
many revolutions have had a dramatic
impact in our lives the French
Revolution for example change the course
of history by replacing absolute
monarchies by modern democracies closer
to my heart I was born just a few months
after a military coup and I grew up
under a dictatorship that took the lives
of thousands of people one of my first
memories as a child is walking through
the unpaved streets of my hometown with
my mom carrying pots filled with
drinking water to take home for the
family while my mother didn’t finish
elementary school and my dad didn’t go
to college
they both knew the importance of
education so one day I remember my mom
told me that if I had homework to do I
wouldn’t have to do the chores little
did she know how much advantage I would
take of that deal
my favorite subject at school was math I
remember falling in love with algebra
and geometry and I remember playing with
everyone in my family was thrilled when
one day I was awarded the Presidential
Scholarship
I was terrified
I had to shake the hands of the dictator
to receive it but not all revolutions
are political in fact many revolutions
have their roots in dramatic scientific
advances that have also changed our
lives like mine I remember I was
finishing my engineering studies in
Chile and there was this team that came
to the u.s. to the Atacama Desert the
Moon Valley because they wanted to test
a robot that they wanted to send to them
I was just fascinated by the challenges
of being able to build machines that
could perceive and navigate the world in
the same way we do and so I moved to
Berkeley to do a PhD in computer vision
machine learning and of course robotics
which are the three key ingredients of
artificial intelligence and there I just
had an awesome time developing
algorithms for flying helicopters on
moving platforms my move to Hopkins was
quite an interesting story I had applied
to many faculty positions in computer
science including Hopkins and one day I
get an interview from Hopkins biomedical
so I arrived to the interview I’m there
with Marie sax and I had to tell him
that the last time I had taken a biology
class was in high school being such a
wonderful man as he was he told me that
the Revolution of the 21st century was
going to be to make medicine a
computational science and he asked me
whether I wanted to join that revolution
humbled by his answer I joined Hopkins
be Amy Ann Marie was right 15 years
after our conversation we are beginning
to see the signs of a dramatic
revolution in data science think about
IBM’s Watson or maybe google’s alphago –
machines that have beaten the world
champions in jeopardy and go or think
about uber who is revolutionizing the
car industry by having cars without
drivers or Cortana or Siri that are
allowing us to talk to machines or maybe
Amazon go who is revolutionizing our
shopping experience by creating
supermarkets that have no lines and
check out but what is enabling these
dramatic changes well on the one hand
data is just growing at an explosive
rate from the millions of pictures and
videos that you and I take every day to
the massive amount of data that maybe
astronomers collect to be able to
decipher the origins of the universe
clearly today’s data is too big and too
complex to be analyzed by humans that is
where machine learning comes in the goal
of machine learning is that you have
data and you want to make predictions
from that data so take for example a
picture and you would like to predict
who is in that picture and in this case
is my daughter
so we humans are really good at making
those predictions but computers are not
and so they dream of artificial
intelligence is to be able to create
machines that have those prediction
abilities as we humans do and to be able
to do that what you do is that you give
a lot of training examples with the kind
of data that you want and the kind of
predictions that you want the machines
to make and then you teach the machine
to make its own predictions pretty much
– they learn like a baby would learn how
to recognize objects now the roots of my
work in computer vision can be traced
back to my elementary school years I
remember that my math teacher would come
with a gigantic ruler and compass and we
just spent hours teaching us about lines
and squares and circles in my research I
use lines to make predictions from data
suppose that I take all of the pictures
of my son it turns out that they can
reasonably be approximated by just a
line of course in big dimensions so if I
take the pictures of my daughter then I
have another
and so if you’ve got pictures of many
people you’ve got multiple lines and
that insight is the basis for the
algorithms of one of the simplest
the origins of my work in biomedical
data science can be traced back to that
conversation with Maurice acts 15 years
ago
sadly Murray passed away exactly one
week though but his dream of making
medicine a computational science is more
alive than ever about three years ago I
began working with a company in Belgium
called my Diagnostics and with a great
team of medical doctors and engineers
from American Hopkins and the goal is to
revolutionize the blood test so imagine
how it works today you go to the doctor
they tell you you need a blood test
you get it the next day you get the
results the day after and then you have
to go to the doctor again to be
diagnosed what if you had a small device
where you could get the results on the
spot at the first visit to the doctor to
be able to do that we need to get some
blood image it in a small device and
then use computer vision algorithms to
detect count and classify those cells
and that is going to be the next
revolution in point-of-care testing in
another project I am working with ten
universities to revolutionize
information I’m sure you all know what
information is about and you can tell
what’s informative from what’s fake news
but do you know what information is to a
computer how our computer measures
information well if you open your
computer I’m sure you’ve seen that the
files are maybe ten megabytes or
something like that
well that’s exactly how a computer
measures by information in
in in data but that’s actually not very
useful let me give you an example
suppose or imagine that you have a
picture and that picture has lots of
people gather in front of a supermarket
and suppose that there is a truck that’s
passing by
well that track could be the track that
killed hundreds of people in Nice
two years ago in which case is very
informative for terrorist or it could
simply be that the track is offloading
the goods in the supermarket and in that
case is not very informative the next
revolution in artificial intelligence
will be to develop machines that can
figure out what’s informative in data
versus what’s not informative and so for
that we’re developing a framework that
is going to redefine information so that
you can characterize the amount of
semantic content in that data set so
this is a little bit like creating a
digital Sherlock Holmes the applications
of these could be endless if we could
record maybe the activity of all neurons
perhaps we’ll figure out a very
connectivity of the brain if we could
figure out all of the activations and
connections between genes maybe we’ll
figure out a cure of cancer or maybe in
50 years from now hospitals will have
all of the data from all patients in the
world and so when you go to the doctor
there will be a question and answering
system that will ask for your symptoms
then you’ll be able to go into a
database to search for all patients that
had similar symptoms and then you will
figure out what treatment was given and
which work and which didn’t saving a lot
of cost in potentially saving millions
of lives step-by-step as we are forging
the future I cannot help but
reminiscence about the many revolutions
that have changed my own life don’t be
afraid
the revolution that is about to forge
your own future it will change your life
in the same way the data science has
changed mine thank you [Applause]
impact in our lives the French
Revolution for example change the course
of history by replacing absolute
monarchies by modern democracies closer
to my heart I was born just a few months
after a military coup and I grew up
under a dictatorship that took the lives
of thousands of people one of my first
memories as a child is walking through
the unpaved streets of my hometown with
my mom carrying pots filled with
drinking water to take home for the
family while my mother didn’t finish
elementary school and my dad didn’t go
to college
they both knew the importance of
education so one day I remember my mom
told me that if I had homework to do I
wouldn’t have to do the chores little
did she know how much advantage I would
take of that deal
my favorite subject at school was math I
remember falling in love with algebra
and geometry and I remember playing with
everyone in my family was thrilled when
one day I was awarded the Presidential
Scholarship
I was terrified
I had to shake the hands of the dictator
to receive it but not all revolutions
are political in fact many revolutions
have their roots in dramatic scientific
advances that have also changed our
lives like mine I remember I was
finishing my engineering studies in
Chile and there was this team that came
to the u.s. to the Atacama Desert the
Moon Valley because they wanted to test
a robot that they wanted to send to them
I was just fascinated by the challenges
of being able to build machines that
could perceive and navigate the world in
the same way we do and so I moved to
Berkeley to do a PhD in computer vision
machine learning and of course robotics
which are the three key ingredients of
artificial intelligence and there I just
had an awesome time developing
algorithms for flying helicopters on
moving platforms my move to Hopkins was
quite an interesting story I had applied
to many faculty positions in computer
science including Hopkins and one day I
get an interview from Hopkins biomedical
so I arrived to the interview I’m there
with Marie sax and I had to tell him
that the last time I had taken a biology
class was in high school being such a
wonderful man as he was he told me that
the Revolution of the 21st century was
going to be to make medicine a
computational science and he asked me
whether I wanted to join that revolution
humbled by his answer I joined Hopkins
be Amy Ann Marie was right 15 years
after our conversation we are beginning
to see the signs of a dramatic
revolution in data science think about
IBM’s Watson or maybe google’s alphago –
machines that have beaten the world
champions in jeopardy and go or think
about uber who is revolutionizing the
car industry by having cars without
drivers or Cortana or Siri that are
allowing us to talk to machines or maybe
Amazon go who is revolutionizing our
shopping experience by creating
supermarkets that have no lines and
check out but what is enabling these
dramatic changes well on the one hand
data is just growing at an explosive
rate from the millions of pictures and
videos that you and I take every day to
the massive amount of data that maybe
astronomers collect to be able to
decipher the origins of the universe
clearly today’s data is too big and too
complex to be analyzed by humans that is
where machine learning comes in the goal
of machine learning is that you have
data and you want to make predictions
from that data so take for example a
picture and you would like to predict
who is in that picture and in this case
is my daughter
so we humans are really good at making
those predictions but computers are not
and so they dream of artificial
intelligence is to be able to create
machines that have those prediction
abilities as we humans do and to be able
to do that what you do is that you give
a lot of training examples with the kind
of data that you want and the kind of
predictions that you want the machines
to make and then you teach the machine
to make its own predictions pretty much
– they learn like a baby would learn how
to recognize objects now the roots of my
work in computer vision can be traced
back to my elementary school years I
remember that my math teacher would come
with a gigantic ruler and compass and we
just spent hours teaching us about lines
and squares and circles in my research I
use lines to make predictions from data
suppose that I take all of the pictures
of my son it turns out that they can
reasonably be approximated by just a
line of course in big dimensions so if I
take the pictures of my daughter then I
have another
and so if you’ve got pictures of many
people you’ve got multiple lines and
that insight is the basis for the
algorithms of one of the simplest
the origins of my work in biomedical
data science can be traced back to that
conversation with Maurice acts 15 years
ago
sadly Murray passed away exactly one
week though but his dream of making
medicine a computational science is more
alive than ever about three years ago I
began working with a company in Belgium
called my Diagnostics and with a great
team of medical doctors and engineers
from American Hopkins and the goal is to
revolutionize the blood test so imagine
how it works today you go to the doctor
they tell you you need a blood test
you get it the next day you get the
results the day after and then you have
to go to the doctor again to be
diagnosed what if you had a small device
where you could get the results on the
spot at the first visit to the doctor to
be able to do that we need to get some
blood image it in a small device and
then use computer vision algorithms to
detect count and classify those cells
and that is going to be the next
revolution in point-of-care testing in
another project I am working with ten
universities to revolutionize
information I’m sure you all know what
information is about and you can tell
what’s informative from what’s fake news
but do you know what information is to a
computer how our computer measures
information well if you open your
computer I’m sure you’ve seen that the
files are maybe ten megabytes or
something like that
well that’s exactly how a computer
measures by information in
in in data but that’s actually not very
useful let me give you an example
suppose or imagine that you have a
picture and that picture has lots of
people gather in front of a supermarket
and suppose that there is a truck that’s
passing by
well that track could be the track that
killed hundreds of people in Nice
two years ago in which case is very
informative for terrorist or it could
simply be that the track is offloading
the goods in the supermarket and in that
case is not very informative the next
revolution in artificial intelligence
will be to develop machines that can
figure out what’s informative in data
versus what’s not informative and so for
that we’re developing a framework that
is going to redefine information so that
you can characterize the amount of
semantic content in that data set so
this is a little bit like creating a
digital Sherlock Holmes the applications
of these could be endless if we could
record maybe the activity of all neurons
perhaps we’ll figure out a very
connectivity of the brain if we could
figure out all of the activations and
connections between genes maybe we’ll
figure out a cure of cancer or maybe in
50 years from now hospitals will have
all of the data from all patients in the
world and so when you go to the doctor
there will be a question and answering
system that will ask for your symptoms
then you’ll be able to go into a
database to search for all patients that
had similar symptoms and then you will
figure out what treatment was given and
which work and which didn’t saving a lot
of cost in potentially saving millions
of lives step-by-step as we are forging
the future I cannot help but
reminiscence about the many revolutions
that have changed my own life don’t be
afraid
the revolution that is about to forge
your own future it will change your life
in the same way the data science has
changed mine thank you [Applause]
Please follow and like us: