Player is loading...
Final pitch, team innovAItor
Lonneke, Idiap Research Institute
Thursday, Aug. 26, 2021 · 3:16 p.m. · 12m 19s
Creating new ideas, products, or services that hit the market is a challenge. Innovation teams have a hard time trying to think out-of-the-box and coming up with exactly that novel product that fills the users’ needs. We will create a AI tool that helps companies to innovate by learning from large time-stamped corpora how new concepts are created and building a model that incorporates domain-specific knowledge and relevant social media context to track trends and users' needs and wishes.
Embed
Transcriptions
Note: this content has been automatically generated.
00:00:01
yeah
00:00:03
okay that's it for me
00:00:11
i think that much of a lot of this i put it out
00:00:19
okay yeah
00:00:24
huh okay if it's come from ah it's it's it's it's it's a it's it's
00:00:36
okay we can kick the
00:00:42
okay well thank yes the the front line
00:00:46
okay welcome everybody i'm happy to uh oh
00:00:52
and uh well i've
00:00:57
happening with the hackers at the ah fit back okay
00:01:05
sell fucking chance so um yeah had very happy to introduce to you are
00:01:11
a actual called innovator it's a data driven innovation too and a chance predict
00:01:17
let me first start by introducing our team for very diverse team in many different ways
00:01:23
um so i'll present all of us so in the middle you see me i'm language processing
00:01:27
expert i've been in the field for about twenty years and now group leader at it yep
00:01:31
that we shall for our business the is not bridgette was working on the algorithms behind the the program
00:01:37
ah there was gloria nah our social media experts and
00:01:41
yeah and this uh uh if responsible for our interface
00:01:44
i would also like to thank our industry partners and a special
00:01:47
thanks goes to work and thank you also for being here today
00:01:50
uh we had a lot of a valuable feedback from you in several
00:01:54
different sessions during and a little bit before also the i. c. started
00:01:58
we had a beverage and followed company also tell uh giving us some feedback and we have lots of informants and a lot
00:02:04
of support and i'd like to sign this people i had an education about the sure i talked to an information signs not profit
00:02:10
institution and i all talked about our product and whether they would be a
00:02:14
willing to use this at about that would be useful also the icy mentors
00:02:18
and you get technical stuff a lot of things goes to u. i. we had to responses and immediately when we ask questions a very great
00:02:25
and four track of course for giving us food all the time it was really really nice and it will be hard to go
00:02:30
back to normal life afterwards so yes innovation it's vital uh they
00:02:35
are destructive innovations that wipe out entire business as we know that happens
00:02:39
so there's a need to innovate fast had of competition and out of the box
00:02:44
thinking is the one thing that's really hard to do when you're under time pressure
00:02:48
also consumer needs are so quickly changing it is really scary
00:02:54
so what if you had someone who could read off your internal data
00:02:59
process all relevant so should media content discover trends come
00:03:02
up with a new ideas predict trends also for the future
00:03:06
i'm present all of that to you in a nice an orderly fashion and if possible like tomorrow yeah
00:03:12
that is something you really cannot ask for any of you and please they would go crazy
00:03:17
so what is our solution our solution isn't a i tool it is based on cognitive
00:03:22
theory so what did we think we think how do people come up with new ideas
00:03:27
well they take existing concepts for example you have carmel on the other hand you have coffee you know all these both these
00:03:33
concepts but bringing them together created something new which was a big success for one of the companies that we're with thinking off
00:03:40
so with computational creativity and a i we have some
00:03:43
additional thing we have we can use a lot of data
00:03:48
we can use all of this data remember everybody postings on social meet yeah
00:03:52
there's lots of ideas floating around that lots of people talking about different concepts
00:03:56
um we can discover patterns there and find trends but what's most important that
00:04:01
we just not not looking only at these patterns and trends but with putting them
00:04:04
together to be able to produce new ideas and produce new ideas based on the
00:04:09
patterns that we've seen before ideas new ideas might be things that humans never fine
00:04:15
so we sat our system we would like to predict the future right
00:04:20
but how can we say it's good in the future we don't know
00:04:23
about the future so we evaluated first on the present so we sat
00:04:28
let's pretend that we are at the end of the nineteen nineties and we are uh and we
00:04:33
have to have predicted new things that will come in two thousand ten so we trained our system
00:04:39
based on on the on the previous huh
00:04:44
on the previous years and that we evaluated it on the two thousand turns
00:04:48
and this system created for us in that most recent that that's that's we've done
00:04:53
it's great for us uh ninety percent accuracy and us some ideas that it came up with yet but when you in the
00:04:59
two thousand times maybe hard to imagine it seems so long ago
00:05:02
right well that i don't think it just just oh just it's
00:05:11
okay cool can vary it
00:05:25
oh
00:05:28
yeah if i'm me i actually like i have such a thing how many interesting
00:05:38
i have the really big one
00:05:43
'cause i have i think that's a have enough of different entrances otherwise i have this huge thing
00:05:51
yeah
00:05:55
yes
00:05:58
yes
00:06:06
oh yeah not for okay like a fact with so yeah i was saying um we can't really evaluate our
00:06:15
system for the future right so we wanted to get an idea of what it what it produced was anything sensible
00:06:21
so we took the any time still the nineteen nineties and then evaluated on the
00:06:25
two thousand tends i want it came up with with things like nature medicine injection
00:06:30
phobia uh antigen p. c. r. injection phobia was an interesting one for work right
00:06:35
maybe find other ways to administer a a a medicine uh urban sustainability for example
00:06:42
and the hop server i thought it's nice to also introduce some technical terms here for the
00:06:47
audience and it's it actually is a type of server i i really have to look that up
00:06:52
okay so what do we want to improve on with our tool we looked a little bit of what is there for other people to use
00:06:58
well one thing we want to improve on his team brainstorming i don't if you've been in
00:07:01
a situation where you ask people for new ideas and everybody's like oh i have no idea right
00:07:07
but we also want to improve on other types of tools that are available to
00:07:11
people to improve off with that team pro brainstorming for example social and media analytic services
00:07:16
they give you a very nice overview of what's happening and what has been happening the
00:07:20
past part where transact going one thinking not doing is what we're doing and that is
00:07:25
creating something new right they just showing you the status quo and it's great for you to look at that to get an idea
00:07:31
but we're actually combine in the things i'm seeing the pattern scroll and learn the
00:07:35
model that is able to predict the the next things that'll come uh actually fan use
00:07:41
there's also innovation management platforms these are platforms where people can input ideas they link up with lots of
00:07:47
external data and that people you know have creative
00:07:51
ideas about uh innovation is feeling like an innovation platform
00:07:54
but yeah they're still relying on people right and we want to help these people with lots of data
00:08:02
so what is our longterm vision but is that we want to be building
00:08:06
creative systems we want to use a yeah right yeah for this next big
00:08:10
forms here where all people we say volume as a creative but systems certainly
00:08:14
aren't right we want to see how far we can get with a creative systems
00:08:18
and we see a big market potential because the global competition creativity market
00:08:23
it's set to grow uh about twenty seven percent if you compare
00:08:27
that to check parts twenty eight percent it's very very strange that
00:08:32
there's so many people working the chat but to sell fuel computation
00:08:35
creativity to this is why we think we should be focusing on that
00:08:40
so flat flight about our journey and the nine days that people here so we started from
00:08:45
uh i'm in an article that we had pretty to write a few years
00:08:49
ago which was just a a first trial and that in these nine days
00:08:54
i brought this team together from people from uh all over the world actually
00:08:57
taking hold to it up with a really nice to work with them again
00:09:00
and we built in the user interface we extended
00:09:03
the algorithm had the domain adaptation uh also into introduced
00:09:08
processed lots and lots of data it was awful yeah i don't know what
00:09:11
how many servers in all different places we use but it was still not enough
00:09:16
so show me the data is is really huge so we fix the demo
00:09:20
we build the business model learned a lot from all the different uh workshops
00:09:24
but yeah and we had a lot of fun don't forget about that the colourful pictures over the phone
00:09:29
what we would like to do after i see see if yeah integrate all the data because we
00:09:34
yeah we did really process a lot of data that we could it integrated into the two
00:09:38
and would like to also then link it up to our interface and uh have a better experience
00:09:43
for the user and evaluate it with our business partners if possible and we already have plans with right
00:09:49
okay that was it for me yeah we are right yeah we need some money back to do that oh i
00:09:54
forgot about that i don't know what it would be very chief that i gave thanks a lot for your uh attention
00:10:00
and uh yeah we're gonna look also at them all um yeah
00:10:07
yeah and it's maybe you want to you know that's janice was a building and so it really but it's good to hear
00:10:15
what's the yeah yeah why don't you do that yeah that that you think that's yeah i can jump in a file yeah
00:10:24
because uh
00:10:26
and as you can see your actions so you can see that could change names so you pretty warm
00:10:32
what kind of company your work at all you would use one of yours for
00:10:37
um what was typing glitch as an example a is recorded a little time because
00:10:45
of the data and gets a inspired for for all the data that you were yours
00:10:50
we we had this running on a simple uh this concerning right or um but we can also cause remember
00:10:56
only because or an remote where there should be didn't focus on those speed of search
00:11:02
for at the moment but of course it was something that can be optimised for them
00:11:07
yeah it's it's uh i thought it was somewhere halfway that uses it became it's my home also um
00:11:15
mm ceases on and strike the but if it's reported you how much
00:11:18
data is processed and then you get the most um a program them
00:11:22
which you decide that in the mood and good related problems from
00:11:27
but over those would be related you would just of our system generated
00:11:32
you can get to them so that it's it's a room remote
00:11:35
viewers and would use much together we could actually get frequencies from rome
00:11:40
which was open both from becoming remote we grew up with
00:11:45
which you knew what you that people are already talking or
00:11:49
and we are out of what related yours or not that already existed
00:11:54
you can also do modems will appropriate would around with that and go
00:11:57
from top to bottom um you can also hard work your portal terms
00:12:03
i think it also select mostly coming to terms are harder to you can focus
00:12:07
on specific things what somebody congestion with them for three four injection needles for other things
Conference Program
Final pitch, team MyLinguist
KHOSRAVANI, Abbas, Idiap Research Institute
Aug. 26, 2021 · 2:15 p.m.
Final pitch, team ProtectID
OTROSHI SHAHREZA, Hatef, Idiap Research Institute
Aug. 26, 2021 · 2:44 p.m.
Final pitch, team innovAItor
Lonneke, Idiap Research Institute
Aug. 26, 2021 · 3:16 p.m.
281 views
Final pitch, team AICIdemo
Michael Liebling, Idiap Research Institute
Aug. 26, 2021 · 3:51 p.m.
Recommended talks
LightAI4Comfort
Aug. 24, 2022 · 12:13 p.m.
119 views