Player is loading...

Embed

Embed code

Transcriptions

Note: this content has been automatically generated.
00:00:01
okay i guess we're going to start to be more less on time
00:00:06
can you really apparently there was some problem with the audio
00:00:11
okay thank you for coming i think we are expecting more people but as usual the recap it later
00:00:19
okay so thank you for joining this years uh innovation the an advisory board so
00:00:25
the first these uh join meeting between innovation they so with all of
00:00:28
you industry representatives and so on and you will see a more like an
00:00:33
overview of our activities and two mortar deposed or advisory board meeting
00:00:40
okay thank you for you to all of you thank you to
00:00:43
the advisory board members to coming from so far sometime
00:00:49
so you have the agenda you should have the agenda in front of you re start with a
00:00:53
short presentation of mine i promise eh uh like i do every year to be short
00:01:00
yeah i uh the uh then uh we will have several group presentations each of them twenty minutes so
00:01:06
we would go like this uh to the whole groups one by one twenty minute per group
00:01:13
interleaved with the coffee break at ten thirty uh_huh he into lobby and doing it next
00:01:19
time we'll have nobody to wells four lines of course but also for interaction
00:01:23
questions and the poster session you may have seen also the posters that we try
00:01:28
to limit the number of posters and the number of told to to minimum
00:01:32
so to to go does not to impress you with the scientific publications and so on but to give you a little
00:01:38
taste of what we are doing it that it yep at a four p. m. will
00:01:44
have a he's not arrived yet is going to write fin couple of hours
00:01:48
but told by a already evolved so who is one of the manager at uh those uh research in uh
00:01:56
sort of paris i forgot the name of the city there
00:01:59
a sort of parts cross with a sigh so airbus industry uh
00:02:03
we are currently starting collaboration with them want and collaboration
00:02:08
and already he will present us the the current uh challenges in uh in the
00:02:13
uh the domain of our human area i own no takes in general
00:02:19
so tonight at four uh_huh okay so inferior words in fifteen minutes
00:02:27
so each up again you know it's been here for twenty seven years twenty eight years and we keep growing
00:02:33
so i'm going to skip all my usual slides about showing all well we are doing financial wise
00:02:39
a project wise and so on and so on we are all doing well
00:02:43
so that's the key message otherwise we're sticking two or three a key missions
00:02:49
with maybe more less emphasis on one or two depending on to your of course research is our primary
00:02:57
priority that's where we put most of our activity and funding you will see you have
00:03:03
an overview of all the research activities going on the t. yep today uh training
00:03:09
so we're doing training miniature interaction with the p. f. l. and other
00:03:13
universities so teaching a welcoming intern intern students and so on
00:03:18
uh you will see and this would be the first time announcement
00:03:21
for officially first of september so it will come uh
00:03:25
two days earlier than planned first announcement of a new a high
00:03:29
master program that will be kick of a at a job
00:03:34
uh first of january uh_huh so this is all the training activities
00:03:39
and innovation and technology transfer it's something that is growing more and more every year
00:03:44
or you have an id the of all the startups are on the
00:03:48
job and more specific early around what is called e. bay arc
00:03:53
which is the aegis incubate awful technology transfer where we have more and more stuff up there
00:03:59
otherwise we're also dealing on the job site about we're dealing with that bigger
00:04:04
a company says but for everything regarding startups and that's where they
00:04:09
sell so many flow between you d. r. and it yep
00:04:12
in terms of realities license agreements collaborations on this where the money goes
00:04:20
but this year we we are very successful uh with one of those spinoffs
00:04:25
it was the second or third time that we had such a success
00:04:28
you may have uh some of you may have for that a key element which must be the somewhere see them on the
00:04:35
apply have left was both by a. m. s. in zurich which
00:04:39
is the main a sense of provider for i phones
00:04:43
and the key lemon was a bored by that company with the condition of remaining in
00:04:50
valet and martin more particular on it and uh now growing even more
00:04:57
and we we're quite lucky to have quite a some good options that can punch a
00:05:03
us so which we could catch him quickly before the move on with their future
00:05:12
okay in terms of organisation i'm not going to spend much time going
00:05:15
through all those boxes that you see there i'm going to skip
00:05:20
without and insulting them of course the left ah so left a pillar which is more i mean although we are
00:05:26
very proud of those people that's where we are doing
00:05:29
code development so interfacing between research and and industry
00:05:34
uh that's where we have all season main that's where we have program managers helping researchers
00:05:38
to develop research projects with that you as a doctor or a european commission
00:05:45
n. s. f. c. c. innovation and so on and so on the got advise or industrial relationships
00:05:52
communication and marketing and so so just get that one and i'm also going to
00:05:57
skip the right one since you will have one told per group today
00:06:02
again in a nut shell today we about ten research groups covering different areas
00:06:08
a speech and what your processing to biggest for historical reasons about twenty eight people know
00:06:13
a computer vision and owning although for also uh is going to
00:06:17
complain is group exchange name recently so it was recorded
00:06:22
machine learning envisioned surprisingly or group led by then you'll get a car you will
00:06:30
or he of the from him a later on social computing
00:06:33
site interaction between i. c. t. and social sciences
00:06:36
perception activity understanding by your signal processing which is a brand new group
00:06:42
that by andre so dealing with the processing of all by all send sauce roberta learning interaction
00:06:48
uncertainty quantification and optimal design dealing with small data
00:06:53
the world is not full of big data remember that all the time most
00:06:58
of the unsolved problems have to deal with small data not big data
00:07:03
there are very few problems where you have access to big data computational by you
00:07:08
imaging for that by michael live link by your matrix security and privacy
00:07:13
sometimes when the group is successful enough we also spin of a research centre
00:07:18
affiliated with each yeah this is the case with the buyer matrix centre
00:07:22
which is more responsible for certification of biometric devices into future all
00:07:27
those devices will have to be certified by one single body
00:07:31
it would not be enough to claim as we all do interfere that we are doing well that ninety nine percent this is
00:07:39
plus it possibly outing people all devices a telecommunication devices and divide by
00:07:44
i. t. u. electronic devices are certified by these c. e. simple
00:07:50
biting into future uh a i devices would have to be satisfied by you
00:07:55
neutral body and i think set so that is a very well positioned to holes to that kind
00:08:00
of body and maybe even better than geneva martin could also be a better place for them
00:08:07
i'm not joking matter language processing the natural complement to speech recognition and
00:08:14
energy informatics is a brand new group also which is starting
00:08:18
first of september now which is going to work in collaboration with uh another research
00:08:24
institute in martin you cram sold other shares on energy when you see button
00:08:30
so today yeah but we started late right
00:08:34
so therefore progress was to lecture us thirteen permanent senior researchers twenty eight pasta talks
00:08:40
thirty five pages to the um software development ingenious sixes does that mean
00:08:45
pen admin staff thirty the trainees and and master students
00:08:50
so about hundred full time equivalent hundred thirty two people in in residence on over
00:08:56
currently forty seven active projects from n. s. f. c. t. i. industries asked
00:09:00
the european commission we we have many of them as you probably know
00:09:05
us so we are probably participants of three big us i up our projects
00:09:12
you will hear some move them in being mentioned in the top
00:09:17
so we're is one in the speech processing and two around by matrix of course not surprisingly
00:09:24
uh about a hundred twenty publications barrier this is an interesting number which was published by the state of away
00:09:30
a few months ago so this is not my number um but uh the state of uh the is reporting that
00:09:36
a job is currently accounting for thirty seven thirty six percent of all
00:09:39
the are in the uh fault in university level instituting that uh_huh
00:09:48
i may skip that seems we i'm already running out of time
00:09:51
but our strengths is independence inflexibility we are affiliated with
00:09:55
the p. f. l. but we are independent and very very flexible so we can move very fast on the ball
00:10:02
regarding projects regarding collaboration with industry we're extremely fast we have
00:10:07
all the tools that we need in house or research
00:10:10
the legal adviser was tech transfer developers
00:10:15
we are so uh at the origin of many open source
00:10:19
libraries some of them you may know apply torch
00:10:24
so they're too big a i open source libraries into will write one it's coal pencil fro
00:10:30
managed by google and the other one call by taught by taught is originating from each up as just one example
00:10:36
but there are many many other examples like that so that currently about fifty open source libraries added yep
00:10:42
and then so about fifty opens open public data sets i would say as available for me yet
00:10:49
and on average we receive about more than one thousand request to access those data sets
00:10:55
those requests that screens by ah says i mean people and uh last year
00:11:01
about five hundred request out of one thousand five hundred requests where granted
00:11:06
on average about five hundred people asking to have access to
00:11:10
some of our pencils databases and uh using it
00:11:14
one of them one of the most popular on and use this is called the i mean
00:11:17
data some of us we know it very well i mean augmented multiparty interaction database
00:11:23
which was awarded to tenure impact i will buy a c. m. the last year or two years ago
00:11:29
ten years impact i worked by easier for one single database
00:11:35
so i'm not following my eyes as you can get them right so i can yeah
00:11:41
i can i think i can skip this one i just mention it right some of our key contributions so if
00:11:46
i look back over the past the last ten years if i asked myself what they've been to maine
00:11:52
the key contributions of the jobs the reasons why a job is very well known
00:11:56
worldwide you see a few of them they mentioned the open source software porch
00:12:02
now call quite often use this by phase book is what they call the the phase bouquet i engine
00:12:08
a bop fall by matrix beat is also very fame use it's on the base on that platform that
00:12:13
we got granted a two weeks ago and you are european project to develop the only one
00:12:19
it was quite concave pick competitive because it's a twenty median funding the only
00:12:24
your pen projects to develop an a. i. europe in part for so the project is called a i for you
00:12:31
so if you got a i for the the digit for
00:12:34
you you will land on the first the website
00:12:40
website of that european projects that is going to stop probably before the end of the year european they
00:12:48
i applied for and that one is going to explode to some of the beat wrestles him
00:12:54
i mentioned uh me i mean data there are many other
00:12:59
so he job is independent flexible affiliated with too strong patton
00:13:04
knows worldwide very would just not in switzerland already worldwide
00:13:08
we have a been a longstanding player in the area of all the our
00:13:12
research area that i will talk about today we have excellent staff uh_huh
00:13:18
you can look at the profile at that age index the list of publications list
00:13:22
of pages to the inciting i'm very very proud to have such an excellent
00:13:27
group of people he added yep we have very sustain port for
00:13:31
you yeah sometimes like in any other places we fail
00:13:36
difficulties by losing funding it happens uh if you look at all part for you this is something we
00:13:43
share tomorrow with the advisory board in private but this output for you is very very diversified
00:13:48
yeah about twenty different funding sources so when we lose one it's not good but is not table either
00:13:57
so that's another advantage basically we try to present our surveys do one stop shop
00:14:02
in the i'm i'm sure we're not the only one in the world
00:14:06
but we just want people to you that this is what we are so right now we are positioning yourself and you
00:14:11
will hear about that into that and to fume next man's we're ready now moving to the next step neck speed
00:14:18
growing much faster and bigger and we want to present all surveys the competence centre in
00:14:24
up by a guy in switzerland and i hope i'll advisory board member will
00:14:30
support this initiative and will push us in that direction
00:14:36
one what i have one more minute right but i think a suppressed a we'll talk about that
00:14:40
right the i master so but just a little signal to our industry parton us here
00:14:47
we have seen from our industrial partner knows over the last year us and this was also by my colleague also for
00:14:53
yeah we must be around so you should talk to him if you meeting so you came up with this idea
00:14:58
of developing a new a. i. master program around a i asked uh
00:15:04
so we have peace it's we have seen for many years that
00:15:06
industry is using a job is a nice reserve reserve wanna
00:15:11
off balance in yeah so each time they grow then you easily if come to egypt and
00:15:17
hire people from us and we are happy about that it's our duty it's fair enough
00:15:22
so when we have a spinoff being both by a bigger company they need to hire way that
00:15:27
they go to find a a skilled people they come to egypt and tasty on people
00:15:32
which is fair enough and that keeps happening then we headed the id that okay we have to
00:15:37
work all together with industry any job to train those people together and share the burden
00:15:44
so that's the model we adopted together with the support financial support of
00:15:47
the state of valet and say go in but on sick call
00:15:51
it was not easy to convince them they finally got confused and they give us funding for that
00:15:56
to kick off a new program based on the the apprenticeship id incessant
00:16:01
apprenticeship is working for if you don't know the apprenticeship in
00:16:04
in u. s. r. in other countries incessant any some model that works very well right many
00:16:09
other countries and virus for that so people here okay yeah they they are trained
00:16:15
i think it's it makes training between in diversity or academia and industry
00:16:22
and that's what we're going to do here so it's a apprenticeship at the master level
00:16:27
so people can hear they already have an engineering bachelor and they will follow courses
00:16:32
you will see the curriculum that we have put together and this is all the courses are
00:16:36
not available on the web with exercise and so on everything is perfectly lined up
00:16:41
so that's the kind of course meeting two ninety e. c. t. s. created after eighteen months
00:16:46
this is going to be very very to have to our students because they
00:16:50
will have to do all that why i so working in industry
00:16:54
that's the idea of apprenticeship but at the end of this curriculum they will
00:16:59
be do you have a happy owners of an a. i. master in
00:17:03
engine well master in a i engine yeah but acting say best yeah i'm down

Share this talk: 


Conference program

Introduction by Hervé Bourlard
BOURLARD, Hervé, Idiap Director, EPFL Full Professor
29 Aug. 2018 · 9:03 a.m.
Presentation of the «Speech & Audio Processing» research group
MAGIMAI DOSS, Mathew, Idiap Senior Researcher
29 Aug. 2018 · 9:22 a.m.
Presentation of the «Robot Learning & Interaction» research group
CALINON, Sylvain, Idiap Senior Researcher
29 Aug. 2018 · 9:43 a.m.
Presentation of the «Machine Learning» research group
FLEURET, François, Idiap Senior Researcher, EPFL Maître d'enseignement et de recherche
29 Aug. 2018 · 10:04 a.m.
Presentation of the «Uncertainty Quantification and Optimal Design» research group
GINSBOURGER, David, Idiap Senior Researcher, Bern Titular Professor
29 Aug. 2018 · 11:05 a.m.
Presentation of the «Perception and Activity Understanding» research group
ODOBEZ, Jean-Marc, Idiap Senior Researcher, EPFL Maître d'enseignement et de recherche
29 Aug. 2018 · 11:24 a.m.
Presentation of the «Computational Bioimaging» research group
LIEBLING, Michael, Idiap Senior Researcher, UC Santa Barbara Adjunct Professor
29 Aug. 2018 · 11:45 a.m.
Presentation of the «Natural Language Understanding» research group
HENDERSON, James, Idiap Senior Researcher
29 Aug. 2018 · 2:03 p.m.
Presentation of the «Biometrics Security and Privacy» research group
MARCEL, Sébastien, Idiap Senior Researcher
29 Aug. 2018 · 2:19 p.m.
Presentation of the «Biosignal Processing» research group
RABELLO DOS ANJOS, André, Idiap Researcher
29 Aug. 2018 · 2:43 p.m.
Presentation of the «Social Computing» research group
GATICA-PEREZ, Daniel, Idiap Senior Researcher, EPFL Adjunct Professor
29 Aug. 2018 · 2:59 p.m.

Recommended talks

Mobile multimedia services based on user attention profiles
Francesca De Simone, EPFL
1 Sept. 2011 · 10:03 a.m.