Les algorithmes du vivant. | Lê Nguyên Hoang | TEDxSaclay

order

can it emerge without authority what is

the nature of the living what is it to

take

these are big questions

philosophical and yet today

we will see that these questions of

more mysterious mathematical answers

again these answers were discovered

and independently by

mathematicians and by nature but

before we get there let’s stop us on

morphogenesis morphogenesis it’s a

the most fascinating phenomena of

living think about it

in the space of a few days or

few months a single cell gets

multiply and become more complex until

take the end of a complete body

which can be as sophisticated as the

human body his eyes his system

immune and its cerebral cortex

how is this possible how is this

possible that a single cell

become such a sophisticated being

so I’m going to talk that facing a

such question

the last thing that comes to you at

the mind these are the algorithms and that

it may be because for many

of you stands a species of

great wall between the living world

and the world of algorithms

if that’s the case it’s a shame because

that often see the world of

algorithms with the concepts of the

biology or conversely see the world of

living with the concepts of

algorithmic is to discover a

exquisite hidden beauty when I observe the

morphogenesis I see units of

calculation that manage to communicate

to tell yourself you’re doing the head – the

arm him his legs and the craziest it is

that this coordination is done without

central authority

that is to say that unlike what I

just illustrated no cell can

become president and dictate to others

what she has to do

the algorithm of morphogenesis is a

decentralized algorithm says more

simply morphogenesis is a little

like internet but in super mega well

fucking

in 1952 alan turing the father of

IT is farther than that

turing a postulate that the cells

communicates via chemicals and

that the key to morphogenesis was the

production broadcasting and

interactions between these products

chemical turing showed that the cases

simplistic of his theory was enough to

explain the welts and the tasks so

typical of the living but especially the

modern computer simulations said

of digital morphogenesis suggests that

the more sophisticated cases of theory

turing are well and truly

to explain the wonders of the

biological morphogenesis

see also extend the horizon of

possible product designers from the

future

today following the mathematics of

turing to the biology of morphogenesis

and numerical simulations

we have an unequivocal answer to our

first philosophical question

yes the order can emerge without authority

nature and even manages to everything

moment in a staggering way and we

we still have a lot to

learn from her so saying that I

you realize that there is a risk that

you imagined that nature can

transcend the laws of physics

intuitively we want to say that the

biological system are way too much

complex to follow rules purely

mechanistic in the 1940s

the mathematician john von neumann has

questioned this intuitive monomane postulate

being a theoretician he completely

ignored the physico-chemical constraints

for a mathematician the physics and the

chemistry it’s already a lot

far too complicated it takes

simplify monomania so imagined a

extremely simplistic alternative universe

which is nothing but an immense

grid every box maybe in

one of 29 possible states here

represented by colors then to

every moment according to the state

present of the universe and laws

physical predetermined a new state

physics of the universe is calculated it says

that the universe of vollmann and

a cellular automaton an automaton

cell is the archetype of a

purely mechanistic universe and one

could believe that in such a universe

the living has no chance of appearing

and yet monomane described a

structure inside his universe

which we now call the

replicator of 20 knots man and whose

evolution was characteristic of the

alive what is van the man

mean by that to simplify monomane

identified with four properties of living

heredity variability the ability to

ever more complex and the

selection even oman showed that his

purely mechanical mechanistic indicator

own all its properties in fact

water applicator 20 knots man this

is not so much the structure that one

see here it’s mostly the long band in

underneath and this long strip this code

genetics is long it’s even super

super super long but they play a role

crucial by following mans mechanics

the laws imposed by his universe the

replicator in front of naumann uses his

genetic code to make a

machine able to copy paste the code

genetics their epilator can so be

replicate even better von neumann

showed that we could change the code

genetics of replicators are however

upset the replication process

including when these changes

were adding genetic code the

replicator of vernon joystick is no

only able to mutate

he was also and above all able to

imagine complexify imagine

now this replicators proliferate to

through the universe of one man and then undergo

natural selection processes

all on thousands of billions

even billions of billions of

generation and you start designing

how the incredible sophistication of the

alive can emerge from rules yet

purely mechanistic but it was not

that the craziest

the craziest thing is that the replication of the

director of 20 no man based on the

same principle as the dn7 molecules of the

living

except that remember you are here in

the 40s and the DNA structure

has not been discovered from

purely algorithmic thinking

theoretical good oman had discovered

the algorithm in the heart of the living and he

had seen amazingly just remember

of this image we saw the code

Genetics valves replicators oman

is super super good what is it about the

dn7 living molecules

the biological equivalent of the small

structure that you see on the left and the

cell that contains dna for humans

the size of this cell can be counted

in micro m here is the length of the dna

once this one unrolled

the human dna is around 2 m

for their naumann wine educators the

genetic code of our cells and super

super good this is the solution

algorithmic who found nature

to allow heredity variability

and the complexification that will have

then allowed the emergence of structures

as sophisticated as the human body

his eyes his immune system and his

cerebral cortex the cerebral cortex

more precisely than the number of its

components

which fascinates me

vis-à-vis the human brain is his

faculty to learn

and that brings us to the third and

last question what does it take

today this question

this question fascinates all kinds of

researchers the computer scientists should

make their algorithm smart the

philosopher will like to guarantee the reliability

of science and students would like

validate the exams without going to lecture

and there I can not help but quote

one of my big heroes the giant rei

solomonoff i quote in science my

interest more about the way

whose things were discovered that

the contents of discoveries the golden egg

was not as exciting as law that

bridges den the contents of

discoveries it’s good but done

to make discoveries

it’s so much more crazy and the genius of

solomonoff was to discover the engine

discoveries by combining

algorithmic with a strange

philosophy called the lease to nice mme

so unfortunately I’m not going

have time to explain all that

in detail

and yet as a blizzard you have to

that I say it’s not envied who

I miss him but for today I’m going

to be satisfied with the completeness theorem

of solomonoff this theorem says that if he

is a regularity in a dataset

then the Bayesian algorithmic of

solomonoff will find this regularity to

for sure and this with surprisingly little

data

so even 9 years old comes back to the

conclusion that learning optimally or

made discoveries

it is to apply these equations

pelvic except that of course you go

tell me that clearly in practice we

does not apply the bazine equations of

solomonoff well if since a small

fifteen years a machine learning

and neuroscience have tremendously

progressed and discovered again and

still that simplified versions of

the Bayesian algorithm of solomonoff

actually seems to be constantly

applied by our brains to such an extent

in 2012 the neuroscientist

stanislas dehaene was forced

to teach a course at the college of

France on the subject he called his

course

the brain statisticians the revolution

bayesian in cognitive science and he

had so much to say about the

subject who even felt compelled

to spread his course

not on two days, not two

months but over two amazing years in

optimizing learning skills

of our brain nature seems to have

discovered the Bayesian equations little

alone in the world off

what I find fascinating in all

case we saw today but also

in many other cases that I have

no time to present is that we do not

knows that we are seeing a convergence

between ultra human works

theoretical about the nature of algorithms

and their performance and solutions

who found the living to proliferate

and survive what to remember what

algorithmic thinking has enormously

brought to the study of the violin

but also what inverse a lot of

our algorithmic problems have without

doubt has already been solved by the living

but above all there is something

absolutely wonderful I think about

contemplate the living with the lenses

of the algorithmic as the saying

Richard Feynman

I can appreciate the beauty of a flower

at the same time I see a lot more than

the flower I can imagine the cells to

interior

the complex interactions that also have

their beauty just as the structure

internal and processes I can only

invite you to contemplate you too the

splendor of the algorithms of the living

thank you

[Applause]