The Deep End of Deep Learning | Hugo Larochelle | TEDxBoston

Artificial Neural Networks are inspired by some of the "computations" that occur in human brains-real neural networks. In the past 10 years, much progress has been made with Artificial Neural Networks and Deep Learning due to accelerated computer power (GPUs), Open Source coding libraries that are being leveraged, and in-the-moment debates and corroborations via social media. Hugo Larochelle shares his observations of what’s been made possible with the underpinnings of Deep Learning.
Hugo Larochelle is a Research Scientist at Twitter and an Assistant Professor at the Université de Sherbrooke (UdeS). Before 2011, he spent two years in the machine learning group at the University of Toronto, as a postdoctoral fellow under the supervision of Geoffrey Hinton. He obtained his Ph.D. at Université de Montréal, under the supervision of Yoshua Bengio. He is the recipient of two Google Faculty Awards. His professional involvement includes associate editor for the IEEE Transactions on Pattern Analysis and Machine Intelligence (TPAMI), member of the editorial board of the Journal of Artificial Intelligence Research (JAIR) and program chair for the International Conference on Learning Representations (ICLR) of 2015, 2016 and 2017.
This talk was given at a TEDx event using the TED conference format but independently organized by a local community. Learn more at ted.com/tedx

[@rickdynes]

NOTE TO INTERNET A talk that is Not for Everyone OR A talk that is (perhaps) Not for You, Can still be Valuable to ...others

[@firefoxmetzger9063]

Its really interesting how fast the field is developing. One year ago he had to introduce deep learning and neural networks and nobody knew. Today virtually everybody seems to know about them...

[@dmitrykalashnikov8637]

Misleading title. I know the history of Deep Learning. I'd never spend time on that. I was expecting some insights on the "end of deep learning"

[@br0cket]

"Deep" end.

[@presa609]

Did you mean something like: ...what about using Deep Learning to create justice where it has been historically denied? Will Deep Learning be subject to protecting vested interests? What if google is culpable of denying justice by removing or allowing information to be denied to persons injured by doctors?

[@tonyreno3168]

Thanks for this comment. Nothing annoys me more than a lying title. Glad I didn't waste my time. BTW, as a rule of thumb I always downvote a misleading title if I see it. Such lying must be punished. People's time is valuable. A title is supposed to let them know what their payoff for spending it on this article/speech/video is. Lies like this must be opposed.

[@RoxanaNoe]

Same here

[@aarmstrong2968]

The neural style transfer algorithm can potentially be used to create a custom visual experience for each user.

[@Brainbuster]

Play at 1.25x playback speed. ;)

[@Leon-pn6rb]

Start watching from 17:58

[@tebisxrod]

it's a shame to use this click bait to get viewers.. "deep end of deep learning". lamentable.

[@seanhardy_]

Rod Tebisx this has like 30,000 views on a channel with 6 million subs lol its not really clickbait

[@tebisxrod]

Sean Hardy yes,It os a clickbait! your 30k views is not a excuse, but a consequence of this! reported!

[@jackmitchell_]

Scene from Transcendence?

[@muzammilnxs]

A whirlwind tour on deep learning. Amazing!

[@Leon-pn6rb]

What I heard: " I WAS INTO AI BEFORE ALL YOU SUCKERS WERE AND YOU MUST KNOW THAT "

[@dogmaradominiczy]

But even then he gave a good insight

[@kkkkjjjj4517]

just play at 2x speed and turn on CC

[@taschke1221]

I don't know if anyone noticed but, it was very telling when he looked down at the red circle in the first min-ish.

[@Alaina420ful]

I'm dumb. What does it tell?

[@taschke1221]

He's conscious, aware, and a little intimidated by the significance and meaning of the circle. He's making a conscious effort to stay within the parameters. Maybe even a little distracted.

[@Alaina420ful]

Interesting. Thank you. I didn't pick up on any of that at all.

[@Bellarium]

I see many complaints, but I find the presenter's presenting style very clear, low on jargon, and insightful. Not for experts I guess.

[@josephdoyle2041]

I thought his presentation style was was clear as well.

[@RafaelQuirinoVex]

I was expecting some information on new learning algorithms (besides backpropagation) or perhaps new types of networks but what i got is boredom... What does this title has to do with this presentation ?...

[@johnstephen52]

I want to see a computer program deal with the "common knowledge" & the common ignoring that humans do instead of those well-defined systems. I WANT TO SEE A COMPUTER PROGRAM ANALYZE & INTERPRET STAGE MAGIC ACTS. Also humor, irony, religious fundamentalism, and other fundamentalisms like computerism. I want to see a computer program that is aware of its own limitations, like I am.

[@toni8675]

How is Alpha Go only the SECOND best Go player in the world? Who is better than Alpha Go?

[@douglasdillman4604]

I'm curious it's 2019now. What happens to all that data can it become a simple thought through a human clone? crow

[@br0cket]

Its actually Hinton himself who came up with the name "Deep Learning"

[@michiganbusinessandtalent161]

louis hay

[@georgegalamb7523]

If you're expecting to see some shocking new technologies from any of these type of videos here, then you have to understand that any new inventions and discoveries in any field and branch of science and engineering will not be revealed, simply because of secrecy to protect their own assets and profitabilities. There is a huge competition out there, why would they reveal any of their latest discoveries. What they are willing to reveal is general knowledge that is freely available for the mass public anyway.

[@marcosrodriguez2496]

it's exactly the oposite actually. these guys are at the forefront of changing how academia. if there is a new idea it's on arxiv immediately. see, only simpletons like you and CEOs think about protecting ideas. It's very myopic. These guys understand that we will be all better off (including them) by making this knowledge available to everyone, as quickly as possible.

[@georgegalamb7523]

Well, if they are supported by non-profit Philanthropists, or if it's open-source, then you are right. But if they aren't, (and most of them are not) then they will protect their new inventions/discoveries/research until it is the right time to reveal them. There's a huge competition going on out there in this dog-eat-dog business-minded monetary-world; and most companies still want to be as successful as possible and so make as much money as possible and own as much assets as possible. This is capitalism. There is no free lunch. Everything cost money. Plus, add to these that most humans are selfish by natural design. Sure, I am with you, it would be a beautiful perfect world "Utopia", if all the people (including politicians and world leaders and company owners) came to their senses and decided one day to freely share all of our resources and ideas; but we're still far from that reality. How far are we from it? It depends on the evolution of the human mind and the desire and demand of the masses. Keep in mind that we are only allowed seeing fractions of new ideas and new discoveries. There are many companies currently working on many different hidden projects that we're only will hear about them when they are ready for the mass market. Sure many "academics" realize and understand that our world would be better off if we humans with our diverse creative minds got together in harmony and mastermind/brainstorm the best possible ideas that we're able to come up with for the advancement of humanity and our precious planet. But the big question is: "Are we there yet?" I'm afraid "no", we are not there yet. Yes, you can hear, here and there that there are open-source projects going on in some areas of research and development, but how widespread it really is? It's still in it's infancy. Yes, in the future we, one day, sooner or later, (hopefully) live in that "monetary-free" and open-minded, fully-intelligent, positive-minded world.

[@twilighttucson2526]

To some extent. TensorFlow and such are being placed into the open source field. The infrastructure of AI is not the profitable part. It's the training material that makes them effective. Which is why Google has become one of the leaders of AI. Their training sample space is staggeringly huge. And why they have no qualms releasing the same framework they use personally.

[@typhoon320i]

When did Johnny Depp learn so much about machine learning?

[@woostep]

6 minutes in and if I hear Artificial Neural Networks again I'm going to lose it, George Katanza style

[@sednove8661]

Très bonne présentation de Hugo.

[@robotguy]

Note to anyone doing public speaking: Don't be like this guy and end every sentence by smacking your lips.

[@marcosrodriguez2496]

LOL, this guy makes your yearly salary in less than a week. I am not sure he should take advice from you on anything.

[@subbuktek]

Note to idiots: when really smart people speak on a complicated topic, they might exhibit some weird quirkiness. Deal with it or go watch some "self-improvement" talks.

[@jarrodstowe805]

Neural networks: In video games we call those skill trees.

[@rgaleny]

HUMAN AWARENESS IS PHYSICAL, PSYCHOLOGICAL AND COGNATIVE

[@rgaleny]

EXPECTATIONS ARE REALITY

[@user-jz7bt2yr6k]

Although i expect Deep Learning Futer, it's ok to listen the histroy of it.anyway, explanation about deep learning algorithm is good.

[@HeyHeyder]

Eventually, If it is not good mathematics, it will be forgotten. I

[@PCarew]

More a history of the politics of ANN

[@jochenm4952]

What is he telling about arXiv.org? In his talk it sounds as if these guys had invented the idea of public preprint servers in 2013. That's ridiculous. arXiv.org has been in place for 20 years or even longer, mostly used by physicist, mathematicians, etc.

[@rgaleny]

CREATE A MAX HEADROOM INTERACTIVE SITE WHERE MILLIONS CAN INTERACT WITH IT TEACHING IT TO RELATE TO PEOPLE ON THE JOB

[@BBBrasil]

very fast, forward into the abyss

[@rfrost4708]

I found the random clicking sounds he makes with his tongue really really annoying. Looking at him somehow gives me a chill in my bones. Very eerie.

[@nykowow]

I guess I'm not a deep learner, since I watched the whole video

[@mbattlecoder]

Everyone and their brother is researching this...

[@CanMetan]

Too much self promotion and boasting of accomplishments. Get to the point.

[@callmeoutlier]

wonderful talk, the idea of running artificial neural networks on GPU it's new to me

[@MichaelDeeringMHC]

GPU's are better than cpu's for running deep learning, but they are not optimised for them. GPU's are optimised for their function as graphics cards. The next step is for some company to design and manufacture hardware optimised for deep learning, DPU's.

[@justinnlk]

Google has TPU (Tensor processing Units)

[@Killuminati23]

The end of Deep A.I will be deep fried humans as a snack for them.

[@adrienpy2386]

Vive le français !!!

[@rakeshmallick27]

This TED talk was given before Alphago beat the world Go champion also before IBM Watson beat human contestans in Jeopardy.

[@maxilosgr]

classic marketing marketing and marketing... deep learning my ass lol

[@bradhaaf4749]

idk why but make a neural network based on a squid's doughnut shaped brain. If you do it tell me what it's good at, so i know why my brain/intuition just came up with that lol hmm mabe a quantum doughnut ?

[@bradhaaf4749]

hmm just occurred to me if you have a circular neural network 200 levels shaped in a circle it could simultaneously compute 100 problems by having an available buffer level between each problem you're working on...

[@MrAndrew535]

With respect to AI, the term deep learning is meaningless. Perhaps this is among the reasons why references to neural networks have a tendency to be immediately dismissed. If this is the case, I would be inclined to dismiss such references too. If for no other reasons than (1) I would be tired of the same references every two minutes and (2) the development of any kind of learning does not equate to intelligence or consciousness and therefore makes no contribution to the subject of AI, Particularly to the I component.

[@marcosrodriguez2496]

I'm not sure what world or time you live in, but references to neural networks are certainly not immediately dismissed. It's currently the leading technology in AI. We solved perception using it and closing in on solving NLP. You sound like you have not the slightest clue what is currently happening.

[@twilighttucson2526]

And pointing it out is equally as meaningless. It's merely a label to cover the field of NN's, RNN's, Markov chains, and other methodologies.

[@manit77]

It takes about 1300 CPUs to run Alpha Go. End of deep learning is just about reached with the current computing and programming platforms. True AI won’t be achieved until we can create a true working quantum computer.

[@redwatch.]

blinded by forehead

[@lonewolfmtnz]

bye bye bipeds

[@shonzhankin5411]

He stole 20 min of my life on this school information. THIEFT!!!

[@yanntiersen2217]

this guy still needs to improve his english. it's a pain to listen

[@jonathangilliam875]

boring

[@jacquesm1652]

What is this bullshit self-promotion?

[@endoflevelboss]

Since he became your boyfriend and you elected to turn a blind eye to his flagrant self-promotion because you're supergay for his back propagating multi-layer networks.

[@ggauche3465]

It's called a TEDx talk. They all seem to about the speaker, using the topic as a background to their own personal story.

[@Leon-pn6rb]

This guy certainly lacks humility

[@lobo25usntrx51]

Mediocre at best, a PhD should have a much better presentation.

[@ac11dc110]

Useless speech