Recent Developments in Deep Learning

Google Tech Talk
March 19, 2010

ABSTRACT

Presented by Geoff Hinton, University of Toronto.

Deep networks can be learned efficiently from unlabeled data. The layers of representation are learned one at a time using a simple learning module that has only one layer of latent variables. The values of the latent variables of one module form the data for training the next module. Although deep networks have been quite successful for tasks such as object recognition, information retrieval, and modeling motion capture data, the simple learning modules do not have multiplicative interactions which are very useful for some types of data.

The talk will show how to introduce multiplicative interactions into the basic learning module in a way that preserves the simple rules for learning and perceptual inference. The new module has a structure that is very similar to the simple cell/complex cell hierarchy that is found in visual cortex. The multiplicative interactions are useful for modeling images, image transformations, and different styles of human walking.

Speaker bio: http://www.cs.toronto.edu/~hinton/bio...

• Category

  Science & Technology

[coastwalker]

codifying learning is a tricky question it turns out, keep up the good work!

[Mark7BD]

oh gosh this brings back memories of university... though I was interested in the material.. I wish I spent more time applying this instead of sitting in a lecture.

[Eduardo thoreum]

ignorance is bliss///

[clockmaker84]

You give autoregressive models an primary eigenval > 1 and they explode, give them one < 1 and they die, so you keep on injecting noise to keep them alive. It's like jiggling a dead horse to get it to move about. ROFL. And SO TRUE!!

[Xusheng Yan]

No, it is an iOS(iPad or iPhone) based app so far, and you can find it with name of iLA(cs) on App Store. thank you！

[Jeremy McMillan]

I don't see it on the App Store. Is it a Mac App?

[Xusheng Yan]

A newly created app for deep learning, iLA(cs), is available on App Store(free).

Excerpt, organise and share, thoughts, ideas and knowledge topics in a natural way, -simple and fast, without the interruption of the flow of ideas by having to think how to organise it.

[wolfgangouille]

Ok I'm done ... I SAID : I'M FINISHED !

*CLAP CLAP CLAP*

[Human20X6]

See:

Hinton, G. E., Osindero, S., & Teh, Y. (2006). A fast learning algorithm for deep belief nets. Neural Computation, 18, 1527–1554.

[wolfgangouille]

It's never too late. Science is accessible now. Do science at home, you have free peer reviewed journals now, you can get and publish your stuff as long as it's thorough.

[CanKaraoke]

Thanks

[Nathan Nifong]

I'm looking for a reference of the claim that is made a 5:23. Can you refer me to a paper?