
I have one thank you so I hope most of
you have heard it’s the term artificial
intelligence or artificial general
intelligence the sort of textbook
definition would be sort of the study of
building agent as we call them to solve
a variety a large variety of very
complex tasks before without needing to
be reprogrammed so these is a very long
quest which we have taken on and there’s
still a long way to go for us to sort of
see it done but crucially it’s already
shown a lot of advancements and a lot of
progress that we already seen our day
today but this thought I actually want
to explain how we the researchers are
actually trying to achieve this goal
so not only like define it but actually
how are we chuckling nowadays building
idea so imagine for a second that I
asked you to be sort of a bit of an
engineer perhaps and do something that
to us looks extremely trivial we do this
every day so there’s a robot and I ask
you okay you build an algorithm a
sequence of instructions to make this
robot open the door now it is very
common to sort of break out this problem
into several pieces and what you could
do is sort of say do that right so to
open up find the door find where the
handle of the door is grab it push it
down and open it I in turns out out that
if you actually strictly speaking do
this sort of rule-based and hierarchical
way to solve a complex problem breaking
it down into smaller not so complex
problems you end up with something that
perhaps is not so optimal
so if you simply try to what’s going to
happen if somebody open you are not
accounting for these the terrain is not
as you know now as you have hope and the
key concept here is that your algorithm
will not generalize to the complexity of
the world this by the way is from the
darker competition in 2015
these programs are not really as simple
as the four lines I showed you and it
really shows that trying to build
algorithms to these like rule-based
chain of rules
Jorge really sort of is bound to failure
in mistakes of course these were like
the funny ones let’s try to go to the
next line yeah so here is how you could
do this perhaps in a slightly different
way I call this the learning approach
and this is studying a lot knowledge so
take the problem of finding the door
right how would you program that right
so if you you were maybe you have these
pixels you try to find these rectangular
objects that look like doors perhaps and
as you saw there it’s not gonna channel
eyes to any conditions you may have not
imagined and then you have to try on
error iterate and this really doesn’t
scale because you’re just now opening
doors and how about doing anything else
so the pattern here is I’m gonna instead
try to get a lot of pictures of doors
and I have human laborers labeling where
the doors are and what I gonna do is I’m
gonna learn what a lot door looks like
instead of programming in terms of rules
and it is very very important so
unfortunately in computer vision people
care about other objects knock doors and
people generally don’t take pictures of
after something I showed you the example
of finding actually cats and dogs which
are very popular objects and in the
research community so how are we
actually gonna do that so the first step
to finding that cat whether an image
contains let’s say a cat or a dog is to
build a big database of pictures that
contain iron cats or dogs and then
literally as humans to label thousands
or tens of thousands
examples of images like this one which
actually is a dog although this one is
pretty tricky I would say so how would
you even you know code these and go
through the pictures and so on so the
first piece is get these database of
labeled examples once you have these the
second step is allow for a generic
program to be able to learn the
characteristics or the patterns on these
database and that’s precisely what deep
learning and in particular give neural
nets are enabling us to do so here you
see a big neural network that actually
processes the image that were pixels and
activates certain neurons in this
architecture that could well be
for instance detecting the size of the
animal the floor enos whether it has pig
ears and so on and crucially these
features that these neural network is
activating on this image and not
prepared by us these are fully learned
from these data backs so again we have
the database we have this model that is
powerful enough to represent all sorts
of interesting features that eventually
lead to an output that is the
probability that the image contains
attack or dock and what you essentially
do is feed one image at a time to this
network of the training database and ask
it to predict the correct label all the
time eventually all these sort of
connections settle onto a configuration
that works well and generalizes beyond
the images that means being trainer and
internally so well that in fact the
billions of pictures every single day
are being analyzed through this kind of
algorithm so the next and one of the
sort of breakthroughs in 2014 which me
incorporators and other groups are sort
of discovered is not only this idea of
taking an image and producing a
probability but also taking a sort of
sequence of symbols and mapping it
through learning examples to another
sequence of symbols and this is a very
powerful
because now we can also learn for
instance to do machine translation so
here the sequence the input sequence
would be the Chinese factors and the
output would be English works and
crucially you learn these without any
knowledge of a language or which words
correspond to which words all you need
is a large database of pairs of
sentences in Chinese and in English
translations of each other and from a
few of these pairs of examples you
essentially achieve a system that
actually generalizes AI improves upon
the real estate of the art so in fact
Google Translate has been upgraded
recently to use these sort of neural
networks approaches that essentially
reach the gap between the previous best
systems and human quality translations
by F by 50% essentially so these again
is already used and it’s sort of already
out there for everyone to benefit from
another such example that is sort of
interesting is generating speech so we
actually do understand how speech is
generated right there is a vocal tract
which essentially is a bunch of tubes
that check changing shape and you could
actually model this and this is one of
these parametric approaches that you see
there on the left but you can also do is
just record this acoustic the audio
files essentially from many people which
will do very complex from fire there’s
like tens of thousands of points per
second of speech but again you just let
a neural network figure out the sort of
patterns in this signal and what you get
end up with is a very good natural
sounding speech system that again beats
the gap between the previous best and
human speech by about 50% factor so this
is some like drastic already sort of
well-developed ivy
that supervised learning enable us to do
and it’s all about this idea of
imitation learning you imitate humans
through these powerful models deep
neural Nets so now there is another
paradigm I wanted to describe which is
what if we cannot get a large database
of these examples or even better we want
to actually go beyond human capabilities
right if we actually want to beat humans
if we take that we are kind of bounded
by how good they are and this is the
reinforcement learning paradigm and for
instance in the game of gold this was
heavily used to for the recent result
that I forgot chief de la fere so it is
extremely simple how this works it’s
almost like obvious the first step is we
have an agent which is in this case the
top layer and this agent lives in an
environment so the imagine would be the
dog or or in the robot case the world
and this agent has a dog for instance it
wants to win at the game of God and the
way this works is quite simple the agent
observes the environment so it observes
let’s say the goal board or the door or
what one not and it can also act upon
the embark it can for instance play a
piece of the port that would be one
action right I play in position four
coma fight for instance in the in that
sort of game and eventually six I cling
to these the game finishes and there my
environment as creation whirring won the
game or not and after it’s been taught
whether or not it updates its ways to
ensure that essentially the the reward
that it gets the probability of winning
is maximized
so this agent is very very similar to
the neural network that was classifying
cancer dogs but in this case it takes as
the inputs these observations so it
would be a goal board
video games cleaner or not and the
outputs are simply the probability of
each actions or which joystick position
you want to play and through this
process you can achieve superhuman
performance in playing at I’ll go and so
on so very crucially it is most of these
algorithms have been developed
developing games um we just kind of
funny because as a kid I used to play
video games and now I sort of work on
making algorithms for video games I
don’t get to play but despite these
profound you know realizations what you
actually get to do when you do through
these two video games is being able to
scale because as you can imagine this
circle of observations and actions
initially it’s not very good so in other
words if you act randomly you would not
gonna get the works very often so you
have to play a lot in order to achieve
good performance but if you go out to
the real world
things will be very slow if you are to
build 10,000 products that you would
crash and eventually you learn how to do
this that wouldn’t scare everyone so
video games are very important because
you can’t run them in parallel in a
cluster and if you lose it’s not a big
deal but eventually you are testing sort
of ideas about intelligence through this
video guys which were designed by making
essentially humans you know imagine and
plan and so on so this is our sort of
research platform it has been so far
until recently Alex it’s gonna be in the
future as well but what’s the problem
well the environment if the environment
is very simple we will learn by simple
agents so the environment and the agents
go side by side and as you can imagine
the
is extremely complex so there are like
you know predators out there that might
want to eat you you need to energy so
you might want to eat plants or other
other human being not human beings
beings and and it’s very diverse and
very complex right so it seems like
creating environments and creating
agents goes both hand by hand and he
didn’t need that so one of the the ideas
here is if we are simulating the the
natural world and we want for instance
to get this guy to like survive the
winter we we need to simulate snow and
food and so on and that is a hard
problem by itself as well however you
know what you can actually try to do is
well we researchers not only build
agents but we are actually also seeing
about the environments that will make
agents advance and amongst many new
environments that people create last
year we created this environment which
is called the mine lab which looks like
three dimensional you know researchers
can set all sorts of tasks to test to
test ideas about planning memory
imagination and so on and I’ll show you
just a simple example where an agent
will essentially go around and try to
eat apples but do not eat lemons this is
another video where there’s actually
like more naturalistic looking video and
there’s an agent out there that is sort
of navigating these complex works
figuring out that apples might appear
again because they grow and so on and so
forth so these environments are becoming
more complex and as a result our agents
will also of course learn better they’re
now a sort of a final note it is kind of
obvious that we need AI it’s very
helpful to you know be able to have
translation systems you might have a
person
system in your phone like helping you
organize streets and so on so forth but
it also seems a friend that an AI needs
nature we need to encode these complex
rules for AI to be useful to us
otherwise we are just bound by the
complexity of the environment in which
agents living and so we done note them
I’d like to just thank you for your
attention I’ll be happy to chat with you
outside Thanks
[Music]
you have heard it’s the term artificial
intelligence or artificial general
intelligence the sort of textbook
definition would be sort of the study of
building agent as we call them to solve
a variety a large variety of very
complex tasks before without needing to
be reprogrammed so these is a very long
quest which we have taken on and there’s
still a long way to go for us to sort of
see it done but crucially it’s already
shown a lot of advancements and a lot of
progress that we already seen our day
today but this thought I actually want
to explain how we the researchers are
actually trying to achieve this goal
so not only like define it but actually
how are we chuckling nowadays building
idea so imagine for a second that I
asked you to be sort of a bit of an
engineer perhaps and do something that
to us looks extremely trivial we do this
every day so there’s a robot and I ask
you okay you build an algorithm a
sequence of instructions to make this
robot open the door now it is very
common to sort of break out this problem
into several pieces and what you could
do is sort of say do that right so to
open up find the door find where the
handle of the door is grab it push it
down and open it I in turns out out that
if you actually strictly speaking do
this sort of rule-based and hierarchical
way to solve a complex problem breaking
it down into smaller not so complex
problems you end up with something that
perhaps is not so optimal
so if you simply try to what’s going to
happen if somebody open you are not
accounting for these the terrain is not
as you know now as you have hope and the
key concept here is that your algorithm
will not generalize to the complexity of
the world this by the way is from the
darker competition in 2015
these programs are not really as simple
as the four lines I showed you and it
really shows that trying to build
algorithms to these like rule-based
chain of rules
Jorge really sort of is bound to failure
in mistakes of course these were like
the funny ones let’s try to go to the
next line yeah so here is how you could
do this perhaps in a slightly different
way I call this the learning approach
and this is studying a lot knowledge so
take the problem of finding the door
right how would you program that right
so if you you were maybe you have these
pixels you try to find these rectangular
objects that look like doors perhaps and
as you saw there it’s not gonna channel
eyes to any conditions you may have not
imagined and then you have to try on
error iterate and this really doesn’t
scale because you’re just now opening
doors and how about doing anything else
so the pattern here is I’m gonna instead
try to get a lot of pictures of doors
and I have human laborers labeling where
the doors are and what I gonna do is I’m
gonna learn what a lot door looks like
instead of programming in terms of rules
and it is very very important so
unfortunately in computer vision people
care about other objects knock doors and
people generally don’t take pictures of
after something I showed you the example
of finding actually cats and dogs which
are very popular objects and in the
research community so how are we
actually gonna do that so the first step
to finding that cat whether an image
contains let’s say a cat or a dog is to
build a big database of pictures that
contain iron cats or dogs and then
literally as humans to label thousands
or tens of thousands
examples of images like this one which
actually is a dog although this one is
pretty tricky I would say so how would
you even you know code these and go
through the pictures and so on so the
first piece is get these database of
labeled examples once you have these the
second step is allow for a generic
program to be able to learn the
characteristics or the patterns on these
database and that’s precisely what deep
learning and in particular give neural
nets are enabling us to do so here you
see a big neural network that actually
processes the image that were pixels and
activates certain neurons in this
architecture that could well be
for instance detecting the size of the
animal the floor enos whether it has pig
ears and so on and crucially these
features that these neural network is
activating on this image and not
prepared by us these are fully learned
from these data backs so again we have
the database we have this model that is
powerful enough to represent all sorts
of interesting features that eventually
lead to an output that is the
probability that the image contains
attack or dock and what you essentially
do is feed one image at a time to this
network of the training database and ask
it to predict the correct label all the
time eventually all these sort of
connections settle onto a configuration
that works well and generalizes beyond
the images that means being trainer and
internally so well that in fact the
billions of pictures every single day
are being analyzed through this kind of
algorithm so the next and one of the
sort of breakthroughs in 2014 which me
incorporators and other groups are sort
of discovered is not only this idea of
taking an image and producing a
probability but also taking a sort of
sequence of symbols and mapping it
through learning examples to another
sequence of symbols and this is a very
powerful
because now we can also learn for
instance to do machine translation so
here the sequence the input sequence
would be the Chinese factors and the
output would be English works and
crucially you learn these without any
knowledge of a language or which words
correspond to which words all you need
is a large database of pairs of
sentences in Chinese and in English
translations of each other and from a
few of these pairs of examples you
essentially achieve a system that
actually generalizes AI improves upon
the real estate of the art so in fact
Google Translate has been upgraded
recently to use these sort of neural
networks approaches that essentially
reach the gap between the previous best
systems and human quality translations
by F by 50% essentially so these again
is already used and it’s sort of already
out there for everyone to benefit from
another such example that is sort of
interesting is generating speech so we
actually do understand how speech is
generated right there is a vocal tract
which essentially is a bunch of tubes
that check changing shape and you could
actually model this and this is one of
these parametric approaches that you see
there on the left but you can also do is
just record this acoustic the audio
files essentially from many people which
will do very complex from fire there’s
like tens of thousands of points per
second of speech but again you just let
a neural network figure out the sort of
patterns in this signal and what you get
end up with is a very good natural
sounding speech system that again beats
the gap between the previous best and
human speech by about 50% factor so this
is some like drastic already sort of
well-developed ivy
that supervised learning enable us to do
and it’s all about this idea of
imitation learning you imitate humans
through these powerful models deep
neural Nets so now there is another
paradigm I wanted to describe which is
what if we cannot get a large database
of these examples or even better we want
to actually go beyond human capabilities
right if we actually want to beat humans
if we take that we are kind of bounded
by how good they are and this is the
reinforcement learning paradigm and for
instance in the game of gold this was
heavily used to for the recent result
that I forgot chief de la fere so it is
extremely simple how this works it’s
almost like obvious the first step is we
have an agent which is in this case the
top layer and this agent lives in an
environment so the imagine would be the
dog or or in the robot case the world
and this agent has a dog for instance it
wants to win at the game of God and the
way this works is quite simple the agent
observes the environment so it observes
let’s say the goal board or the door or
what one not and it can also act upon
the embark it can for instance play a
piece of the port that would be one
action right I play in position four
coma fight for instance in the in that
sort of game and eventually six I cling
to these the game finishes and there my
environment as creation whirring won the
game or not and after it’s been taught
whether or not it updates its ways to
ensure that essentially the the reward
that it gets the probability of winning
is maximized
so this agent is very very similar to
the neural network that was classifying
cancer dogs but in this case it takes as
the inputs these observations so it
would be a goal board
video games cleaner or not and the
outputs are simply the probability of
each actions or which joystick position
you want to play and through this
process you can achieve superhuman
performance in playing at I’ll go and so
on so very crucially it is most of these
algorithms have been developed
developing games um we just kind of
funny because as a kid I used to play
video games and now I sort of work on
making algorithms for video games I
don’t get to play but despite these
profound you know realizations what you
actually get to do when you do through
these two video games is being able to
scale because as you can imagine this
circle of observations and actions
initially it’s not very good so in other
words if you act randomly you would not
gonna get the works very often so you
have to play a lot in order to achieve
good performance but if you go out to
the real world
things will be very slow if you are to
build 10,000 products that you would
crash and eventually you learn how to do
this that wouldn’t scare everyone so
video games are very important because
you can’t run them in parallel in a
cluster and if you lose it’s not a big
deal but eventually you are testing sort
of ideas about intelligence through this
video guys which were designed by making
essentially humans you know imagine and
plan and so on so this is our sort of
research platform it has been so far
until recently Alex it’s gonna be in the
future as well but what’s the problem
well the environment if the environment
is very simple we will learn by simple
agents so the environment and the agents
go side by side and as you can imagine
the
is extremely complex so there are like
you know predators out there that might
want to eat you you need to energy so
you might want to eat plants or other
other human being not human beings
beings and and it’s very diverse and
very complex right so it seems like
creating environments and creating
agents goes both hand by hand and he
didn’t need that so one of the the ideas
here is if we are simulating the the
natural world and we want for instance
to get this guy to like survive the
winter we we need to simulate snow and
food and so on and that is a hard
problem by itself as well however you
know what you can actually try to do is
well we researchers not only build
agents but we are actually also seeing
about the environments that will make
agents advance and amongst many new
environments that people create last
year we created this environment which
is called the mine lab which looks like
three dimensional you know researchers
can set all sorts of tasks to test to
test ideas about planning memory
imagination and so on and I’ll show you
just a simple example where an agent
will essentially go around and try to
eat apples but do not eat lemons this is
another video where there’s actually
like more naturalistic looking video and
there’s an agent out there that is sort
of navigating these complex works
figuring out that apples might appear
again because they grow and so on and so
forth so these environments are becoming
more complex and as a result our agents
will also of course learn better they’re
now a sort of a final note it is kind of
obvious that we need AI it’s very
helpful to you know be able to have
translation systems you might have a
person
system in your phone like helping you
organize streets and so on so forth but
it also seems a friend that an AI needs
nature we need to encode these complex
rules for AI to be useful to us
otherwise we are just bound by the
complexity of the environment in which
agents living and so we done note them
I’d like to just thank you for your
attention I’ll be happy to chat with you
outside Thanks
[Music]
Please follow and like us: