Brains, Sex, and Machine Learning

Hinton's team (SuperVision) uses an interesting 'dropout' technique. He gave a Google Tech Talk on this back in June. [VIDEO] And an older talk that covers some of what a deep convolutional net is: [LINK]

Here is a GREAT talk by Geoffrey Hinton (the Prof running said lab) [VIDEO] where he explains the method. Unfortunately, even though it was posted three times to HN [LINK] it never made the fron page. Here is the my summary and comment: " Great talk. I don't know much about artificial neural networks (ANN) and even less about natural ones, but I have the feeling that I learnt a lot from this video. If I understand correct, Hinton uses so many artificial neurons compared to the amount of learning data, that you would usually see an overfitting effect. However, his ANN's randomly shut of a substantial part (~50%) of the neurons during each learning iteration. He calls this "dropout". Therefore, a single ANN represents many different models. Most models never get trained, but they exist in the ANN, because they share their weights with the trained models. This learning method avoids over specializing and therefore improves robustness with respect to new data but it also allows for arbitrary combination of different models which tremendously enlarges the pool of testable models. When using or testing these ANNs you also "dropout" neurons during every prediction. Practically, every rerun predicts a different result by using a different model. Afterwards, these results are averaged. The more results, the higher the chance, that the classification is correct. Hinton argues, that our brains work in a similar way. This explains among other things a) Why are neurons firing in a random

In addition to what bravura said, Geoff Hinton's group at UT has recently introduced a new approach to training neural networks that they call "dropout". You can read about it in their paper ( [LINK] ) or watch Hinton describe it in a recent talk he gave at Google ( [VIDEO] ). Roughly speaking, dropout training provides a strong regularizing effect through a sort of model averaging that is conceptually related to the well-known bagging approach from which random forests derive their power and flexibility. Dropout training has already produced state-of-the art results on several time-worn standard benchmarks and helped Hinton's group win a recent Kaggle competition (for an overview of their approach, see: [LINK] ). I've played around with this a bit over the last few weeks, and have a Matlab implementation publicly available from my Github at: [LINK] .

Something really interesting must be happening with AI at Google, because in the past few months both Ray Kurzweil (the best-known proponent of the singularity) and Geoff Hinton (the crazy-talented individual who invented deep-belief networks using interconnected "restricted Boltzman machines"[1]) have joined the company. -- [1] For an overview of deep belief networks, see these videos: [LINK] , [LINK] , and [VIDEO]

It seems the relationship between the brain and deep learning has evolved in such a way that the later can help with insights into how the former works. In this regard, I thought I would mention the extraordinary simple and elegant talk by G. Hinton last summer: [VIDEO] It starts from a simple and clever improvement to an existing deep learning method and ends up with beautiful (and simple!) insights on why neurons are using simple spikes to communicate.

Using dropout with deep neural networks is an (extremely cheap) way to gain many of the benefits of using an ensemble. If you want a great overview of the dropout technique, watch this tech talk by Hinton: [VIDEO] Also, the recent maxout algorithm (from Montreal group, authors of Theano) that got state of the art results on several datasets is essentially just an algorithm designed to do particularly well with dropout (as far as I understand it).

Speaking of bagging: Geoff Hinton talks a bit about model averaging and has a very stimulating pair of hypotheses related to this, which are: a) genetic recombination in the form of sex might be "bagging across genes" that prevents tight co-adaptation (= overfitting on the evolutionary timescale) b) the brain uses noisy discrete firing instead of continuous communication because it allows for "bagging across network topologies". The bulk of his talk is about how dropout in neural nets can be used to accomplish some of the same performance benefits that model averaging gives on other algorithms (at a much smaller relative computational cost). Here's the talk: [VIDEO]

As far as I know there is a decent amount of information against the fact that brains are quantum computers. On the other hand there is a talk by Hinton on youtube [1] that sheds interesting light on some variants of deep learning and the way the brain uses (classical) noise. [1] [VIDEO]

>As for contrastive divergence with Restricted Boltzmann Machines overcoming the vanishing gradient problem. That's true, but has anyone even tried doing this with 7-bit floating point and demonstrated it even works? I wasn't referring to RBMs here, although it does look like I misremembered the details. See slides 74-76 of this presentation for a brief sketch [0]. In these feed-forward MLPs, the information content from each neuron is capped at one bit (which is quite a bit less than seven). I seem to recall him saying something similar (and with more detail) in a Google Tech Talk a few months later [1], although I'm not sure. [0]: [LINK] [1]: [VIDEO]

Geoff Hinton has a theory that Naive Bayes is just logistic regression with dropout: [VIDEO] And that if you divide all the parameters by a certain amount, you would get accurate probability estimates.

I would highly recommend this video [VIDEO] from Geoffrey Hinton in 2012 that I think is related to this research.

Geoffrey Hinton described doing this back in 2012. He noted that it learned more slowly but generalized better, he described how this is nearly isomorphic with dropout, and he presented a biological motivation: [VIDEO] [VIDEO]