# MIT China Summit 2018

Data: 11-01-2025 21:46:46

## Lista de Vídeos

1. [MIT China Summit: L. Rafael Reif](https://www.youtube.com/watch?v=5tPgOGK3huo)
2. [MIT China Summit: Chunli Bai](https://www.youtube.com/watch?v=0DaO-sz_bb8)
3. [MIT China Summit: Gang Chen](https://www.youtube.com/watch?v=gTTXjHRG4HM)
4. [MIT China Summit: Robert Desimone](https://www.youtube.com/watch?v=JIhyWyLNfx0)
5. [MIT China Summit: Jian-Wei Pan](https://www.youtube.com/watch?v=NgzJOkMz2LI)
6. [MIT China Summit: Nergis Mavalvala](https://www.youtube.com/watch?v=8vikez91czo)
7. [MIT China Summit: Yang Lan](https://www.youtube.com/watch?v=w3BRm6SL8RQ)
8. [MIT China Summit: W. Eric L. Grimson](https://www.youtube.com/watch?v=6VqSQxLNWCc)
9. [MIT China Summit: Xiao’ou Tang](https://www.youtube.com/watch?v=1NnuAb_f4RQ)
10. [MIT China Summit: Dina Katabi](https://www.youtube.com/watch?v=9nVZwLqG6aI)
11. [MIT China Summit: Daniela Rus](https://www.youtube.com/watch?v=DqimU04S06s)
12. [MIT China Summit: Daniela Rus and Qingfeng Liu](https://www.youtube.com/watch?v=5ck1oJcLMOA)
13. [MIT China Summit: The Quest for Intelligence Panel](https://www.youtube.com/watch?v=rHB3tscNUkU)
14. [MIT China Summit: Martin A. Schmidt](https://www.youtube.com/watch?v=hbvB_kC9JcU)
15. [MIT China Summit: Maria Zuber](https://www.youtube.com/watch?v=pziELAxLpo8)
16. [MIT China Summit: Retsef Levi](https://www.youtube.com/watch?v=jN5XiIIDveg)
17. [MIT China Summit: Carlo Ratti](https://www.youtube.com/watch?v=vHfKHFKBS4A)
18. [MIT China Summit: Siqi Zheng](https://www.youtube.com/watch?v=vis0sNTcZcM)
19. [MIT China Summit: Melissa Nobles](https://www.youtube.com/watch?v=oVztM-gjujo)
20. [MIT China Summit: Next Generation Leaders -- MIT Technology Review Innovators under 35](https://www.youtube.com/watch?v=OOozm3yA9Y0)
21. [MIT China Summit: Yasheng Huang](https://www.youtube.com/watch?v=C6P97__rJBQ)
22. [MIT China Summit: Robert C. Merton](https://www.youtube.com/watch?v=uYABw7h07tc)
23. [MIT China Summit: Jessica Tan](https://www.youtube.com/watch?v=EGqadx4azgA)
24. [MIT China Summit: Sanjay Sarma](https://www.youtube.com/watch?v=bZMBPtpdZGs)
25. [MIT China Summit: New Visions of Education and Research for the Benefit of Humankind](https://www.youtube.com/watch?v=KPjvV1iywpk)
26. [MIT China Summit: Richard Lester](https://www.youtube.com/watch?v=t5SAnTPh8f4)

## Transcrições

### MIT China Summit: L. Rafael Reif
URL: https://www.youtube.com/watch?v=5tPgOGK3huo

Idioma: en

you
[Music]
[Applause]
good morning everyone morning I'm ruff
arrived I'm mi t--'s president and I'm
honoured and delighted to welcome you to
the first MIT China summit plenary
conference I would favour that this
truly distinguished audience includes
some of China's most influential and
innovative business leaders investors
entrepreneurs government leaders and
academics and I'm proud to say that some
of those people are also graduates of
MIT as we begin I wish to express our
deep thanks to the Chinese Academy of
Sciences a superb institution of higher
learning in science and technology for
joining us in this opportunity for
dialogue and exploration I'm also deeply
grateful to these events generous
sponsors I offer my grateful admiration
to everyone who worked so hard across so
many international time zones to bring
this event to life in special credit
very special credit goes to associate
provost richard lester and director of
the China summit bethe to pool for the
leadership and brilliant execution so
please join me
actually three an execution up to this
moment let's see about the rest of the
day were all eager to dive into the
fascinating content of today's program
so I'll take just a few moments to
reflect on what inspired us to create
this summit in a way it is surprising
that this is the first MIT China summit
because the ties between MIT and China
have been long-standing as we heard in
the video I might be enrolled its first
student from China a hundred and forty
years ago when MIT itself was still very
new since then students faculty alumni
staff and mi t--'s many great friends in
China have built a wonderful bridge of
connection a bridge that with busy
traffic in both directions a bridge of
great benefit to the people of MIT and
hopefully to the people of China as well
as just a few examples forty-one members
of the current MIT faculty were born in
China MIT enrolls about 11,500 students
total and more than seven percent of
them come from China most of them in a
graduate program in fact more MIT
faculty and students come from China and
from any other nation besides the United
States and through our online teaching
platform EDX more than a hundred and
five thousand Chinese students have
enrolled in courses from MIT join the
other way MIT sends more than a hundred
and fifty students from to China every
year to learn often through internships
with leading Chinese companies and many
MIT faculty members have built their
careers doing research in China or with
Chinese
there are many other links between us
that I will mention just one more
it is the MIT China connection that has
without question touched the most people
and that's the fact the important fact
the director of mi t--'s McGovern
Institute for brain research has for
five seasons served as a judge on the
most popular game show in China the
brain Bob DeSimone is very well known in
China in fact back at MIT tourists
visiting from China often stopped him
when he's just walking across campus so
I'm delighted to advertise that in just
a few minutes
professor DeSimone sitting in the front
row in person will join other amazing
brains from China and MIT here on stage
professor Lisa - a pleasure to have you
here with us I gotta take advantage of
any MIT famous person I can find these
connections between MIT and China has
been in place for so long that we could
simply have allowed them to continue to
flourish and attended but we saw this
summit as a powerful way to respond to
an emerging reality mit has always been
focused on what we call inventing the
future and over the last few decades MIT
has become a truly global University as
well as an American one with retiring
from around the globe a hundred and
thirty five nations and our graduates
live and work around the world to our
faculty count on the ability to build
constructive international relationships
what's more our mission statement
inspires us to tackle humanity's great
global challenges how to advance clean
energy fight climate change
reverse environmental degradation how to
design sustainable sea
and final route to water and food
security for all how to fight cancer
Alzheimer's disease and other challenges
of an aging population and how to
establish ethical guidelines for
advanced technologies so that they are
used for the good of all this will be by
the way a center of focus of the new
Schwartzman college of computing at MIT
there are problems and these are
problems of concern to everyone on earth
and problems that no nation or
institution and hope to solve alone so
it is increasingly clear that if we at
MIT continue to aspire to invent the
future and to make progress against
great global challenges we must work
with the most talented people all over
the world who share those aspirations
now people with advanced skill and
expertise in science and technology and
with such serious aspirations exist all
over the world but no other nation has
as large a pool of first-rate scientific
and technical talent as China this is a
moment of some friction between our two
nations but will come to you through the
medium of this summit in the spirit of
wishing to learn together through our
mutual benefit in order to benefit the
world
the MIT faculty you'll hear from today
are Trailblazers they may not be quite
as famous at Bob DeSimone
but all of them are stars in their field
yet were equally eager to learn from our
colleagues and counterparts from China
in academia and in business as everyone
in this room understands China today is
a formidable global player in many
aspects of science and technology on the
academic side China recently moved to
number one in the volume of research
papers published on artificial
intelligence in quantum computing 5g
networks and high-speed rail technology
being developed in China is world-class
in mobile payment and in facial and
spoken language recognition Chinese
companies are global leaders because
they have capitalized on their advanced
algorithms and advantages in scale and
data access in recognizing that such
technologies are capable of both
enormous social benefit and terrible
abuse it will be vital to define their
ethical application in another area of
leadership the Chinese government is
also investing significantly in research
and directly supporting startups in
China has unrivaled capacity to rapidly
ramp up large-scale production of
advanced technology technology products
which has the fast lane from innovation
to market even a respect for China's
enormous strengths we come to you eager
to imagine how we can best make progress
together on the serious problems of the
world we see our own strength as rooted
in a national culture of openness
opportunity and entrepreneurship inspire
by an atmosphere of intellectual freedom
and supported by the rule of law and
most importantly we know
we reach new heights of creativity by
United brilliant reuniting brilliant
talent from every corner of society in
every corner of the world I'm eager to
see all that we may learn from one
another and all that we may jointly
achieve of course there is only so much
we can aspire to do in a single summit
but I'm encouraged and thrilled to have
established such a promising place to
begin and now I am extremely honored to
introduce professor chun-li Bey
president of the Chinese Academy of
Sciences
[Applause]

---

### MIT China Summit: Chunli Bai
URL: https://www.youtube.com/watch?v=0DaO-sz_bb8

Transcrição não disponível

---

### MIT China Summit: Gang Chen
URL: https://www.youtube.com/watch?v=gTTXjHRG4HM

Transcrição não disponível

---

### MIT China Summit: Robert Desimone
URL: https://www.youtube.com/watch?v=JIhyWyLNfx0

Transcrição não disponível

---

### MIT China Summit: Jian-Wei Pan
URL: https://www.youtube.com/watch?v=NgzJOkMz2LI

Transcrição não disponível

---

### MIT China Summit: Nergis Mavalvala
URL: https://www.youtube.com/watch?v=8vikez91czo

Transcrição não disponível

---

### MIT China Summit: Yang Lan
URL: https://www.youtube.com/watch?v=w3BRm6SL8RQ

Idioma: en

[Music]
you
ok ladies and gentlemen girls and boys
please get seated again our session will
start now let me first of all introduce
myself my name is yang LAN have been a
media person for more than 25 years and
I kept asking myself the question why am
I here well this is of course this is
the gathering of some of the most
brilliant scientists and entrepreneurs
and we are here to share some of the
frontier of the knowledge about our
planet about the universe about our
brain and above all about our future so
I figured out the reason I'm here
because I'm curious
I think being curious is a very
important and precious part of human
nature
the curiosity brought me and my team to
search for the implication and history
of artificial intelligence
two years ago we traveled to five
countries interviewed more than 80
people in this field from neuroscience
to artificial intelligence from
companies and and to researchers etc and
we produced a documentary called in the
search of artificial intelligence which
was broadcast last year now we are in
production of the second season so I
guess that's the reason I'm here
actually my search of artificial
intelligence started from MIT in the
beginning of 2016 that we started our
interviews from the school from computer
science and artificial intelligence
laboratory and I'm at MIT and I'm very
happy to see some of the familiar faces
here so 60 years ago researchers at MIT
and elsewhere lit the fuse on the very
big question what is intelligence and
how does it work and can machines think
so artificial intelligence basically
took off in MIT and of course in the
following session you will see some of
the most brilliant minds from both US
and China to elaborate and deepen our
understanding of human intelligence and
machine intelligence another finding
that I had from MIT is that we all know
that's a place that gathered some of the
most curious and the smartest minds of
the world and we Chinese think sometimes
there's a very fine line between genius
and craziness so when I was visiting see
cell for example I even find that
building is a little bit peculiar the
inspiration was I was told from dancing
you know so the building's like they're
rocking row so I told my cameraman I
said can you see that it's rock and roll
it's dancing buting and my camera one
cameraman looked up and said are you
kidding I thought they were drunk
uh anyway ecstasy ecstasy of the probing
and it and forever exploration of
frontier of knowledge and innovation how
exciting that is so the following
session will showcase some of the most
exciting vision and goals of MIT quest
for intelligence a multidisciplinary
project of neuroscience and machine
learning will share the view with some
of the key players in this field and
they will they are not only going to
present their latest work and research
and findings and their implications but
also above all how they will bring this
transformative technology to address to
some of the most challenging issues of
our planet and of human society let me
introduce them one by one so that their
presentations will not be interrupted by
my introduction any more our first
speaker will be Professor Eric Grimson
who is the chancellor for academic
advancement of MIT he's a professor of
computer science
and the medical engineering and
specialize in computer vision and
medical image analysis he will be
followed by mr. Tong Xiao ou professor
from Chinese University of Hong Kong and
the founder of sense time Shang honker
gee professor town will give a
presentation about his research and its
applications he's a expert in computer
vision and the deeper learning he's also
got his PhD from MIT and then we'll have
professor Dina kitabi who's a professor
of Electrical Engineering and computer
science from MIT the presentation is
entitled from wearables to Invisibles
she's the director of MIT Center for
wireless networks and mobile computing
and then the Nilla Russ who is the
director of computer science and
artificial intelligence laboratory C
cell of MIT will make a presentation on
engineering intelligence by the way
although she champions the crazy
building she's very reasonable person
and then we'll have two conversations
one is between danila with a Ching from
Leo who is the founder and chairman and
CEO of I fly tech could a student fate
which is a very successful company in
China and abroad and then I'll moderate
a conversation between ya Qing Jung the
president of Baidu
and Professor Grimson so let's enjoy our
first speakers presentation professor
Eric Grimson please
you

---

### MIT China Summit: W. Eric L. Grimson
URL: https://www.youtube.com/watch?v=6VqSQxLNWCc

Idioma: en

you
[Music]
I want to tell you about a major
initiative that you just heard about
launched earlier this year at MIT we
call it the quest for intelligence and
it involves faculty and students from
virtually every department of the
Institute and as you heard it builds on
over 60 years of work from MIT from the
founding of the field of artificial
intelligence by researchers at MIT and
colleagues elsewhere let me start by
giving you a vision of the kinds of
grand challenges that we're tackling and
the kinds of places where we think AI
and machine learning are going to make
big impact over the next few decades we
see breakthroughs in a our machine
learning coming from many places new
algorithms new mathematical theories
faster computers but a key idea for us
is to couple an understanding of the
neuroscience and cognitive science of
human intelligence with advances in
computational methods not only will this
increase our understanding of our own
intelligence but we think that current
and future insights into how our brain
and mind work can dramatically influence
future applications of AI a wonderful
example of a moonshot here using this
perspective is to use our knowledge of
how young children learn to serve as a
basis for creating a computational
equivalent imagine building a computer
system that starts like a newborn and
learns to be a two-year-old maybe not
misbehave the way a two-year-old does
but it learns at least the intelligence
of a two-year-old such a platform could
not only help us devise and execute
experiments on understanding human
intelligence
it could also serve as the foundation
for the development of a new and
possibly radically different generation
of machine learning methods and it could
provide insights into educational
environments that better support
learning but we want to do more than
just explore the foundations of human
intelligence we want to apply those
insights to virtually every domain of
science engineering design and social
sciences and to seek solutions that have
impact on people's lives can we use
models of how we as humans process
natural language to create systems that
combine information from many sources
images lab reports doctor's notes
genetic studies family histories all to
create highly accurate detection methods
and personalized treatment plants for
patients suffering from cancer heart
disease neurodegenerative ailments and
other situations we also see AI and
machine learning methods impacting
discovery of new knowledge in other
domains for example the task of
determining the best sequence of
reactions to produce a new chemical
compound like a drug is an art rather
than a science but by training on
hundreds of thousands of examples of
chemical reactions and their outcomes
new machine learning systems at MIT are
showing a remarkable ability to predict
a chemical reactions major product and
by inverting that process one can if
dramatically improves the the ability to
create new drugs by knowing what you
want to get to and working backwards to
decide what are the precursor chemicals
you want to use to get there same ideas
apply in material design synthetic
biology many other domains new AI and
machine learning systems are not only
going to impact science and engineering
they're going to impact organizations
financial social medical civil so
systems at MIT can look at credit card
data to determine who is most likely to
be delinquent and allow a bank to think
about how best to handle risk management
and in this and many other cases is
going to be crucial that systems can
access data without invading our privacy
so workers at MIT have created new
techniques that allow mi s re allow
machine learning systems to Incred to
attack
roxrite wrong phrase to access encrypted
information without breaking the
encryption in order to extract
information while preserving that
privacy something that we think is
crucial in order to preserve our rights
thus
MIT is coordinating this Institute wide
initiative to test these challenges
focus focusing on the science and
engineering of intelligence but also on
that application in every domain and
doing it while worrying about the
ethical moral and societal implications
of those methods the quest has two parts
to accomplish this the first part the
core focuses on basic questions does
neuroscience and cognitive science
provide new insights into learning
algorithms can one-shot learning be
captured algorithmically can we use
computation as a basis for understanding
our own intelligence do insights from
young children how they learn actually
help us create the next generation of AI
systems the bridge focuses on making
machine learning tools easily applicable
to every discipline at MIT basic science
engineering finance social science
design let me give you a couple examples
of both of these parts so understanding
our own intelligence we think might lead
to very different ways of thinking about
smart applications and this idea is
letting us attack some really
fundamental questions about human
intelligence how is it that babies learn
from so few examples how is it that
young children can recognize objects and
words in a naturally embedded situation
separating out the word in the object
without any supervision how do we
achieve common-sense reasoning how do
you deal with perceptual awareness the
fact that we can perceive things in
natural settings let me give you one
example of a dressing this fundamental
system current deep learning methods
show impressive performance on very
specific tasks but they need hundreds of
millions of training examples and yet
you can show a two-year-old ten examples
of something and she learns it very
quickly how is this one-shot learning
done well colleagues like Josh Tenenbaum
and others at MIT are exploring this
idea using an approach that suggests
that perhaps young children have in
their heads the equivalent
of a simulation engine they can simulate
what happens with the physical setting
by using that one can create a
probabilistic program that when seeing a
new situation can very reliably predict
what is going to happen when you couple
that with models of intent you
potentially have the way of learning
from very very small numbers of examples
very effectively and we'll see if we can
build the computational equivalent of a
young child the bridge it's the second
element and it aims to make machine
learning tools easy to use by anybody
anywhere at MIT or anywhere else my way
of thinking about it is we'd like those
tools to be as easy to use in the future
as today we use Word or Excel as a
simple tool to quote president rife we
want to create bilingual experts that
use the language of machine learning as
easily as today we use the language of
mathematics in a few minutes you're
going to hear two great examples of
applying these kinds of a tools in other
domains I want to just give you two very
simple examples of ways in which people
are using them to young faculty at MIT
Elsa Olivetti and Stephanie Jagielka
combined to tackle an interesting
problem Elsa wants to create new
materials and she'd like to do it by
simply specifying theoretically what are
the properties you want of those
materials but to create the material is
still an art by building a natural
processing or natural language
processing system their system has read
over a million articles from the
scientific literature and has learned to
detect patterns between precursor
materials and crystal structure why
should that matter by now inverting the
process one can describe properties of a
material and the system will generate
suggested recipes of how to create those
new materials many of which have already
been verified in the literature well the
project is not done it suggests that
design and materials will now be much
easier to do and the same idea would
apply to drug design and other creations
of other processes we give you a second
example
FinTech an area of obvious interest to
many people Andrew Lowe has take a
machine learning tools and applied them
to credit card data looking at different
banks and using this to predict which
customers are most likely to be
delinquent most likely to cause problems
for the banks and showing as a part of
this process that not only can he
predict how the banks can manage their
risk but that in fact each bank should
use a different process there is no one
solution to it these kinds of results
these kinds of tools are going to change
the way finance is done by applying
these machine learning techniques
throughout all aspects of how we deal
with human life in one last example most
of my examples have been in software but
we also know that we need to build
physical devices physical
implementations of these most current
speech and vision systems and you're
going to hear some examples of that rely
on neural Nets through densely
interconnected modules that interact
through simple processors the problem is
those nets are huge they're expensive
and they use a lot of energy and so most
cellphones will not use embedded
processors they will simply ship the
data up to the cloud it's process there
and it shipped back down we'd love to
create processors that are designed
specifically for neural net computation
so now that trying to Khasan and others
at MIT have done exactly that
creating chips designed to implement
neural net computations that are 7 to 10
times faster and use 95 percent less
energy allowing them to be embedded in
household devices and cell phones and
will create a rapid amplification of the
ability to use these devices in many
places around all of our daily
activities thus the quest will look at
the fundamentals of understanding our
intelligence it will look at
applications of intelligent systems in
every aspect of science engineering
design and social sciences and what we
hope to see is over the next few decades
this will have an impact that changes
the way we think about
we've create new knowledge apply that
knowledge tackle the world's great
challenges and finally do it in a way
that pays attention to the ethical moral
and societal implications of using all
of that and we hope we'll turn it around
to use our insights to think about what
does it tell us about how we learn how
young children learn and how might we
actually educate them better and with
that that gives you a sense of mi t--'s
quest for intelligence che che
[Applause]

---

### MIT China Summit: Xiao’ou Tang
URL: https://www.youtube.com/watch?v=1NnuAb_f4RQ

Idioma: en

[Music]
good morning ladies and gentlemen Thank
You MIT and Thank You Raphael for
inviting me to be here it's a huge honor
he would not take it in hand till noon
on top when Zuzu go to work ja ja da na
na so he was intended to the talker who
you know may like job you know with your
dad to go see go how does your aunt
Angela Lee na young man now you know
what the human aesthetic Etobicoke
injury be a lot easier so the title of
my talk is in the mood for ai ai ai ai
the pronunciation of this word in
Chinese is AI which means love so
basically I borrowed this name from a
very beautiful Hong Kong movie in the
mood for love and I know many of you
have not really seen this movie so for
the next two hours I invite you to watch
the movie the whole movie
[Music]
if you see the two lines down there
is we are analyzing every frame of the
movie who attacked in the mood of the
movie in a particular scene we're doing
face detection face recognition for
tracking and we are analyzing in the
relationship what they are doing you
know with each other in this movie so
basically what is the purpose are we
teaching the Machine to watch movie for
us certainly for bad movies that will be
useful but mostly we would like to watch
movie by ourselves so the real
application is not ready to teach the
Machine to watch a movie actually we
don't really have the time for the whole
movie so let me skip this so what is the
application so one of the application is
advertising so we want to you know find
out the right place the right mood to
insert other word heisting you know this
is a sense time I our platform
[Music]
you know since time has me existing for
like ninety two hundred years already
you see from the movie that's our
autonomous driving car and of course the
real adult Iseman I want to plug in is
this this is my son because I understand
that there are many professor from MIT
is here today I just want you to
remember my son this face one more time
[Music]
someday he's going to apply to college
and I need the reference letters
actually the last week he called home he
said mom and dad I have a green news I
got a b-plus in math and we said oh okay
and he said mom p+ is a great great
great for any subject and we said okay
fine then a few days later he got a she
in French so I think just take whatever
you can you can do just don't push for
it so anyway so that's why I need a
lighter and no pressure is but I know
he's not going to make it why my tea and
the letter is really for application to
Harvard
[Applause]
so let's come back to the subject of I
really sorry but congratulations to
Boston Red Sox a week more than a week
ago they became the championship
champion again and I think they bring
the first championship in 1998 then you
take them 86 years to we another wine so
it's really the dedication the hard work
you know the perseverance all this
really you know happened to go through
all those years it's kind of like a I
you know from for 50 years it never
worked
then suddenly it starts work and so
since then they have been winning I
don't know it's five championship so
it's really hard work and I'm a big fan
of Red Sox and all the other big sports
teams in Boston I was at MIT from 91 to
97 so I watch a lot of their their games
and at that time it is really were
miserable and from 91 to 97
they got totally zero championship
altogether and then from 98 to 218 you
know it's 11 altogether so this through
this big data analysis I have to draw
the conclusion it is really not the hard
work you know the dedication it's maybe
just me you know maybe I should have
left MIT sooner but actually it does
have something to do with sports we
actually use AI to do an analysis of the
activities you know tracking the you
know actions of the statistics so sports
is one area that one industry we are
working on using AI and another area we
work on is they are
this is the first mobile phone
application that actually using a are in
real game you know you have you when
you're taking a we are taking a real
video scene and you can actually
interact with game characters in real
time on a mobile phone and this is I
think the King of glory is the most
popular game in the world to work with
the Oppo and Tencent who produce
this realistic game playing applications
so as I'm showing you a few applications
actually we are working on IP a lot more
than this we are also working on
autonomy driving medical imaging and
quite number of other industries and
even you know AI is really related to
quantum mechanics quantum computing
there we heard it before
so since I'm Rudy has been working in
not just not ai ai is really not
independent industry so AI plus is so
that's really our work we're working
with many many industries to help to
improve the efficiency of those
industries so that's our work and we are
very fortunate to became the fifth
national open innovation platform for
artificial intelligence and computer
vision intelligent vision right after
Baidu $0.10 Alibaba and I fell attack so
it's a huge honor that today I'll have
the five national platform we have three
right here so one area I want to
particulars
is education
thank you for any industries the
challenge winning is the most important
thing so we actually published the first
textbook for high school along with a
lab together with our partner we hope
not just attract the interest of student
to AI but also to all stamp subject and
we hope you know in the future we can
work with MIT on this and finally we are
very very fortunate to be able to work
with MIT to form this alliance to work
together on AI
I want to thank MIT for coming to China
and for coming to China at this
particular time actually I think a
couple months ago when I met professor
with he told me that they are coming to
China for the MIT summit and many people
asked him him are you crazy
at this time and he told me exactly this
is a time to go to China
so thank you
[Music]
[Applause]
so I believe AI is really the purpose of
AI is really helping us to break
boundaries so before we have so that
might happen to break boundary among
different scientific subject and
secondly it can also help him to break
the boundary between AI and the
traditional industries and finally most
importantly is helping us to Brick's a
boundary between countries so I think
MIT along with many other top university
in the US has made American a great
country and I hope in the future we can
work with MIT to make humanity great
again
thank you
[Applause]

---

### MIT China Summit: Dina Katabi
URL: https://www.youtube.com/watch?v=9nVZwLqG6aI

Idioma: en

[Music]
you
hello everyone it's my great pleasure to
be here my name is Dena kitabi and I
want to tell you about health care and
particularly how we can move for about
two Invisibles so I'm really interested
in automating health care that is I want
to bring the care from the hospital and
the clinic to the patient's home and if
you think about it this is really
important particularly now because we
have Asian societies so many people are
old there are so many patients and there
are not enough doctors and nurses so one
way to automate healthcare is we need to
use video conferences between the doctor
and the patient I mean we all know how
to do this at the problem you can't
really do this because you can't do it
at scale you can't have the doctor
having videoconference with all of his
patients all of the time that would be
too much and there are not enough
doctors not enough time what we really
need is the ability to monitor health in
the home continuously it is you wanna
monitor the patient breathing heartbeats
motion days sleep or physiological
signal detect when health is degrading
and allow the doctor so they can
intervene and pay attention to the
patient who needs it but how do you do
is how do you do this I mean you might
be saying oh but wait a minute right
monitoring health at home is really
cumbersome today if you want to monitor
something like breathing you want
someone to wear a chest band or a nasal
poke if you want to monitor sleep you
ask them to wear all of these sensors on
their head and body and sleep with them
if you want to March a false for grandma
or grandpa you ask them to worry upon
dn't and push a button and if you want
to monitor a patient's Parkinson you
have to ask them to wear accelerometers
and move with them this is not a happy
picture you don't want your patients to
be wearing these device
what if somebody comes and tells you
that we can model all of these things
and many more without a single sensor on
the person's body this is exactly what
my group at MIT does we have invented a
machine or device that looks very much
like a Wi-Fi box it sits in the
background of the home and it marcher
breathing heartbeat gate sleep and many
other things without a single sensor on
the person just using the wireless
signal in the environment toilet it
might be me Andre okay how can how can
you do this how can you matter how can
amount to your breathing your heartbeats
you're sitting in there without touching
it but if you think about it you guys
are here sitting in a sea of wireless
signals you agree with me on this Wi-Fi
cellular every single thing everything
that you do affects the electromagnetic
waves around you you took breath it
affects the electromagnetic waves
you took a step changes the
electromagnetic waves and our device is
smart is sit in the background and uses
machine learning advanced models of
machine learning to analyze these
electromagnetic waves so that would know
it took a breath or you moved or
whatever else in fact it can matter
breathing heartbeat gate sleep and many
other things so I want to show you a few
videos to illustrate this here so here
you see the home wireless did not spread
inside the home they actually reflect
off our human bodies because our bodies
are full of water and they come back to
our device which analyzes them using
machine learning here it would detect
the fall and alert the caregiver via
text email or phone message let me show
you some examples from our lab
technology doesn't always work either
for MIT
yeah so what we have here our device is
going to march is this person okay so
this red dot is where the device thing
this person is but the device is not
even in the same room as a person it is
actually in the adjacent office
monitoring him from behind the wall so
imagine if somebody is marching us from
behind the wall from Adafruit but
wireless signals traverse walls and as
he walks you can see how the red dot is
tracking him now remember this is purely
based on how the wireless signal
interacts with his body he doesn't have
any sensor no cell phone no
accelerometer proximity accurately as
you can see so it turned out that if you
can measure don't know that if you can
measure gait speed you actually can
measure very important health metric
because gait speed is important for
Parkinson's for multiple sclerosis for
sarcopenia for so many of the motion
related diseases but also chose out that
it is a predictor of hospitalization in
heart diseases that congestive heart
disease or COPD so today he emerges this
in the hospital when the patient goes
there with stopwatch but you can measure
this 24/7 in the home what else can we
measure sleep Hey sleep is a very big
problem for many people now if you think
about it you can see why our device
would be able to measure sleep so the
device sees a person as he walks to bed
when he stops tossing around in bed when
he steps out of bed so we should be able
to measure sleep based on on motion
something called actigraphy
but it turns out that we can measure
sleep much deeper than that we can
measure sleep stages so when we go to
sleep our brainwaves change and we enter
different stages awake like sleep deep
sleep
and REM or rapid eye movements these
sleep stages are very important for
sleep disorders
there are also important for variety of
diseases for example rapid eye movement
disturbances are related to depression
the slow waves during deep sleep are
related to Alzheimer's today if you want
to march or sleep station would you do
you send the patient you sent the
patient is a sleep lab and they put all
of these sensors on his body and face
and ask him to sleep like this okay say
I mean of course you can see that he's
not happy sleeping like this is not
great so what can we do let me show you
our scenario so this is our device
transmits a very low-power wireless
signal monitors of reflection using
machine learning and spits out the sleep
stages throughout the night it knows
when this person is dreaming in the REM
stage okay what else can we monitor
using wireless signals in the
environment so we can monitor you
breathing so this guy is sitting like
you guys and what you see here is
nothing but his inhales exhales inhales
exhales and we asked him to stop
breathing to hold his breath and you see
the signal stays at the steady state
because he exhaled he did not inhale so
let me zoom in on the signal now what
you see here is the same breathing
signal these are the inhales these are
the exhales and you see some blips on
the signal these are not noise these are
actually his heartbeats beat by beat you
can count them so when we started
working on this so many people from the
healthcare industry got very interested
and today we have deployed with more
than 200 patients in their homes those
patients are in Parkinson's
in Alzheimer's some people have
depression some people have pulmonary
diseases like COPD and we are working
with hospitals and actually foundation
like the Michael J Fox foundation on
changing health care using this device I
want to show you some of our results
with the patients so here I'm going to
show you a patient who live in assisted
living community and I'm going to show
you the difference between how the
patient moves and how the nurse moves so
let's see this so as you can see the
nurse is faster than the patient but
also the nurse is smoother the patient
is slow but also his wiggling though
what we do with this so we deployed with
multiple patients let me show you some
results from our Parkinson patients so
you see the device up there we call it
the Emerald device and what you see here
is two hours of trajectory from this
patient so we take these trajectories
and we look at the life of the patient
so here I'm taking those trajectory and
plotting them on this graph in terms of
what the patient was doing and let me
explain this graph so every circle is 24
hours vo is midnight and 12 is noontime
the innermost circle is the first day of
the experiment and the outermost circle
is the last day of the experiment this
is eight weeks in the life of this
Parkinson patient and what you see here
is that his life has changed a lot
during these eight weeks like for
example at the beginning of the
experiment here look at the blue which
is sleep look how bad it is at the
beginning of the experiment very
disrupted sleep but as the experiment
goes on at the end the outermost circle
the blue becomes less disrupted so sleep
is less disruptive you see all I think
you see all of that green stuff this is
the patient sitting on his chair doing
nothing so this patient is very sedan
chewy which is really bad for older
people you can see
so for example here yellow at 8:00 a.m.
in the morning which is when the nurse
comes to help this patient go into the
bathroom to do his toilet in and
dressing so he's very dependent on his
nurse let me show you the impact of
medication so what you see here is the
speed of the walking of the patient as a
function of the hour in the day so what
you see is that there is an inflection
point where his pee suddenly increases
around 5:00 a.m. can you guess what
happens
takes his medications so you immediately
see the impact of that medication on him
and let me plot for you the speed of his
wife which is of course she's faster
because she doesn't have Parkinson's but
it doesn't show the same inflection
point so you can actually see the impact
of medication and start changing the
dose of the medication and adjusting the
medication for that particular patient
how about breezing abnormalities so here
you see apnea so the first thing you see
the patient's breathing and then he
stops breathing
this is apnea and then breezes again and
stops breathing and so on though I want
to end by going back to the beginning
and asking you what else can we do here
whose wireless signal so you see that
red dot it tells us that this person
location and that he's not moving he's
standing there but it doesn't tell me
whether he's actually standing or
sitting and also as he moves I was able
to show you the dot moving with him but
you don't know whether he moves with his
right foot or left so let's see our most
recent results ok so here the bigger
frame is the the vision of the wireless
device and the small frame is the vision
of the camera inside the room and as you
can see we get the full skeleton when he
sits on the chair the wireless device
through the wall knows that he's sitting
on the chair he stand up he walks we see
every single movement
which give us much more information
about motion completely without putting
any device through wireless signals
through walls so if you think about it
what we did here is that we change the
vision most people think about digital
health about wearables but note that
there is a new thing this is the
invisible we can just use the wireless
signal in the environment to monitor
health and that is how I think we can
automate healthcare thank you very much
[Applause]

---

### MIT China Summit: Daniela Rus
URL: https://www.youtube.com/watch?v=DqimU04S06s

Transcrição não disponível

---

### MIT China Summit: Daniela Rus and Qingfeng Liu
URL: https://www.youtube.com/watch?v=5ck1oJcLMOA

Transcrição não disponível

---

### MIT China Summit: The Quest for Intelligence Panel
URL: https://www.youtube.com/watch?v=rHB3tscNUkU

Idioma: en

[Music]
you
okay Thank You danila and Qing phone
very fast speeches and I hope one day
our AI translators can help us to bridge
different languages so I'd like to
invite our next two panelists one is a
professor kimsen again and the other is
joshing Jung who is the president of
Baidu and also you know an expert in
this field too please gentlemen we have
about 15 minutes before that everyone
can go for lunch please
yeah please so I guess gashing will or
choose to speak English right because
your Chinese has some Seattle accent
what kind of Juji I turn to and her us
that it will heart I'm human but we are
SWANA what are you and me or do be call
you what you're saying security well
great so we were talking about
collaborations and and also competitions
and also some of the new frontier of a
eyes research oh my first question is
mostly for Eric basically so we have
learned that the study of human brain or
human intelligence has given the
inspiration to artificial intelligence
but then what's the other way around
how does AI help us to understand better
how human brain works so it's a great
question and I think there's a couple of
ways it can happen one is that
computation can provide an experimental
basis you can simulate things in a
computational system that might suggest
experiments you want to try on people to
understand what works or what doesn't
work and I think AI systems can expose
surprising results connections that you
hadn't seen before that then allow
neuroscientists or cognitive scientists
to go back and look at do humans think
that way there may be somewhere we don't
but
maybe somewhere what an AI system does
let's neuroscientist cognitive
scientists asked questions about the
human system that deepens our
understanding well something that people
usually say is that since we know so
little about the mystery of human
intelligence there's no way that
machines can copy that but there are
also other arguments by saying that the
machines don't have to understand how
the system works so long as they can
simulate the reactions what what do you
think I'm always nervous about that
argument and maybe it's just you know
training as a scientist if you think
about current deep learning systems
they're impressive what they can do but
we don't always understand exactly what
insight is leading to that result and as
a scientist I would really like to know
what are the things that are happening
if I make a small change to that system
how does that deal with it you're right
that we don't know a lot about what goes
on inside of our brain but having a
system that just solves a problem
without giving us an explanation of why
is it challenged and if I may I think if
you think about applications of AI in
the United States today many healthcare
companies make decisions about what is
covered for us for a patient based on an
AI system I'd like to know how it made
that decision I'd like to know what was
the intelligence behind that that
conclusion so I'd always prefer to have
something that does allow an explanation
of how it got to that decision is that
also the training of yeah a team right
always understand why it works that way
not just how absolutely and yes it is
your turn to you know because we we are
now also talking about in the deep
learning as a very effective algorithm
to advance many technologies and their
applications but is that the only way
what do you see as the future other
paths in our you know intellectual or
scientific research well in the first of
all I want to commend MIT you know Eric
Raphael arrived an entire and
mighty management for your foresight and
vision to start this audition request
indeed in the future of AI you know will
be a confluence of many different things
burning its bring science neuroscience
and physics let me also suggest you add
other disciplines like you know ethics
and the policies because you know that
we'll need a lot of the collaboration
and knowledge that whatever field and
I'd be thinking too about oh by the way
today
you know you normally in sessions like
this I consider myself as a visionary
for the longer term today you know after
session especially the last one and I
feel I'm very pragmatic and a short term
in any case now if you're looking at the
next five years 10 years indeed machine
learning or deep learning as useful as
valuable as it is there is you know
there's major limitations and we need a
better understanding of course of how
brain and how mind works and and and
also to bring that knowledge back to and
our existing and a technology and
algorithms our discovery understanding
of bringing in mind it's actually pretty
superficial but even you know that
knowledge has now been well applied to
what do we do today
I recently I read some papers and a book
in the human brain in the research just
our understanding of the memory model
actually you know we've made a huge
progress but we really have not used
that to the things like that on striving
but we talked with this in the morning
there that to kind of some short-term
memory which are very helpful making
decisions one is the declarative
explicit memory and the other is the
implicit and non-declarative and the
first one is actually quite well
understood and we already taking
some of that in our convolutional neural
network or REM but the second part is
instincts which which are kind of muscle
memory which are not well understood in
fact when we develop technologies for
driving it it's a second part of that
memory and decision process it's more
important but you know it's totally not
what we're doing right now so I think
there's a lot of things we're going to
need understand from different field
especially bringing research the other
thing is on the on the quantum computing
how are we talked about that will
address a change and disrupt and our
existing architecture and will certainly
change it how are we going to compute
and provide services but you know again
it takes a while to see
commercialization that's one area we
hope to work with the third thing we
talked about before is 5g which is not
exactly I but will help
AI to to apply in a much faster manner
infrastructure 13 Isis it's very
interesting to to see our study things
from different perspectives for example
when I was doing the documentary on AI I
was interviewing a professor tomas so
that's right from MIT so I asked him so
how can you explain why our human
memories show so bad we have spent hours
hours in in learning but we just forgot
many of them but then Tommaso answered
by saying that you should ask that
question in a different way why our
human brain is so good at learning you
know sometimes forgetting things it's
also part of our very precious
intelligence so I think changing
perspectives sometimes give us total
different insight about intelligent
itself so my next question is actually
how do you see the importance of a
collaboration between research
institutes and companies the the private
sector of this area
what are the things that both sides put
on the table Eric would you stop so I
think it's a great question and I think
many ways to look at it but for me one
of the biggest ones is a university
explores ideas they take a long term
view they will try unusual things that a
company cannot always do because their
shareholders want to make sure that
they're doing things well what a
university isn't good
often at scaling so the engineering that
needs to take place and that's not meant
as as a negative it's meant as a
positive to take an idea and make it
really work well in in massive scale is
something that I think companies do much
better now of course companies have
their own research labs they're not that
they're independent but I think that
trade-off that back and forth of an idea
being explored in a university and then
working with the company to say all
right how do we make it actually work
24/7 365 days a year how do we make it
work in all of these conditions that
feeds back to the university to ask new
questions and also I think the in terms
of the scale and impact of the
possibilities here in China is just
beyond imagination maybe a Qing can
explain a little bit about the
potentials of your collaboration with
the city of Xuan for example a new city
which will be a another Center a future
in northern China and how does that
apply to the application of a I heard
you saying I completely agree with the
Eric's the model in a university and an
industry there are different different
stages you know when I worked for GT
labs or Microsoft for of I do I find the
collaboration with edenia hugely
valuable by the way I spent five years
in Boston between 1990 and 95 I used to
think that was my thoughts for the poor
responses of the poor performance of the
sporting team today I knew in a shallow
was responsible there's a different
different stage for you know for example
just in the other in fact
machine learning AI seneschal all the
early breakthrough in theory in the
models and algorithms started from
University nineties Sam use Toronto and
back Montreal and and it was was
companies that actually scale and apply
I remember one of the first application
of deep learning and speech was from
Geoff Hinton he actually worked in
Microsoft Research and made that as you
know as and he's the one of the early
tangible results shown that show an
interesting initiative here for some of
you from Massachusetts from Boston you
probably know this is a dish Asst a move
or from the Chinese government a beauty
a city for the future so we have an
opportunity to design the entire
architectural the city whether is energy
or roads or transportation or public
services so we actually worked with them
to help you know build a road for the
future the city for the future in terms
of not only the course autonomous cars
but also a intelligent and Connecticut
roads that's a central part of the
future city so if you have a chance to
visit you man you will see our mini
buses so it's a level four automation
with no drivers wheels or brakes it's
already operational there were just the
course you can buy your coke or or or
your soda from there and and also all
the other infrastructure is being peeled
from I just visit there about two weeks
ago and there are very few traffic
lights because the oh the streets are so
connected and so smart so the cars
actually can afford to be summer so it's
quite interesting to see how AI is
applied at that scale that's right let
me just add you know we are in the mists
of
huge a transformation in terms of
digitization and in fact the previous
speakers already touched on that point
you know the first way is to digitize
contents you know video imagery
documents a speech in fact I spent many
years to be that video compression you
know transmission the suddenly was to
digitize Enterprise CRM CRPS your this
is a logic all this processes now we are
in the midst of digitization which is
this resident of the world of the
physical world
are a body and are bring a course roads
and everything so this wave is going to
transform economy the world society much
more profound way okay my next question
is about the challenges and concerns you
scientists tend to believe that our
future will be improved by science and
technology is that true to my
interviewees from the science background
to write anything on my guest Bo because
they all write something very rosy about
our future the possibilities in living
much better life but when I interviewed
you know professors and scholars from
liberal arts they all look very worried
so what are your worries do you what do
you think we should start addressing to
some of the questions what are some of
the priorities that you want to address
- I know ya Qing is also attending an
international conference on the
accountabilities and research abilities
give us more information on that yeah
professor please two points one is to
think about I'm gonna call them ethical
issues ethical may be the wrong word
because they're implies as a right or a
wrong but I think one of the things we
have to think about and I hope that
universities companies government's come
together on this is to think about what
are the implications of using these
systems where are there places where we
should
use an AI system because that should be
a decision still made by a human or as
we saw with one of the examples I think
from from Daniella where are places
where the combination can make a better
decision than just one person or another
but we really have to think about what
are the ethics of using these places and
the other one is something that that
president rife said yesterday but I
think is really important and that is
what is the disruption on jobs
what is that disruption going to do sit
societally and MIT has a very large
working group looking at and I have to
do this carefully Raphael right it is
the work of the future not the future of
work will always have work in the future
but what what do we think about where
will this change how do we deal better
with training people for that how do we
do our children to understand how to
deal with that but I would say thinking
about standards for ethical use is
really important and it should be global
and thinking about how do we plan for
that future now so that when those
systems are there they're working with
people not replacing people you know I
am optimist and the technologists so I a
long term I'm confident this wall served
the society and the human being possibly
as big companies and we need to be aware
we need to be responsible and we need to
pivot the direction to the wheel and we
redesigned 10-4 so you know we recently
we publicize our principles of AI you
know called ethics or principles you
know AI should be a safe and a
controllable and we need be is
empowerment not a replacement of human
writing right an AI wall even more
option choices for a human being now to
not a limit that so so I think those are
the the principles we have in fact you
know after this after lunch I'm going to
fly to San Francisco
to attend a conference which is a new or
partnership set by a few large companies
got a partnership for AI for the benefit
of people in a society so it's a
Microsoft Google Amazon Facebook Apple
Baidu is the first Chinese company to be
part of this partnership so it's not
only about technology it is about
implication an impact of technology on
the society people but let me also say
this I think part of a human
intelligence is to go beyond what we
have today it is to invent technologies
to build tools but also to be smart
enough to use these tools so will will
survive and thrive after this the
presidential administration I guess well
my last question since we only have like
a minute and a half is to address our
audience I think we have many thought
you know leaders entrepreneurs and and
mostly young people what's your
suggestion to them when they are
planning their own personal life or
Korea for the next five or ten years
what's your best advice for them Wow I
have two sons 28 and 30 and I haven't
figured out how to tell them to do
reference letters anymore that's good to
know I would say you know we'll borrow a
statement from from president rice I
think the next generation needs to be
bilingual earlier bilingual was he
needed to know math and something is
handy divided lingual in languages not
just English and different different
categories or disciplines but the
bilingual now is that no matter what
field you want to go in even creative
fields creating things if you don't have
a solid grasp of computation a solid
grasp of what AI and machine learning
can do for you you're going to be left
behind and
so rather than saying I can't do
computing you should be able to say I'm
going to use computing I need to
understand it it is a new mode of
thinking that everybody has to have
without it you will be left behind with
it you'll thrive watching have a 2 kids
both actually I got a Columbia night
they asked my advice and I give them a
device never listen I keep telling them
machines learning you must be learning
always always learn be open-minded be
adaptive
actually what happened it's not only
about learning self its ability to learn
new things knowledge we have today or
become irrelevant in five years in ten
years so its ability to learn things
that you open minded not only about
computer science but also about other
fields but ah well I would I would say
that for scientists and technologists I
think liberal arts or humanities studies
can also be a very good bilingual skill
to add to that because as Einstein said
mystery stays in the middle of true art
and true science so when we are probing
the very field of the unknown I think we
should have both the heart and brain and
then the mind to lead us ahead well
thank you very much to gentlemen and I
would like to ask all of you to give a
big round of applause to our wonderful
panelists and all the speakers this
morning thank you
now our human brain tell us we need food
enjoy a lunch thank you

---

### MIT China Summit: Martin A. Schmidt
URL: https://www.youtube.com/watch?v=hbvB_kC9JcU

Transcrição não disponível

---

### MIT China Summit: Maria Zuber
URL: https://www.youtube.com/watch?v=pziELAxLpo8

Transcrição não disponível

---

### MIT China Summit: Retsef Levi
URL: https://www.youtube.com/watch?v=jN5XiIIDveg

Transcrição não disponível

---

### MIT China Summit: Carlo Ratti
URL: https://www.youtube.com/watch?v=vHfKHFKBS4A

Transcrição não disponível

---

### MIT China Summit: Siqi Zheng
URL: https://www.youtube.com/watch?v=vis0sNTcZcM

Transcrição não disponível

---

### MIT China Summit: Melissa Nobles
URL: https://www.youtube.com/watch?v=oVztM-gjujo

Transcrição não disponível

---

### MIT China Summit: Next Generation Leaders -- MIT Technology Review Innovators under 35
URL: https://www.youtube.com/watch?v=OOozm3yA9Y0

Transcrição não disponível

---

### MIT China Summit: Yasheng Huang
URL: https://www.youtube.com/watch?v=C6P97__rJBQ

Transcrição não disponível

---

### MIT China Summit: Robert C. Merton
URL: https://www.youtube.com/watch?v=uYABw7h07tc

Transcrição não disponível

---

### MIT China Summit: Jessica Tan
URL: https://www.youtube.com/watch?v=EGqadx4azgA

Transcrição não disponível

---

### MIT China Summit: Sanjay Sarma
URL: https://www.youtube.com/watch?v=bZMBPtpdZGs

Transcrição não disponível

---

### MIT China Summit: New Visions of Education and Research for the Benefit of Humankind
URL: https://www.youtube.com/watch?v=KPjvV1iywpk

Transcrição não disponível

---

### MIT China Summit: Richard Lester
URL: https://www.youtube.com/watch?v=t5SAnTPh8f4

Transcrição não disponível

---

