Hacker News Board Hacker News, browsed like a classic message board
Live data from Hacker News

Does Deep Learning Have Deep Flaws?

kdnuggets.com

Post reply on HN
Print view View original

21–30 of 55 posts

Re: Does Deep Learning Have Deep Flaws?

#21

Post by agibsonccc »

JabavuAdams wrote: post #10
agibsonccc wrote: post #5

I think just like any machine learning algorithm, especially with computer vision, you need to prepare things properly. More robust data and images such as moving window on image patches and image rotations, even going as far as applying filters like sobel for edge detection will make your algorithm a lot better. Any algorithms have weaknesses. It's a matter of being aware of them and compensating for them in your mo…

The value proposition of deep learning is to eliminate these kind of hand-coded features and to discover the features automagically. However, maybe there's a middle ground. I.e. maybe we don't need the more esoteric features that SIFT uses, but it just makes sense to do edge detection, and say a Fourier transform for audio.

I wrote [1], I'm plenty aware of the "feature discovery" that goes on, but it can still be an enhancer. See my recent talk[2] for a good overall idea of the situation.

Normalization and other data transforms are still required for discovery of features.

[1]: http://deeplearning4j.org/

[2]: https://www.youtube.com/watch?v=hykoKDl1AtE

Re: Does Deep Learning Have Deep Flaws?

#22

Post by streptomycin »

cpa wrote: post #6
jrochkind1 wrote: post #3

First thought: Can I turn all digital pictures of me into 'adversarial examples', so the eye of sauron can't identify me from pictures? I'm sure it's not as simple as that, presumably any algorithmic modification to an 'adversarial' nature can be countered by other algorithms. But I predict a new realm of 'arms race' here in the future.

Adversarial examples are tied to a specific algorithm, they didn't produce any universal adversarial examples from what I understand.

From the article: What’s more surprising is that the same perturbation can cause a different network, which was trained on a different training dataset, to misclassify the same image. It means that adversarial examples are somewhat universal.

Re: Does Deep Learning Have Deep Flaws?

#23

Post by magicalist »

varelse wrote: post #14

This problem was observed 20+ years ago with linear models used for protein structure prediction. For any given model of what described a properly folded protein, one could locate conformations of the same protein that were rated as folded even better than the correct conformation (I called them doppelgangers, but the name "decoy" is what caught on). The statistical naivete of the field led to all sorts of inadverten…

> If it's likely overfitting, attempt an expectation maximization-like fix to the problem. Train the network. Generate adversaries, Add them to the training set, train again and then lather rinse repeat until either the network can't be trained or the problem goes away.

As I quoted in my other comment, the paper suggests doing exactly that.

Re: Does Deep Learning Have Deep Flaws?

#24

Post by dave_sullivan »

magicalist wrote: post #20

The concept is very cool, but it's not surprising that dimensionality reduction through a non-linear process is going to result in sections of input parameters that yield incorrect (and weird) results. Our visual system, while not the same as these systems, is extremely well developed and robust, yet the list of optical illusions that can fool us is quite long. In this study, the optical illusions are really just sur…

> This summary completely neglects the fact that the paper also recommends that -- just like distorted images are added to training sets today (you wouldn't want something common like optical aberration from the camera lens screwing up your classifier) -- in the future, these adversarial examples should be added to training sets to mitigate their effects.

I've seen several articles citing this paper as proof that deep learning is deeply flawed, yet they all seem to miss the point you make above.

The other interesting result is that the neurons are not in fact individual features you can just grab and drop into another algo--the entire space defined by the model works together, through all the layers. Honestly, that was a more interesting result for me, although I don't know that it negates anything, I've just got to stop telling people that the individual units are features.

Not that deep learning is the end-all-be-all of machine learning--it's not. It's just that this paper isn't saying what reporters are saying it's saying... As per usual?

Never let the truth get in the way of a good story...

Re: Does Deep Learning Have Deep Flaws?

#25

Post by kilbasa »

codeflo wrote: post #13
kilbasa wrote: post #7

IIRC even the human brain has the 'adversarial' image flaw (these images will be unique to each person), but one simple workaround is to alter the input image via eye movement (which happens unconsciously).

IIRC? Can you provide any source or example? This sounds very interesting.

There was a discussion on slashdot (take it with a grain of salt perhaps) about a similar article.

http://slashdot.org/story/14/05/27/1326219/the-flaw-lurking-...

The comment I recalled was written by someone with the handle "presidenteloco".

Re: Does Deep Learning Have Deep Flaws?

#26

Post by dekhn »

varelse wrote: post #14

This problem was observed 20+ years ago with linear models used for protein structure prediction. For any given model of what described a properly folded protein, one could locate conformations of the same protein that were rated as folded even better than the correct conformation (I called them doppelgangers, but the name "decoy" is what caught on). The statistical naivete of the field led to all sorts of inadverten…

Thanks for saying this. I can't really comment on your experiments (I;m not qualified) but you can be assured that some people are working in machine learning today specifically having learnt the lessons of pre- and post-CASP. I don't know that I agree CASP was founded specifically because people found decoys, but...

it was an special shock when I learned about ensemble methods (I think they were just called "combined servers" at the time) at CASP and saw that all our hard work (manual alignments, lots of expert analysis of models, etc) wasn't really better (far worse in fact) than a few simply trained ensemble systems that memorized what they were bad at and classified their predictions with the appropriate probabilities.

See also: http://www.nature.com/nchem/journal/v6/n1/nchem.1821/metrics... http://googleresearch.blogspot.com/2012/12/millions-of-core-... (note, 4 of the 6 projects awarded specifically involved physical modelling of proteins and the fifth was a drug-protein binding job) http://research.google.com/archive/large_deep_networks_nips2...

none of the above are coincidental: the first two links are specifically because I went to Google to use those GPUs and CPUs for protein folding and design and drug discovery. The third project is now something I am experimenting with.

Re: Does Deep Learning Have Deep Flaws?

#30

Post by thisisdave »

magicalist wrote: post #20

The concept is very cool, but it's not surprising that dimensionality reduction through a non-linear process is going to result in sections of input parameters that yield incorrect (and weird) results. Our visual system, while not the same as these systems, is extremely well developed and robust, yet the list of optical illusions that can fool us is quite long. In this study, the optical illusions are really just sur…

An overlapping group of authors (compared with the ones that wrote the "intriguing properties" paper) recently published a neat preprint on using adversarial networks as part of the training process. The goals of the paper are different from the networks discussed here, but I thought it was really interesting.

http://arxiv.org/abs/1406.2661

Post reply on HN