where you
Google like what’s the game where you
Google like what’s the game where you have a like a Markov chain with Google
have a like a Markov chain with Google
have a like a Markov chain with Google Translate you start with a thing you
Translate you start with a thing you
Translate you start with a thing you translate to one language you translate
translate to one language you translate
translate to one language you translate back eventually becomes garbage and
back eventually becomes garbage and
back eventually becomes garbage and nothing like the original thing you said
nothing like the original thing you said
nothing like the original thing you said and you just avoid that problem entirely
and you just avoid that problem entirely
and you just avoid that problem entirely with this image classification this is
with this image classification this is
with this image classification this is like the bread and butter of deep
like the bread and butter of deep
like the bread and butter of deep learning it’s what made deep learning a
learning it’s what made deep learning a
learning it’s what made deep learning a big deal people it was kind of a not
big deal people it was kind of a not
big deal people it was kind of a not mainstream thing until about two twenty
mainstream thing until about two twenty
mainstream thing until about two twenty twelve when deep learning one this
twelve when deep learning one this
twelve when deep learning one this imagenet competition and beat all of the
imagenet competition and beat all of the
imagenet competition and beat all of the other things by a fairly large margin
other things by a fairly large margin
other things by a fairly large margin and made everyone realize hey this
and made everyone realize hey this
and made everyone realize hey this solves problems that nothing else could
solves problems that nothing else could
solves problems that nothing else could solve before and there’s real-world
solve before and there’s real-world
solve before and there’s real-world applications to this like google photos
applications to this like google photos
applications to this like google photos an example there’s a lot of api’s where
an example there’s a lot of api’s where
an example there’s a lot of api’s where people have made a business of telling
people have made a business of telling
people have made a business of telling you what’s in an image people do face
you what’s in an image people do face
you what’s in an image people do face classification Faith’s
classification Faith’s
classification Faith’s detection there’s a lot of money and
detection there’s a lot of money and
detection there’s a lot of money and sentiment recognition you know like have
sentiment recognition you know like have
sentiment recognition you know like have a camera here and look at the room tell
a camera here and look at the room tell
a camera here and look at the room tell them if they’re enjoying the talk or not
them if they’re enjoying the talk or not
them if they’re enjoying the talk or not based on like people’s smiles and stuff
based on like people’s smiles and stuff
based on like people’s smiles and stuff maybe not for talks but like for ads and
maybe not for talks but like for ads and
maybe not for talks but like for ads and stuff something that can’t be done yet
stuff something that can’t be done yet
stuff something that can’t be done yet though is unbiased image classification
though is unbiased image classification
though is unbiased image classification or it’s still a lot of work this was a
or it’s still a lot of work this was a
or it’s still a lot of work this was a huge issue for Google photos actually
huge issue for Google photos actually
huge issue for Google photos actually like I think it’s like a few days after
like I think it’s like a few days after
like I think it’s like a few days after they released it people were complaining
they released it people were complaining
they released it people were complaining on Twitter that
on Twitter that
on Twitter that their friends were being classified as
their friends were being classified as
their friends were being classified as gorillas due to a lack of diversity in
gorillas due to a lack of diversity in
gorillas due to a lack of diversity in the training data and this is kind of
the training data and this is kind of
the training data and this is kind of unavoidable when you have imperfect
unavoidable when you have imperfect
unavoidable when you have imperfect datasets I actually don’t know how they
datasets I actually don’t know how they
datasets I actually don’t know how they solve this they might have just removed
solve this they might have just removed
solve this they might have just removed some of the classes that could be been
some of the classes that could be been
some of the classes that could be been taken as offensive but that’s just a
taken as offensive but that’s just a
taken as offensive but that’s just a hack right like we want like real
hack right like we want like real
hack right like we want like real algorithms that don’t make these kinds
algorithms that don’t make these kinds
algorithms that don’t make these kinds of stupid mistakes talking about not
of stupid mistakes talking about not
of stupid mistakes talking about not making stupid mistakes a problem near
making stupid mistakes a problem near
making stupid mistakes a problem near and dear to my heart is medical imaging
and dear to my heart is medical imaging
and dear to my heart is medical imaging they’ve been a bunch of huge successes
they’ve been a bunch of huge successes
they’ve been a bunch of huge successes on medical imaging in particular there’s
on medical imaging in particular there’s
on medical imaging in particular there’s been some really cool stuff done reading
been some really cool stuff done reading
been some really cool stuff done reading x-rays and CT scans cool stuff with
x-rays and CT scans cool stuff with
x-rays and CT scans cool stuff with segmenting pathology scans detecting
segmenting pathology scans detecting
segmenting pathology scans detecting diabetic retinopathy all of these things
diabetic retinopathy all of these things
diabetic retinopathy all of these things it’s people have been getting superhuman
it’s people have been getting superhuman
it’s people have been getting superhuman results like better than what seems to
results like better than what seems to
results like better than what seems to be the best doctors and hopefully very
be the best doctors and hopefully very
be the best doctors and hopefully very soon this kind of stuff will be like
soon this kind of stuff will be like
soon this kind of stuff will be like reaching the end users and helping
reaching the end users and helping
reaching the end users and helping people so this is a really exciting area
people so this is a really exciting area
people so this is a really exciting area of deep learning progress similar in
of deep learning progress similar in
of deep learning progress similar in that vein it’s not limited to either 2d
that vein it’s not limited to either 2d
that vein it’s not limited to either 2d images or having a single prediction per
images or having a single prediction per
images or having a single prediction per image you can do what’s called semantic
image you can do what’s called semantic
image you can do what’s called semantic segmentation where you label each pixel
segmentation where you label each pixel
segmentation where you label each pixel or in this case voxel in an image and
or in this case voxel in an image and
or in this case voxel in an image and you can also it also works for high
you can also it also works for high
you can also it also works for high dimensional data so this for example is
dimensional data so this for example is
dimensional data so this for example is 3d segmentation of I believe in neuron
3d segmentation of I believe in neuron
3d segmentation of I believe in neuron and this algorithm actually is iterative
and this algorithm actually is iterative
and this algorithm actually is iterative and how it like expands over time and
and how it like expands over time and
and how it like expands over time and this is very similar to how a human
this is very similar to how a human
this is very similar to how a human would segment a neuron it would not just
would segment a neuron it would not just
would segment a neuron it would not just say all those months here’s a neuron if
say all those months here’s a neuron if
say all those months here’s a neuron if it starts at something and being like
it starts at something and being like
it starts at something and being like okay this is close to this other thing
okay this is close to this other thing
okay this is close to this other thing this is maybe a neuron so we are as we
this is maybe a neuron so we are as we
this is maybe a neuron so we are as we were like expanding the reach of deep
were like expanding the reach of deep
were like expanding the reach of deep learning more people are designing more
learning more people are designing more
learning more people are designing more and more of these priors to build into
and more of these priors to build into
and more of these priors to build into the architectures to do much smarter
the architectures to do much smarter
the architectures to do much smarter things so whenever I say not yet on
things so whenever I say not yet on
things so whenever I say not yet on something it might be that technology is
something it might be that technology is
something it might be that technology is there P we just haven’t tried hard
there P we just haven’t tried hard
there P we just haven’t tried hard enough talking about and not yet there’s
enough talking about and not yet there’s
enough talking about and not yet there’s been some really cool work on image
been some really cool work on image
been some really cool work on image captioning so instead of given an image
captioning so instead of given an image
captioning so instead of given an image output a object in the image it’s given
output a object in the image it’s given
output a object in the image it’s given an image describe the image
an image describe the image
an image describe the image and this is a much harder task because
and this is a much harder task because
and this is a much harder task because there’s a lot of things that can go on
there’s a lot of things that can go on
there’s a lot of things that can go on in an image and there’s a lot of
in an image and there’s a lot of
in an image and there’s a lot of possible ways to describe an image so
possible ways to describe an image so
possible ways to describe an image so how do you say something is riot and
how do you say something is riot and
how do you say something is riot and what what set of things do you choose to
what what set of things do you choose to
what what set of things do you choose to have something described and this is
have something described and this is
have something described and this is pretty good like these descriptions are
pretty good like these descriptions are
pretty good like these descriptions are actually this is a good case there’s
actually this is a good case there’s
actually this is a good case there’s many bad cases of this but they still do
many bad cases of this but they still do
many bad cases of this but they still do make some really dumb mistakes it might
make some really dumb mistakes it might
make some really dumb mistakes it might reflect underlying issues with our
reflect underlying issues with our
reflect underlying issues with our imaging models or it might be due to
imaging models or it might be due to
imaging models or it might be due to dataset size but this is still an open
dataset size but this is still an open
dataset size but this is still an open research problem similarly to that it’s
research problem similarly to that it’s
research problem similarly to that it’s not very good at answering questions
not very good at answering questions
not very good at answering questions about images or stories it can be good
about images or stories it can be good
about images or stories it can be good at finding specific things in the images
at finding specific things in the images
at finding specific things in the images but there’s other things that seem to be
but there’s other things that seem to be
but there’s other things that seem to be easier than finding a thing or just as
easier than finding a thing or just as
easier than finding a thing or just as easy as finding a thing that deepening
easy as finding a thing that deepening
easy as finding a thing that deepening currently it’s not good at like counting
currently it’s not good at like counting
currently it’s not good at like counting if you ask good this is I don’t have a
if you ask good this is I don’t have a
if you ask good this is I don’t have a counting example here but if you have
counting example here but if you have
counting example here but if you have like a bowl of oranges and you ask like
like a bowl of oranges and you ask like
like a bowl of oranges and you ask like how many oranges are in this bowl this
how many oranges are in this bowl this
how many oranges are in this bowl this sounds like a very easy task but it’s
sounds like a very easy task but it’s
sounds like a very easy task but it’s quite hard for models right now so
quite hard for models right now so
quite hard for models right now so that’s a big problem talking about big
that’s a big problem talking about big
that’s a big problem talking about big problems we definitely are nowhere close
problems we definitely are nowhere close
problems we definitely are nowhere close to automating research this is a great
to automating research this is a great
to automating research this is a great tweet we’re the researchers were the
tweet we’re the researchers were the
tweet we’re the researchers were the ones that wanted to make the AI do all
ones that wanted to make the AI do all
ones that wanted to make the AI do all the work and play games and while they
the work and play games and while they
the work and play games and while they play games but instead it’s the opposite
play games but instead it’s the opposite
play games but instead it’s the opposite right the yeah is just playing games all
right the yeah is just playing games all
right the yeah is just playing games all day and researchers are working harder
day and researchers are working harder
day and researchers are working harder than ever it’s a tough life I think the
than ever it’s a tough life I think the
than ever it’s a tough life I think the comments on this were equally great
comments on this were equally great
comments on this were equally great because maybe this is a sign that the AI
because maybe this is a sign that the AI
because maybe this is a sign that the AI is actually intelligent you know maybe
is actually intelligent you know maybe
is actually intelligent you know maybe it’s like just pretending to be dumb and
it’s like just pretending to be dumb and
it’s like just pretending to be dumb and being like why would I want to do all
being like why would I want to do all
being like why would I want to do all the work I’m just gonna keep playing
the work I’m just gonna keep playing
the work I’m just gonna keep playing games all day some aspects of research
games all day some aspects of research
games all day some aspects of research might be automated something that some
might be automated something that some
might be automated something that some people consider to be either boring or a
people consider to be either boring or a
people consider to be either boring or a waste of time or hard is designing these
waste of time or hard is designing these
waste of time or hard is designing these architectures in the first place and
architectures in the first place and
architectures in the first place and there has been some work in using deep
there has been some work in using deep
there has been some work in using deep learning to automate the architect the
learning to automate the architect the
learning to automate the architect the design of architectures for more deep
design of architectures for more deep
design of architectures for more deep learning and you get like these crazy
learning and you get like these crazy
learning and you get like these crazy things that no one would ever design
things that no one would ever design
things that no one would ever design yeah I would definitely not
yeah I would definitely not
yeah I would definitely not think to do that in the right so this
think to do that in the right so this
think to do that in the right so this stuff has had some fairly promising
stuff has had some fairly promising
stuff has had some fairly promising results I I put this under a maybe of
results I I put this under a maybe of
results I I put this under a maybe of what can be plausible because it’s both
what can be plausible because it’s both
what can be plausible because it’s both it was both very expensive and not quite
it was both very expensive and not quite
it was both very expensive and not quite as good as a state of the art but this
as good as a state of the art but this
as good as a state of the art but this seems like a really promising Avenue and
seems like a really promising Avenue and
seems like a really promising Avenue and a potential place that it could make a
a potential place that it could make a
a potential place that it could make a big impact so maybe all of our learning
big impact so maybe all of our learning
big impact so maybe all of our learning about architecture and studying this and
about architecture and studying this and
about architecture and studying this and trial and error maybe all of this will
trial and error maybe all of this will
trial and error maybe all of this will be outsourced to you know farms of
be outsourced to you know farms of
be outsourced to you know farms of computers somewhere and we could just
computers somewhere and we could just
computers somewhere and we could just you know stick to the high level tasks
you know stick to the high level tasks
you know stick to the high level tasks but life is rarely that kind a despite
but life is rarely that kind a despite
but life is rarely that kind a despite fake news to the contrary we are long
fake news to the contrary we are long
fake news to the contrary we are long ways away from automating software
ways away from automating software
ways away from automating software development there were some articles on
development there were some articles on
development there were some articles on algorithms automating coding and III
algorithms automating coding and III
algorithms automating coding and III think that some people were a little bit
think that some people were a little bit
think that some people were a little bit panicked on this maybe all of the
panicked on this maybe all of the
panicked on this maybe all of the articles when deepening automates X
articles when deepening automates X
articles when deepening automates X causes some panic but I hang around I
causes some panic but I hang around I
causes some panic but I hang around I hang out with lots of software engineers
hang out with lots of software engineers
hang out with lots of software engineers so they were worried for like a second
so they were worried for like a second
so they were worried for like a second until they realized that this thing was
until they realized that this thing was
until they realized that this thing was actually really really dumb not that the
actually really really dumb not that the
actually really really dumb not that the the work was done but how the algorithm
the work was done but how the algorithm
the work was done but how the algorithm did it was nowhere close to software
did it was nowhere close to software
did it was nowhere close to software engineering it was a slightly better
engineering it was a slightly better
engineering it was a slightly better heuristic for picking random bits of
heuristic for picking random bits of
heuristic for picking random bits of code together and doing trial and error
code together and doing trial and error
code together and doing trial and error on that code and as we all know that is
on that code and as we all know that is
on that code and as we all know that is absolutely not how we do software
absolutely not how we do software
absolutely not how we do software engineering right like we design stuff
engineering right like we design stuff
engineering right like we design stuff upfront not trial and error it’s like
upfront not trial and error it’s like
upfront not trial and error it’s like all done by the books yeah this
all done by the books yeah this
all done by the books yeah this algorithm definitely can’t do that
algorithm definitely can’t do that
algorithm definitely can’t do that so we our jobs are safe right guys
ok so some people some people know what
ok so some people some people know what I’m talking about some great memes this
I’m talking about some great memes this
I’m talking about some great memes this is not really model output but if you do
is not really model output but if you do
is not really model output but if you do follow the field people love to have fun
follow the field people love to have fun
follow the field people love to have fun things in there some people actually
things in there some people actually
things in there some people actually wrote a conf a fait paper I think that’s
wrote a conf a fait paper I think that’s
wrote a conf a fait paper I think that’s pretty incredible it’s someone can like
pretty incredible it’s someone can like
pretty incredible it’s someone can like dedicate a research paper with I assume
dedicate a research paper with I assume
dedicate a research paper with I assume a real idea I didn’t read this but I
a real idea I didn’t read this but I
a real idea I didn’t read this but I assume it’s a real idea
assume it’s a real idea
assume it’s a real idea to a troll name I think it’s great and
to a troll name I think it’s great and
to a troll name I think it’s great and it shows like the speed of publishing in
it shows like the speed of publishing in
it shows like the speed of publishing in the field common Silicon Valley problem
the field common Silicon Valley problem
the field common Silicon Valley problem no deep learning will not solve all of
no deep learning will not solve all of
no deep learning will not solve all of your problems especially not your
your problems especially not your
your problems especially not your product definition problems it won’t
product definition problems it won’t
product definition problems it won’t find something useful for you to do and
find something useful for you to do and
find something useful for you to do and it will not make you magically rich
it will not make you magically rich
it will not make you magically rich despite a lot of belief to the contrary
despite a lot of belief to the contrary
despite a lot of belief to the contrary and similar to this image general chat
and similar to this image general chat
and similar to this image general chat bots are actually quite difficult you’d
bots are actually quite difficult you’d
bots are actually quite difficult you’d think that you just give you no model a
think that you just give you no model a
think that you just give you no model a dataset of two people talking and it’ll
dataset of two people talking and it’ll
dataset of two people talking and it’ll be able to replicate those people
be able to replicate those people
be able to replicate those people talking but it turns out that our
talking but it turns out that our
talking but it turns out that our language models are quite good at making
language models are quite good at making
language models are quite good at making things that look grammatically correct
things that look grammatically correct
things that look grammatically correct but are semantically quite terrible so
but are semantically quite terrible so
but are semantically quite terrible so they don’t have like they don’t have any
they don’t have like they don’t have any
they don’t have like they don’t have any history involved they don’t have there’s
history involved they don’t have there’s
history involved they don’t have there’s lots of issues of them and this like a
lots of issues of them and this like a
lots of issues of them and this like a misunderstanding too that led to a lot
misunderstanding too that led to a lot
misunderstanding too that led to a lot of companies starting products that
of companies starting products that
of companies starting products that ended up pivoting away from using deep
ended up pivoting away from using deep
ended up pivoting away from using deep learning at all and ended up using like
learning at all and ended up using like
learning at all and ended up using like an army of workers in the Philippines
an army of workers in the Philippines
an army of workers in the Philippines this manually doing the chatting for
this manually doing the chatting for
this manually doing the chatting for them which turns out to be a pretty
them which turns out to be a pretty
them which turns out to be a pretty economical way to do things but specific
economical way to do things but specific
economical way to do things but specific chat bots are very doable so when if you
chat bots are very doable so when if you
chat bots are very doable so when if you turn the problem from hey let’s generate
turn the problem from hey let’s generate
turn the problem from hey let’s generate arbitrary text to hey let’s pick among a
arbitrary text to hey let’s pick among a
arbitrary text to hey let’s pick among a small set of valid responses things
small set of valid responses things
small set of valid responses things become a lot easier this is in boxes
become a lot easier this is in boxes
become a lot easier this is in boxes smart reply which apparently is used by
smart reply which apparently is used by
smart reply which apparently is used by over ten percent of mobile infox replies
over ten percent of mobile infox replies
over ten percent of mobile infox replies which sounds like a lot of
which sounds like a lot of
which sounds like a lot of qualifications but I just think it’s
qualifications but I just think it’s
qualifications but I just think it’s cool that something started out as an
cool that something started out as an
cool that something started out as an April Fool’s Day is now real April
April Fool’s Day is now real April
April Fool’s Day is now real April Fool’s Day joke it’s also sweet
Fool’s Day joke it’s also sweet
Fool’s Day joke it’s also sweet animation but this kind of stuff is very
animation but this kind of stuff is very
animation but this kind of stuff is very plausible and I think that people who do
plausible and I think that people who do
plausible and I think that people who do use machine learning for chat BOTS will
use machine learning for chat BOTS will
use machine learning for chat BOTS will end up constraining the problem quite a
end up constraining the problem quite a
end up constraining the problem quite a bit and that’s actually very doable if
bit and that’s actually very doable if
bit and that’s actually very doable if you’re trying to classify like do I have
you’re trying to classify like do I have
you’re trying to classify like do I have enough information or is this person
enough information or is this person
enough information or is this person satisfied or do I need to pull another
satisfied or do I need to pull another
satisfied or do I need to pull another human in to actually chat through this
human in to actually chat through this
human in to actually chat through this person that’s much more doable than hey
person that’s much more doable than hey
person that’s much more doable than hey automatically solve this person’s IT
automatically solve this person’s IT
automatically solve this person’s IT issues which sounds really hard and
issues which sounds really hard and
issues which sounds really hard and similar in vain to that coherent text is
similar in vain to that coherent text is
similar in vain to that coherent text is quite a challenge like long any long
quite a challenge like long any long
quite a challenge like long any long amount of text a lot of journalists
amount of text a lot of journalists
amount of text a lot of journalists journalists I feel like a big victim to
journalists I feel like a big victim to
journalists I feel like a big victim to hype because it’s kind of their fault
hype because it’s kind of their fault
hype because it’s kind of their fault uh-huh and they’re kind of worried about
uh-huh and they’re kind of worried about
uh-huh and they’re kind of worried about their jobs about RDR deep nets and like
their jobs about RDR deep nets and like
their jobs about RDR deep nets and like start writing articles for us and the
start writing articles for us and the
start writing articles for us and the answer seems to be no so if you’re a
answer seems to be no so if you’re a
answer seems to be no so if you’re a journalist
journalist
journalist don’t worry sorry
did you say it’s hard to investigate
did you say it’s hard to investigate journalists as an AI
I couldn’t quite hear the last part of
I couldn’t quite hear the last part of that oh yeah for sure
that oh yeah for sure
that oh yeah for sure he said as may I it’s hard to do
he said as may I it’s hard to do
he said as may I it’s hard to do investigative journalism that is
investigative journalism that is
investigative journalism that is definitely true debatable about how much
definitely true debatable about how much
definitely true debatable about how much investigative journalism current
investigative journalism current
investigative journalism current journalists do but that’s definitely the
journalists do but that’s definitely the
journalists do but that’s definitely the case I think in this case it is even the
case I think in this case it is even the
case I think in this case it is even the worry that a lot of journalism is read
worry that a lot of journalism is read
worry that a lot of journalism is read stuff on Twitter turn it into an article
stuff on Twitter turn it into an article
stuff on Twitter turn it into an article hope to get lots of clicks make click
hope to get lots of clicks make click
hope to get lots of clicks make click vadie headline topic definitely not all
vadie headline topic definitely not all
vadie headline topic definitely not all of it but some fraction of it is that
of it but some fraction of it is that
of it but some fraction of it is that and I think that there is some worry
and I think that there is some worry
and I think that there is some worry about this like I believe in finance
about this like I believe in finance
about this like I believe in finance there’s a big race to like who can
there’s a big race to like who can
there’s a big race to like who can publish these articles first based on
publish these articles first based on
publish these articles first based on various data sources and if you’re not
various data sources and if you’re not
various data sources and if you’re not talking about quality but speed these
talking about quality but speed these
talking about quality but speed these things definitely have a speed advantage
yeah III I’m not I’m personally not
yeah III I’m not I’m personally not worried about journalists jobs being
worried about journalists jobs being
worried about journalists jobs being taken for sure cool sorry sorry I
taken for sure cool sorry sorry I
taken for sure cool sorry sorry I couldn’t hear you very well but thank
couldn’t hear you very well but thank
couldn’t hear you very well but thank you for yelling that time yeah so
you for yelling that time yeah so
you for yelling that time yeah so something that seems to be really
something that seems to be really
something that seems to be really promising I actually think that this is
promising I actually think that this is
promising I actually think that this is one of the most promising upcoming uses
one of the most promising upcoming uses
one of the most promising upcoming uses of deep learning that’s like not quite
of deep learning that’s like not quite
of deep learning that’s like not quite there but might be there and it’s like a
there but might be there and it’s like a
there but might be there and it’s like a would be a really sexy field to get into
would be a really sexy field to get into
would be a really sexy field to get into with robotics it seems like there’s a
with robotics it seems like there’s a
with robotics it seems like there’s a lot of really good stuff happening with
lot of really good stuff happening with
lot of really good stuff happening with imitation learning and a lot of people
imitation learning and a lot of people
imitation learning and a lot of people are invested just working oh it is
are invested just working oh it is
are invested just working oh it is working this is pretty cool people are
working this is pretty cool people are
working this is pretty cool people are investing in a lot of the the research
investing in a lot of the the research
investing in a lot of the the research labs are investing and getting the
labs are investing and getting the
labs are investing and getting the robots training together in like how do
robots training together in like how do
robots training together in like how do we collect lots of data for robots to
we collect lots of data for robots to
we collect lots of data for robots to get them working automatically because
get them working automatically because
get them working automatically because right now
right now
right now at least to my understanding I’m no
at least to my understanding I’m no
at least to my understanding I’m no roboticist is that majority of the work
roboticist is that majority of the work
roboticist is that majority of the work done by robots is done manually and if
done by robots is done manually and if
done by robots is done manually and if we can like make it a lot easier to
we can like make it a lot easier to
we can like make it a lot easier to train robots to do things that we care
train robots to do things that we care
train robots to do things that we care about maybe
about maybe
about maybe all of a sudden we’re going to have like
all of a sudden we’re going to have like
all of a sudden we’re going to have like more general programmable robots that
more general programmable robots that
more general programmable robots that people can do stuff with so that I think
people can do stuff with so that I think
people can do stuff with so that I think is really really promising and at least
is really really promising and at least
is really really promising and at least from a research perspective of someone
from a research perspective of someone
from a research perspective of someone who reads the papers and like keeps up
who reads the papers and like keeps up
who reads the papers and like keeps up with what people are doing it seems very
with what people are doing it seems very
with what people are doing it seems very plausible that this kind of thing could
plausible that this kind of thing could
plausible that this kind of thing could make a breakthrough in the near term
make a breakthrough in the near term
make a breakthrough in the near term especially with what’s called imitation
especially with what’s called imitation
especially with what’s called imitation learning where robots rather than
learning where robots rather than
learning where robots rather than learning by trial and error which can be
learning by trial and error which can be
learning by trial and error which can be very hard they just learn to copy humans
very hard they just learn to copy humans
very hard they just learn to copy humans which is goes back into the rule of
which is goes back into the rule of
which is goes back into the rule of thumb I was talking about we’re giving
thumb I was talking about we’re giving
thumb I was talking about we’re giving giving these algorithms supervised data
giving these algorithms supervised data
giving these algorithms supervised data telling them what to do generally works
telling them what to do generally works
telling them what to do generally works a lot better than hoping for magic that
a lot better than hoping for magic that
a lot better than hoping for magic that you know hoping that they will magically
you know hoping that they will magically
you know hoping that they will magically figure out the thing to do which is what
figure out the thing to do which is what
figure out the thing to do which is what a lot of the field is trying to get
a lot of the field is trying to get
a lot of the field is trying to get working right now depending on who you
working right now depending on who you
working right now depending on who you ask
ask
ask game playing I would count categories
game playing I would count categories
game playing I would count categories that is not yet there there have been
that is not yet there there have been
that is not yet there there have been some amazing successes in game playing
some amazing successes in game playing
some amazing successes in game playing but a lot of those successes aren’t
but a lot of those successes aren’t
but a lot of those successes aren’t quite super general a lot of it it’s
quite super general a lot of it it’s
quite super general a lot of it it’s like very input simple input output
like very input simple input output
like very input simple input output mapping like I was mentioning so Atari
mapping like I was mentioning so Atari
mapping like I was mentioning so Atari seemed to be a lot of that debatable
seemed to be a lot of that debatable
seemed to be a lot of that debatable whether or not go was that even though
whether or not go was that even though
whether or not go was that even though that was definitely a huge win but
that was definitely a huge win but
that was definitely a huge win but there’s been other games where models
there’s been other games where models
there’s been other games where models are nowhere near as successful so things
are nowhere near as successful so things
are nowhere near as successful so things like even like very simple Minecraft
like even like very simple Minecraft
like even like very simple Minecraft mazes it’s still the model still aren’t
mazes it’s still the model still aren’t
mazes it’s still the model still aren’t quite there yet or recently there’s been
quite there yet or recently there’s been
quite there yet or recently there’s been a bunch of work on doom visual doom like
a bunch of work on doom visual doom like
a bunch of work on doom visual doom like do them from the pixels and this model
do them from the pixels and this model
do them from the pixels and this model actually super cool let me see here does
actually super cool let me see here does
actually super cool let me see here does this work you can skip to the fighting
this work you can skip to the fighting
this work you can skip to the fighting so there’s been a lot of progress on
so there’s been a lot of progress on
so there’s been a lot of progress on that really recently this is like the
that really recently this is like the
that really recently this is like the what was the state of the art in 2013 if
what was the state of the art in 2013 if
what was the state of the art in 2013 if you can tell it’s like pretty dumb like
you can tell it’s like pretty dumb like
you can tell it’s like pretty dumb like shooting a wall right now let’s see here
shooting a wall right now let’s see here
shooting a wall right now let’s see here yeah this is so this is a little bit
yeah this is so this is a little bit
yeah this is so this is a little bit smarter this was state of the art in I
smarter this was state of the art in I
smarter this was state of the art in I would say 2016 ish mid 2016 this is
would say 2016 ish mid 2016 this is
would say 2016 ish mid 2016 this is still pretty dumb
still pretty dumb
still pretty dumb and people been making a lot more
and people been making a lot more
and people been making a lot more progress recently with this we’re look
progress recently with this we’re look
progress recently with this we’re look at this this is actually intimidating
at this this is actually intimidating
at this this is actually intimidating it’s like moving around its shooting
it’s like moving around its shooting
it’s like moving around its shooting intelligently etc etc there’s been a lot
intelligently etc etc there’s been a lot
intelligently etc etc there’s been a lot more progress being made in this and it
more progress being made in this and it
more progress being made in this and it seems that we’re nowhere near close to
seems that we’re nowhere near close to
seems that we’re nowhere near close to or at least to my knowledge solving
or at least to my knowledge solving
or at least to my knowledge solving something like Starcraft but it’s really
something like Starcraft but it’s really
something like Starcraft but it’s really promising and people are putting a lot
promising and people are putting a lot
promising and people are putting a lot of effort into this so the likelihood
of effort into this so the likelihood
of effort into this so the likelihood that we make some big breakthroughs in
that we make some big breakthroughs in
that we make some big breakthroughs in the coming years seems to be likely cool
the coming years seems to be likely cool
the coming years seems to be likely cool and this is category of stuff it’s like
and this is category of stuff it’s like
and this is category of stuff it’s like what I am one of the most excited about
what I am one of the most excited about
what I am one of the most excited about just because I would never consider
just because I would never consider
just because I would never consider using like these extremely powerful
using like these extremely powerful
using like these extremely powerful classification models for artsy things
classification models for artsy things
classification models for artsy things maybe that’s just me but some of these
maybe that’s just me but some of these
maybe that’s just me but some of these use cases are incredibly creative and
use cases are incredibly creative and
use cases are incredibly creative and incredibly cool and like this stuff is
incredibly cool and like this stuff is
incredibly cool and like this stuff is amazing this one came out pretty
amazing this one came out pretty
amazing this one came out pretty recently and it learns to transform
recently and it learns to transform
recently and it learns to transform images from different domains so
images from different domains so
images from different domains so transforming like a zebra into a horse
transforming like a zebra into a horse
transforming like a zebra into a horse or vice versa so image transformation it
or vice versa so image transformation it
or vice versa so image transformation it can be done can even be done with videos
can be done can even be done with videos
can be done can even be done with videos so this is like actually really done by
so this is like actually really done by
so this is like actually really done by a model it’s not like cherry pick data
but so it’s actually a transform this
but so it’s actually a transform this video and like this is not seamless but
video and like this is not seamless but
video and like this is not seamless but that’s pretty good better than I could
that’s pretty good better than I could
that’s pretty good better than I could do with Photoshop which is not saying
do with Photoshop which is not saying
do with Photoshop which is not saying much but like this is like pretty
much but like this is like pretty
much but like this is like pretty impressive and I would not even have
impressive and I would not even have
impressive and I would not even have thought of this as a use case like hey
thought of this as a use case like hey
thought of this as a use case like hey I’m a machining researcher at Google I
I’m a machining researcher at Google I
I’m a machining researcher at Google I have a you know giant cluster
have a you know giant cluster
have a you know giant cluster I’m gonna transform a horse into a zebra
I’m gonna transform a horse into a zebra
I’m gonna transform a horse into a zebra right
unfortunately this kind of thing is not
unfortunately this kind of thing is not completely reliable but this is pretty
completely reliable but this is pretty
completely reliable but this is pretty amusing so like with all machine
amusing so like with all machine
amusing so like with all machine learning like making it completely
learning like making it completely
learning like making it completely reliable can be challenging talking
reliable can be challenging talking
reliable can be challenging talking about that there’s been an app that’s
about that there’s been an app that’s
about that there’s been an app that’s been gaining popularity called face app
been gaining popularity called face app
been gaining popularity called face app that
that
that does facial transformation so in the top
does facial transformation so in the top
does facial transformation so in the top left you see the original photo top
left you see the original photo top
left you see the original photo top right you see like a more manly
right you see like a more manly
right you see like a more manly transformation you know like more edgy
transformation you know like more edgy
transformation you know like more edgy chin bill beardy then bottom left and
chin bill beardy then bottom left and
chin bill beardy then bottom left and old miss transformation and bottom right
old miss transformation and bottom right
old miss transformation and bottom right a smiling transformation and this is
a smiling transformation and this is
a smiling transformation and this is actually pretty good pretty good and
actually pretty good pretty good and
actually pretty good pretty good and like an app can do it on your phone no
like an app can do it on your phone no
like an app can do it on your phone no human input it just does it it’s pretty
human input it just does it it’s pretty
human input it just does it it’s pretty impressive that you can do this and this
impressive that you can do this and this
impressive that you can do this and this is a really cool use case unfortunately
is a really cool use case unfortunately
is a really cool use case unfortunately it’s not perfect in particular it’s also
it’s not perfect in particular it’s also
it’s not perfect in particular it’s also suffers from that bias problem like with
suffers from that bias problem like with
suffers from that bias problem like with many other things when you turn a cool
many other things when you turn a cool
many other things when you turn a cool model into a product it there’s a
model into a product it there’s a
model into a product it there’s a different set of requirements in this
different set of requirements in this
different set of requirements in this case they had a transformation which
case they had a transformation which
case they had a transformation which makes a person’s face hotter and one of
makes a person’s face hotter and one of
makes a person’s face hotter and one of the things it did was it always lighten
the things it did was it always lighten
the things it did was it always lighten skin which was offensive to some people
skin which was offensive to some people
skin which was offensive to some people yeah that they had to pull that feature
yeah that they had to pull that feature
yeah that they had to pull that feature I think or change I think they actually
I think or change I think they actually
I think or change I think they actually changed the name from hot to something
changed the name from hot to something
changed the name from hot to something else I can’t remember art is doable this
else I can’t remember art is doable this
else I can’t remember art is doable this is art from scratch or unconditional art
is art from scratch or unconditional art
is art from scratch or unconditional art like it’s like these models can just
like it’s like these models can just
like it’s like these models can just create these artsy things and I think
create these artsy things and I think
create these artsy things and I think that this is really really cool I
that this is really really cool I
that this is really really cool I personally think that these both look
personally think that these both look
personally think that these both look really good can I get a show of hands of
really good can I get a show of hands of
really good can I get a show of hands of who thinks the one on the left is better
who thinks the one on the left is better
who thinks the one on the left is better what about the right oh it looks like a
what about the right oh it looks like a
what about the right oh it looks like a tie I made the one on the left so I was
tie I made the one on the left so I was
tie I made the one on the left so I was hoping that people would vote for that
hoping that people would vote for that
hoping that people would vote for that one cool I think they’re both really
one cool I think they’re both really
one cool I think they’re both really cool I would definitely have a poster of
cool I would definitely have a poster of
cool I would definitely have a poster of that in my room or a painting of that in
that in my room or a painting of that in
that in my room or a painting of that in my house then this kind of thing like
my house then this kind of thing like
my house then this kind of thing like who would have even thought that as a
who would have even thought that as a
who would have even thought that as a side effect of these really powerful
side effect of these really powerful
side effect of these really powerful actually useful things we would get art
oh yeah I’m definitely not claiming that
oh yeah I’m definitely not claiming that this is the like the the the first thing
this is the like the the the first thing
this is the like the the the first thing in terms of algorithmic arts it’s just I
in terms of algorithmic arts it’s just I
in terms of algorithmic arts it’s just I find to just be a really cool use case
find to just be a really cool use case
find to just be a really cool use case of deep learning because when III don’t
of deep learning because when III don’t
of deep learning because when III don’t think ten years ago people would have
think ten years ago people would have
think ten years ago people would have imagined like yeah imagine all of these
imagined like yeah imagine all of these
imagined like yeah imagine all of these cool pictures will make and every
cool pictures will make and every
cool pictures will make and every actually every time there’s a new use
actually every time there’s a new use
actually every time there’s a new use case in art I’m just amazed like who
case in art I’m just amazed like who
case in art I’m just amazed like who thought of this like who spent their
thought of this like who spent their
thought of this like who spent their time on this and I’m thankful for that
time on this and I’m thankful for that
time on this and I’m thankful for that because I wouldn’t have done it but I
because I wouldn’t have done it but I
because I wouldn’t have done it but I think it’s really awesome and I think in
think it’s really awesome and I think in
think it’s really awesome and I think in some ways it’s also kind of cool that
some ways it’s also kind of cool that
some ways it’s also kind of cool that unlike fractals or something like that
unlike fractals or something like that
unlike fractals or something like that it feels like there’s there’s a lot more
it feels like there’s there’s a lot more
it feels like there’s there’s a lot more like there’s more unknown unknowns in
like there’s more unknown unknowns in
like there’s more unknown unknowns in this right now which makes it really
this right now which makes it really
this right now which makes it really promising as well maybe that’s from my
promising as well maybe that’s from my
promising as well maybe that’s from my misunderstanding of art though or
misunderstanding of art though or
misunderstanding of art though or algorithms or anything I’m not expert in
algorithms or anything I’m not expert in
algorithms or anything I’m not expert in any of this stuff sketching with another
any of this stuff sketching with another
any of this stuff sketching with another recent use case where you just train in
recent use case where you just train in
recent use case where you just train in a data set of humans drawing little
a data set of humans drawing little
a data set of humans drawing little things and the things in the the top
things and the things in the the top
things and the things in the the top corner up there is things that the model
corner up there is things that the model
corner up there is things that the model drew and in the bottom here you can
drew and in the bottom here you can
drew and in the bottom here you can actually do math on sketches so you take
actually do math on sketches so you take
actually do math on sketches so you take like a cat face you add in a pig with a
like a cat face you add in a pig with a
like a cat face you add in a pig with a body you subtract a pig face and you end
body you subtract a pig face and you end
body you subtract a pig face and you end up with like a cat with a body and like
up with like a cat with a body and like
up with like a cat with a body and like it’s kind of cool that it works I mean
it’s kind of cool that it works I mean
it’s kind of cool that it works I mean the math checks out so awesome for for
the math checks out so awesome for for
the math checks out so awesome for for the non artists in the room you can also
the non artists in the room you can also
the non artists in the room you can also turn what is arguably not art in the
turn what is arguably not art in the
turn what is arguably not art in the bottom-left into something that is
bottom-left into something that is
bottom-left into something that is potentially art so this is another very
potentially art so this is another very
potentially art so this is another very cool use case where you like it can like
cool use case where you like it can like
cool use case where you like it can like enable people to it becomes almost like
enable people to it becomes almost like
enable people to it becomes almost like you know a new artistic medium right
you know a new artistic medium right
you know a new artistic medium right where you can now use these things to
where you can now use these things to
where you can now use these things to enhance existing art to do things maybe
enhance existing art to do things maybe
enhance existing art to do things maybe that people wouldn’t have done before or
that people wouldn’t have done before or
that people wouldn’t have done before or enable people who couldn’t have done
enable people who couldn’t have done
enable people who couldn’t have done this before
this before
this before or maybe just make it more faster
or maybe just make it more faster
or maybe just make it more faster thing like that it feels almost like you
thing like that it feels almost like you
thing like that it feels almost like you know like a new instrument from the
know like a new instrument from the
know like a new instrument from the musical sense so this stuff is really
musical sense so this stuff is really
musical sense so this stuff is really cool style transfer I think this is
cool style transfer I think this is
cool style transfer I think this is crazy because a year and a half ago this
crazy because a year and a half ago this
crazy because a year and a half ago this was already looking really good and it’s
was already looking really good and it’s
was already looking really good and it’s just gotten better and better so this is
just gotten better and better so this is
just gotten better and better so this is like going so well I actually I should
like going so well I actually I should
like going so well I actually I should have put the old pictures here as well
have put the old pictures here as well
have put the old pictures here as well but like these are the new what I think
but like these are the new what I think
but like these are the new what I think is the latest in style transfer and this
is the latest in style transfer and this
is the latest in style transfer and this is pretty good like you can see
is pretty good like you can see
is pretty good like you can see transferring the style of a fire into a
transferring the style of a fire into a
transferring the style of a fire into a bottle like this is like a professional
bottle like this is like a professional
bottle like this is like a professional Photoshop job and I guess and this is
Photoshop job and I guess and this is
Photoshop job and I guess and this is like impressive and I would want to do
like impressive and I would want to do
like impressive and I would want to do this and I look forward to this being
this and I look forward to this being
this and I look forward to this being able to be done for me because I don’t
able to be done for me because I don’t
able to be done for me because I don’t want to implement it myself but there’s
want to implement it myself but there’s
want to implement it myself but there’s a lot of really cool stuff being done
a lot of really cool stuff being done
a lot of really cool stuff being done with style transfer and this stuff is
with style transfer and this stuff is
with style transfer and this stuff is really pragmatic because like
really pragmatic because like
really pragmatic because like aesthetically this is already like very
aesthetically this is already like very
aesthetically this is already like very high quality this is my crowning
high quality this is my crowning
high quality this is my crowning achievement actually mixing my face with
achievement actually mixing my face with
achievement actually mixing my face with that of a Pokemon probably my best
that of a Pokemon probably my best
that of a Pokemon probably my best achievement and deep learning definitely
achievement and deep learning definitely
achievement and deep learning definitely works would recommend trying it again
works would recommend trying it again
works would recommend trying it again and probably newer stuff will work even
and probably newer stuff will work even
and probably newer stuff will work even better and as far as specific go there’s
better and as far as specific go there’s
better and as far as specific go there’s all sorts of other things a rough
all sorts of other things a rough
all sorts of other things a rough formula is taken input that is similar
formula is taken input that is similar
formula is taken input that is similar to another input that deep learning has
to another input that deep learning has
to another input that deep learning has succeeded on like images audio raw text
succeeded on like images audio raw text
succeeded on like images audio raw text other domains like that pick a response
other domains like that pick a response
other domains like that pick a response that is a relatively simple mapping from
that is a relatively simple mapping from
that is a relatively simple mapping from that input so nothing too complicated
that input so nothing too complicated
that input so nothing too complicated but simple mappings like are they human
but simple mappings like are they human
but simple mappings like are they human faces and collected data set trainer
faces and collected data set trainer
faces and collected data set trainer model usually something like that gets
model usually something like that gets
model usually something like that gets just something that works quite well and
just something that works quite well and
just something that works quite well and as far as what it can do if you if you
as far as what it can do if you if you
as far as what it can do if you if you pick the right things generally it
pick the right things generally it
pick the right things generally it generally makes it is the valgar isn’t
generally makes it is the valgar isn’t
generally makes it is the valgar isn’t helped you a lot in doing
helped you a lot in doing
helped you a lot in doing a lot of the easy work for you getting
a lot of the easy work for you getting
a lot of the easy work for you getting like the last bit of presents always a
like the last bit of presents always a
like the last bit of presents always a lot of work but you you’ll know if you
lot of work but you you’ll know if you
lot of work but you you’ll know if you can get it which is it which makes it a
can get it which is it which makes it a
can get it which is it which makes it a little bit easier
little bit easier
little bit easier oh so back to the big questions how will
oh so back to the big questions how will
oh so back to the big questions how will the world change I think this is a great
the world change I think this is a great
the world change I think this is a great tweet like Andrew Inge
tweet like Andrew Inge
tweet like Andrew Inge I do believe automation and steroids is
I do believe automation and steroids is
I do believe automation and steroids is the right way to think about it not
the right way to think about it not
the right way to think about it not sentience or AI overlords or anything
sentience or AI overlords or anything
sentience or AI overlords or anything else like that I actually would be quite
else like that I actually would be quite
else like that I actually would be quite pleasantly surprised to see generally I
pleasantly surprised to see generally I
pleasantly surprised to see generally I in my lifetime just because I think
in my lifetime just because I think
in my lifetime just because I think that’s so unlikely and that’s not
that’s so unlikely and that’s not
that’s so unlikely and that’s not because despite what my current pants
because despite what my current pants
because despite what my current pants might imply that I’m one of the live
might imply that I’m one of the live
might imply that I’m one of the live fast die young types it’s it’s that I I
fast die young types it’s it’s that I I
fast die young types it’s it’s that I I think that it’s quite quite a ways away
think that it’s quite quite a ways away
think that it’s quite quite a ways away though I would love to be wrong as far
though I would love to be wrong as far
though I would love to be wrong as far as how the world change I won’t claim to
as how the world change I won’t claim to
as how the world change I won’t claim to be an expert on the societal effects of
be an expert on the societal effects of
be an expert on the societal effects of automation but luckily this guy would he
automation but luckily this guy would he
automation but luckily this guy would he had a TED talk called will automation
had a TED talk called will automation
had a TED talk called will automation take away all our jobs all seems like a
take away all our jobs all seems like a
take away all our jobs all seems like a little bit of a weasel word here it has
little bit of a weasel word here it has
little bit of a weasel word here it has over a million views and bet riches law
over a million views and bet riches law
over a million views and bet riches law of headlines applies here any headline
of headlines applies here any headline
of headlines applies here any headline that ends in a question mark can be
that ends in a question mark can be
that ends in a question mark can be answered by no so the answer is no
answered by no so the answer is no
answered by no so the answer is no saving you 18 minutes the claim is that
saving you 18 minutes the claim is that
saving you 18 minutes the claim is that AI automation just like other automation
AI automation just like other automation
AI automation just like other automation will take some jobs away but it will
will take some jobs away but it will
will take some jobs away but it will probably do much more transformation of
probably do much more transformation of
probably do much more transformation of jobs because a lot of jobs aren’t these
jobs because a lot of jobs aren’t these
jobs because a lot of jobs aren’t these just simple mappings and there’s more
just simple mappings and there’s more
just simple mappings and there’s more complicated new instincts to them but
complicated new instincts to them but
complicated new instincts to them but automation increases the leverage of any
automation increases the leverage of any
automation increases the leverage of any person and there will be a lot more jobs
person and there will be a lot more jobs
person and there will be a lot more jobs that we just can’t imagine will happen
that we just can’t imagine will happen
that we just can’t imagine will happen so that’s his view I don’t have strong
so that’s his view I don’t have strong
so that’s his view I don’t have strong views in this yeah what can deep
views in this yeah what can deep
views in this yeah what can deep learning do the fields really exciting
learning do the fields really exciting
learning do the fields really exciting there’s a whole lot of things that we
there’s a whole lot of things that we
there’s a whole lot of things that we can do now that we previously couldn’t a
can do now that we previously couldn’t a
can do now that we previously couldn’t a lot of things that were once thought to
lot of things that were once thought to
lot of things that were once thought to be really really hard are now doable
be really really hard are now doable
be really really hard are now doable people fought out of solving go was like
people fought out of solving go was like
people fought out of solving go was like a hundred years away or more and there’s
a hundred years away or more and there’s
a hundred years away or more and there’s all sorts of fields that this could
all sorts of fields that this could
all sorts of fields that this could affect
affect
affect as far as specifics ago which is what
as far as specifics ago which is what
as far as specifics ago which is what would actually be more useful to you
would actually be more useful to you
would actually be more useful to you guys
guys
guys the answer is it’s kind of complicated
the answer is it’s kind of complicated
the answer is it’s kind of complicated my advice would be to look at example
my advice would be to look at example
my advice would be to look at example failure and success cases network with
failure and success cases network with
failure and success cases network with researchers and people in industry or
researchers and people in industry or
researchers and people in industry or use someone’s rule of thumb maybe my own
use someone’s rule of thumb maybe my own
use someone’s rule of thumb maybe my own maybe take it and change it but really
maybe take it and change it but really
maybe take it and change it but really you want to build your own mental
you want to build your own mental
you want to build your own mental classifier of what isn’t isn’t possible
classifier of what isn’t isn’t possible
classifier of what isn’t isn’t possible and refine that classifier around the
and refine that classifier around the
and refine that classifier around the set of problems that you specifically
set of problems that you specifically
set of problems that you specifically care about so if you want like a
care about so if you want like a
care about so if you want like a specific like I want to figure out if
specific like I want to figure out if
specific like I want to figure out if deep learning can find X in a genomic
deep learning can find X in a genomic
deep learning can find X in a genomic data set you like go into the research
data set you like go into the research
data set you like go into the research look at that figure out like what seems
look at that figure out like what seems
look at that figure out like what seems possible what’s not and there’s so many
possible what’s not and there’s so many
possible what’s not and there’s so many problems out there that you might have
problems out there that you might have
problems out there that you might have you might end up being the world expert
you might end up being the world expert
you might end up being the world expert in knowing if deep learning works for
in knowing if deep learning works for
in knowing if deep learning works for your problem so it’s just there’s so
your problem so it’s just there’s so
your problem so it’s just there’s so much opportunity right there that if you
much opportunity right there that if you
much opportunity right there that if you ask anyone there probably will tell you
ask anyone there probably will tell you
ask anyone there probably will tell you an answer because people love to give
an answer because people love to give
an answer because people love to give answers but they probably won’t give you
answers but they probably won’t give you
answers but they probably won’t give you a very good one including myself so
a very good one including myself so
a very good one including myself so right now it’s unavoidable to do some of
right now it’s unavoidable to do some of
right now it’s unavoidable to do some of that work unless you do something that
that work unless you do something that
that work unless you do something that someone’s already solved but that’s kind
someone’s already solved but that’s kind
someone’s already solved but that’s kind of a cop-out answer the last big
of a cop-out answer the last big
of a cop-out answer the last big question it was how do I take advantage
question it was how do I take advantage
question it was how do I take advantage of these trends I think it’s a lot like
of these trends I think it’s a lot like
of these trends I think it’s a lot like learning softer engineering especially
learning softer engineering especially
learning softer engineering especially back in like when the internet was young
back in like when the internet was young
back in like when the internet was young and my answers scratch your own itch
and my answers scratch your own itch
and my answers scratch your own itch play around with it a lot of the work on
play around with it a lot of the work on
play around with it a lot of the work on art specifically was done by hobbyists
art specifically was done by hobbyists
art specifically was done by hobbyists and not researchers and we have no idea
and not researchers and we have no idea
and not researchers and we have no idea yet what can be done on you know the
yet what can be done on you know the
yet what can be done on you know the problems you care about and for all you
problems you care about and for all you
problems you care about and for all you know you might be sitting on depending
know you might be sitting on depending
know you might be sitting on depending the next killer app that no one else is
the next killer app that no one else is
the next killer app that no one else is thought of and it’s scratching your own
thought of and it’s scratching your own
thought of and it’s scratching your own itch leads to something valid for others
itch leads to something valid for others
itch leads to something valid for others start a company on it there’s lots of
start a company on it there’s lots of
start a company on it there’s lots of money for companies going around right
money for companies going around right
money for companies going around right right now and the world needs more AI
right now and the world needs more AI
right now and the world needs more AI companies that actually provide value
companies that actually provide value
companies that actually provide value I’m not gonna name names and also
I’m not gonna name names and also
I’m not gonna name names and also prepare for like a sweet transition
prepare for like a sweet transition
prepare for like a sweet transition consider joining Google it’s the best or
consider joining Google it’s the best or
consider joining Google it’s the best or any other like AI focused company which
any other like AI focused company which
any other like AI focused company which is like interesting impactful problems
is like interesting impactful problems
is like interesting impactful problems and the resources to solve those
and the resources to solve those
and the resources to solve those problems
problems
problems thank you
thank you
thank you you
”
Be First to Comment