9 ms·

What are the limits of deep learning?

102 points by pizza 8y ago

hprotagonist 8y ago
"NNs that rely on backprop only really work on continuous functions and many interesting things like language[0] are not products of those" is real high on my list of limits. [0]: https://www.linkedin.com/pulse/google-hyping-why-deep-learning-cannot-applied-easily-berkan-ph-d https://www.linkedin.com/pulse/google-hyping-why-deep-learni...
- evrydayhustling 8y ago
  State of the art on most NLP tasks is now provided by deep learning, thanks to a host of techniques for dealing with both categorical and structurally complex data, such as LSTMs. Here's a great resource for keeping up with development in that area: https://github.com/sebastianruder/NLP-progress https://github.com/sebastianruder/NLP-progress
- yorwba 8y ago
  That argument seems to be based on the idea that because written language uses a discrete encoding, it's impossible to use continuous functions. But as soon as you replace deterministic discrete values by continuous probabilities of those values, you're back in the space of continuous functions. Even non-NN approaches to NLP use that representation, because optimizing continuous parameters is something we can do very well. Discrete optimization is much harder.
  - YeGoblynQueenne 8y ago
    And how well do all those techniques work when it comes to representing meaning, rather than structure?
    - yorwba 8y ago
      I think meaning isn't really separable from structure. But assuming you're talking about semantics vs. syntax, then continuous representations can certainly do a lot more than just syntax. They can model how words change in meaning over time [1], they can identify words that mean the same in different languages [2] and use that to translate whole sentences [3] without ever seeing a direct translation. Of course none of those tasks are solved to perfection yet, but some kind of meaning is definitely being captured. [1] https://arxiv.org/abs/1703.00607 https://arxiv.org/abs/1703.00607 [2] https://arxiv.org/abs/1710.04087 https://arxiv.org/abs/1710.04087 [3] https://arxiv.org/abs/1710.11041 https://arxiv.org/abs/1710.11041
      - YeGoblynQueenne 8y ago
        I disagree. If I tell you that argoubr + vliugo - viyt = plqtv, do you now understand what any of those words mean? If I say that "gye" must always precede "turo", do you know what "gye turo" means, or why it doesn't make sense to say "turo" without "gye"? What kind of meaning is captured by the relations between these words that these rules I just described, define? Would it help if I gave you all the possible ways that every word in the language those words are drawn from are to be articulated together, without telling you what any of those wors mean? And how would I explain those words' meaning without recourse to words whose meaning you already know? Also, that's what I mean by "structure" - the context in which tokens are found in text. I don't want to say "syntax" or "semantics" because it seems to me those are loaded terms that have very specific, er, meaning and I am not a linguist to be able to unpick it.
        yorwba 8y ago
        > Would it help if I gave you all the possible ways that every word in the language those words are drawn from are to be articulated together, without telling you what any of those wors mean? And how would I explain those words' meaning without recourse to words whose meaning you already know? That's why I linked the two papers on unsupervised machine translation. Given enough text in an unknown language, it is already possible to turn those unknown words into words which humans understand.
evrydayhustling 8y ago
Yet another piece that cherrypicks AI failures in order to make vague thesis that "deep learning is not enough". I can't decide if it makes the article more or less honest that the "Beyond Deep Learning" section is essentially a catalog of... recent developments in Deep Learning. The need to (selectively) pidgeonhole and then judge (on impossible to operationalize criteria) a fast-evolving field is purely rhetorical and, worse, aimed at folks who don't understand the way the field is progressing. I've got nothing against raising platforms for alternative lines of AI research, especially including highlighting the ways they complement current deep learning But this article format is so transparent and tired.
- eli_gottlieb 8y ago
  If you actually read the papers they cite, you'll notice they bring in probabilistic programming as well as other alternatives to deep learning.
  - evrydayhustling 8y ago
    I didn't say they didn't! Yes, they "bring in" probabilistic programming, graphical representations of symbolic structures, and more. But all of the citations 6-10 are from papers that integrate new approaches into a deep neural model. Most are from self- described deep learning teams, and all would be welcome at conferences and publications within the deep learning community. Integrating with other inference paradigms has been a major driver in neural network research for 30y... so trying to cast progress like this as some sort of comeuppance that proves the insufficiency of the field is BS.
    - eli_gottlieb 8y ago
      >But all of the citations 6-10 are from papers that integrate new approaches into a deep neural model. Having read citation 8, "Building Machines that Learn and Think like People", a number of times over, I don't see where that paper includes a neural network architecture.
endymi0n 8y ago
I think the current critique of Deep Learning as a technique is a dead end. Now I'm only a semi-professional in a sea of recent development and research, but if you ask me, the technique itself can take us all the way to General Intelligence and more. What's missing from this discussion is how we humans don't just use a single, vision based "deep net" in order to identify the banana, we use lots of very different deep nets that run in parallel and cross-validate each other. We have a vision net like computers that can be fooled with the same tricks (just google "optical illusion"). But a second net does three-dimensional inference of the banana shape from stereo vision and cross-validates that with a toaster. If in doubt, unnoticed parallax movement will be commanded to our body in order to get a better shape estimation. Another net parses the content. A banana in a jungle or on a table would be highly plausible. A banana in an operating room or on a roller coaster would be implausible. Another net parses context and timeline. If no banana was there and no banana has been incoming through our continuous estimated 3d space, there is probably no banana there. There's smell, taste, all kinds of things running in parallel that continuously cross-validate our perception of this being a banana, probably a lot of different models just from vision alone that feed back into the main classification thread. We'll even happily do looped cross validation if our eyes rest on a scene. Our memory will fetch reference data for all first order predictions and start subconciously processing this into the main prediction. It will predict how a banana feels, how soft it is, how warm. If we see a banana on lava that is not burning, our logic and reasoning will challenge the prediction. All of this happens subconciously, and there's no magic behind. If you ask me, there are no limits of deep learning. There's just still a limit of imagination in feature extraction and processing.
- dmreedy 8y ago
  If there's one lesson to take away from our history of studying the brain, the phenomenon of human intelligence, and the halting 'progress' of AI, it's that it always seems as easy as you propose. At first.
- tabtab 8y ago
  It could be like Turing Equivalency in that deep nets of deep nets may be able to do the job, but are not the most efficient and/or easiest AI for humans to tune and tame. I'm not saying gluing bunches of deep nets together is bad R&D, only that it may not be the only or best AI game in town. I suppose if we got a wad of deep nets to give General Intelligence (including "common sense"), but it took a big room of servers, that may be a good enough springboard, but I'm not sure putting all our eggs in the deep net basket is the best route to fuller AI just because it's been the most promising up to this point. It's roughly comparable to spending all R&D perfecting vacuum tubes in the 1940's because they were the best known route back then to general digital electronics. The best starters are not always the best finishers.
alexandernst 8y ago
I'm not an IA expert in any way; I just merely started learning some basic stuff, but so far the only thing I see from all the algorithms I have covered so far is that IA, deep learning, ML and so on are just statistical databases or sets, so to call them. They don't learn anything, they just repeat. From that I can say that IA is limited only to the data you can label and then feed it.
cwkoss 8y ago
The limit of deep learning is the quantity and accuracy of training data. For some problems, you can't run 10^x trials: you can synthesize training data but it may be of questionable real-world relevance and lead to overfitting of the generation algorithm.
- randcraw 8y ago
  The other big limit to DL is its inability to learn a new concept from a single training example, AKA one-shot learning. Given that humans do most of their supervised learning this way underscores that, even though DL nets intriguingly resemble the architecture of the brain, in fact, DL's current learning process clearly departs from the one used by the brain. (Unless the way a baby fiddles with each new object is in fact a form of data augmentation, increasing the visual sample space by changing the object's angles. But I digress...) Better to drop the "Neural Net as Brain" metaphor before we Procrustes it to death.
  - randyrand 8y ago
    is that a limit of DL? or is that a limit of how we currently do our training?
  - hnick 8y ago
    It also has no imagination, unless I've missed something (I admit I really need to catch up in this area). Humans are experts at imagining 'what if' scenarios and learning from things that never even happened. "If this was bright red, is it still a cat? Or do all cats have a limited range of colours on a fixed cat-colour-spectrum?" I'm not sure if any DL systems do that?
    - randcraw 8y ago
      Proposing what-if scenarios and asking questions of your own and using your imagination generally falls under the name 'counterfactuals' -- a big area of AI research. But to date, the priorities of deep learning have focused elsewhere (mostly in perception of images and speech via discrete classification). Likewise, 99% of the computing strategy used by DL practitioners and researchers continues to use large scale supervised learning -- because it continues to make advances in many problem spaces just as it is. Is it possible to escape DL's apparent limits and use it differently to solve problems that lie in the counterfactual domain? Perhaps. But so far, there's been relatively little work or modest results along those lines. Given the architectural model used by DL (large matrix multiplications and iterations using SIMD operations on GPUs), the prospect of applying it to alternative AI objectives that are more conventionally symbolic and whose methods are typically algebraic (and unnatural to share the same computational model as DL) do not seem to be especially promising places for DL to thrive. Not so far anyway.
YeGoblynQueenne 8y ago
>> And it could make the networks far less vulnerable to adversarial attacks simply because a system that represents things as objects, as opposed to patterns of pixels, isn’t going to be so easily thrown off by a little noise or an extraneous sticker. But, representing objects and their relations is how Good, Old-Fashioned AI works and the inability to deal with noisy input was, exactly, where GOFAI systems stumbled. How are relational representations going to suddendly help neural nets (famously robust to noise already) better handle noise? In any case, if we really want to learn relations there are symbolic learning techniques that can do that very cheaply from very few examples. Are we reinventing the wheel, only with more artificial neurons, now? And what's the point of that?
- AstralStorm 8y ago
  The problem is that neither AI can differentiate between something new and noise. Even probabilistic models have this problem to a lesser degree. So the idea is indeed to have an ANN learn a kind of a semantic network - then the problem is still pruning it and clustering reliably as well as detecting the new.
darawk 8y ago
> Then there’s the opacity problem. Once a deep-learning system has been trained, it’s not always clear how it’s making its decisions. This is no less true of human decision making.
otabdeveloper1 8y ago
What do you mean 'limits'?? There are no limits to deep learning! Singularity here we come! (That was sarcasm. Sad that a disclaimer is needed nowadays.)

102 points by pizza 8y ago

"NNs that rely on backprop only really work on continuous functions and many interesting things like language[0] are not products of those" is real high on my list of limits. [0]: https://www.linkedin.com/pulse/google-hyping-why-deep-learning-cannot-applied-easily-berkan-ph-d https://www.linkedin.com/pulse/google-hyping-why-deep-learni...

evrydayhustling 8y ago
State of the art on most NLP tasks is now provided by deep learning, thanks to a host of techniques for dealing with both categorical and structurally complex data, such as LSTMs. Here's a great resource for keeping up with development in that area: https://github.com/sebastianruder/NLP-progress https://github.com/sebastianruder/NLP-progress
yorwba 8y ago
That argument seems to be based on the idea that because written language uses a discrete encoding, it's impossible to use continuous functions. But as soon as you replace deterministic discrete values by continuous probabilities of those values, you're back in the space of continuous functions. Even non-NN approaches to NLP use that representation, because optimizing continuous parameters is something we can do very well. Discrete optimization is much harder.
- YeGoblynQueenne 8y ago
  And how well do all those techniques work when it comes to representing meaning, rather than structure?
  - yorwba 8y ago
    I think meaning isn't really separable from structure. But assuming you're talking about semantics vs. syntax, then continuous representations can certainly do a lot more than just syntax. They can model how words change in meaning over time [1], they can identify words that mean the same in different languages [2] and use that to translate whole sentences [3] without ever seeing a direct translation. Of course none of those tasks are solved to perfection yet, but some kind of meaning is definitely being captured. [1] https://arxiv.org/abs/1703.00607 https://arxiv.org/abs/1703.00607 [2] https://arxiv.org/abs/1710.04087 https://arxiv.org/abs/1710.04087 [3] https://arxiv.org/abs/1710.11041 https://arxiv.org/abs/1710.11041
    - YeGoblynQueenne 8y ago
      I disagree. If I tell you that argoubr + vliugo - viyt = plqtv, do you now understand what any of those words mean? If I say that "gye" must always precede "turo", do you know what "gye turo" means, or why it doesn't make sense to say "turo" without "gye"? What kind of meaning is captured by the relations between these words that these rules I just described, define? Would it help if I gave you all the possible ways that every word in the language those words are drawn from are to be articulated together, without telling you what any of those wors mean? And how would I explain those words' meaning without recourse to words whose meaning you already know? Also, that's what I mean by "structure" - the context in which tokens are found in text. I don't want to say "syntax" or "semantics" because it seems to me those are loaded terms that have very specific, er, meaning and I am not a linguist to be able to unpick it.
      - yorwba 8y ago
        > Would it help if I gave you all the possible ways that every word in the language those words are drawn from are to be articulated together, without telling you what any of those wors mean? And how would I explain those words' meaning without recourse to words whose meaning you already know? That's why I linked the two papers on unsupervised machine translation. Given enough text in an unknown language, it is already possible to turn those unknown words into words which humans understand.

evrydayhustling 8y ago

Yet another piece that cherrypicks AI failures in order to make vague thesis that "deep learning is not enough". I can't decide if it makes the article more or less honest that the "Beyond Deep Learning" section is essentially a catalog of... recent developments in Deep Learning. The need to (selectively) pidgeonhole and then judge (on impossible to operationalize criteria) a fast-evolving field is purely rhetorical and, worse, aimed at folks who don't understand the way the field is progressing. I've got nothing against raising platforms for alternative lines of AI research, especially including highlighting the ways they complement current deep learning But this article format is so transparent and tired.

eli_gottlieb 8y ago
If you actually read the papers they cite, you'll notice they bring in probabilistic programming as well as other alternatives to deep learning.
- evrydayhustling 8y ago
  I didn't say they didn't! Yes, they "bring in" probabilistic programming, graphical representations of symbolic structures, and more. But all of the citations 6-10 are from papers that integrate new approaches into a deep neural model. Most are from self- described deep learning teams, and all would be welcome at conferences and publications within the deep learning community. Integrating with other inference paradigms has been a major driver in neural network research for 30y... so trying to cast progress like this as some sort of comeuppance that proves the insufficiency of the field is BS.
  - eli_gottlieb 8y ago
    >But all of the citations 6-10 are from papers that integrate new approaches into a deep neural model. Having read citation 8, "Building Machines that Learn and Think like People", a number of times over, I don't see where that paper includes a neural network architecture.

endymi0n 8y ago

I think the current critique of Deep Learning as a technique is a dead end. Now I'm only a semi-professional in a sea of recent development and research, but if you ask me, the technique itself can take us all the way to General Intelligence and more. What's missing from this discussion is how we humans don't just use a single, vision based "deep net" in order to identify the banana, we use lots of very different deep nets that run in parallel and cross-validate each other. We have a vision net like computers that can be fooled with the same tricks (just google "optical illusion"). But a second net does three-dimensional inference of the banana shape from stereo vision and cross-validates that with a toaster. If in doubt, unnoticed parallax movement will be commanded to our body in order to get a better shape estimation. Another net parses the content. A banana in a jungle or on a table would be highly plausible. A banana in an operating room or on a roller coaster would be implausible. Another net parses context and timeline. If no banana was there and no banana has been incoming through our continuous estimated 3d space, there is probably no banana there. There's smell, taste, all kinds of things running in parallel that continuously cross-validate our perception of this being a banana, probably a lot of different models just from vision alone that feed back into the main classification thread. We'll even happily do looped cross validation if our eyes rest on a scene. Our memory will fetch reference data for all first order predictions and start subconciously processing this into the main prediction. It will predict how a banana feels, how soft it is, how warm. If we see a banana on lava that is not burning, our logic and reasoning will challenge the prediction. All of this happens subconciously, and there's no magic behind. If you ask me, there are no limits of deep learning. There's just still a limit of imagination in feature extraction and processing.

dmreedy 8y ago
If there's one lesson to take away from our history of studying the brain, the phenomenon of human intelligence, and the halting 'progress' of AI, it's that it always seems as easy as you propose. At first.
tabtab 8y ago
It could be like Turing Equivalency in that deep nets of deep nets may be able to do the job, but are not the most efficient and/or easiest AI for humans to tune and tame. I'm not saying gluing bunches of deep nets together is bad R&D, only that it may not be the only or best AI game in town. I suppose if we got a wad of deep nets to give General Intelligence (including "common sense"), but it took a big room of servers, that may be a good enough springboard, but I'm not sure putting all our eggs in the deep net basket is the best route to fuller AI just because it's been the most promising up to this point. It's roughly comparable to spending all R&D perfecting vacuum tubes in the 1940's because they were the best known route back then to general digital electronics. The best starters are not always the best finishers.

alexandernst 8y ago

I'm not an IA expert in any way; I just merely started learning some basic stuff, but so far the only thing I see from all the algorithms I have covered so far is that IA, deep learning, ML and so on are just statistical databases or sets, so to call them. They don't learn anything, they just repeat. From that I can say that IA is limited only to the data you can label and then feed it.

cwkoss 8y ago

The limit of deep learning is the quantity and accuracy of training data. For some problems, you can't run 10^x trials: you can synthesize training data but it may be of questionable real-world relevance and lead to overfitting of the generation algorithm.

randcraw 8y ago
The other big limit to DL is its inability to learn a new concept from a single training example, AKA one-shot learning. Given that humans do most of their supervised learning this way underscores that, even though DL nets intriguingly resemble the architecture of the brain, in fact, DL's current learning process clearly departs from the one used by the brain. (Unless the way a baby fiddles with each new object is in fact a form of data augmentation, increasing the visual sample space by changing the object's angles. But I digress...) Better to drop the "Neural Net as Brain" metaphor before we Procrustes it to death.
- randyrand 8y ago
  is that a limit of DL? or is that a limit of how we currently do our training?
- hnick 8y ago
  It also has no imagination, unless I've missed something (I admit I really need to catch up in this area). Humans are experts at imagining 'what if' scenarios and learning from things that never even happened. "If this was bright red, is it still a cat? Or do all cats have a limited range of colours on a fixed cat-colour-spectrum?" I'm not sure if any DL systems do that?
  - randcraw 8y ago
    Proposing what-if scenarios and asking questions of your own and using your imagination generally falls under the name 'counterfactuals' -- a big area of AI research. But to date, the priorities of deep learning have focused elsewhere (mostly in perception of images and speech via discrete classification). Likewise, 99% of the computing strategy used by DL practitioners and researchers continues to use large scale supervised learning -- because it continues to make advances in many problem spaces just as it is. Is it possible to escape DL's apparent limits and use it differently to solve problems that lie in the counterfactual domain? Perhaps. But so far, there's been relatively little work or modest results along those lines. Given the architectural model used by DL (large matrix multiplications and iterations using SIMD operations on GPUs), the prospect of applying it to alternative AI objectives that are more conventionally symbolic and whose methods are typically algebraic (and unnatural to share the same computational model as DL) do not seem to be especially promising places for DL to thrive. Not so far anyway.

YeGoblynQueenne 8y ago

>> And it could make the networks far less vulnerable to adversarial attacks simply because a system that represents things as objects, as opposed to patterns of pixels, isn’t going to be so easily thrown off by a little noise or an extraneous sticker. But, representing objects and their relations is how Good, Old-Fashioned AI works and the inability to deal with noisy input was, exactly, where GOFAI systems stumbled. How are relational representations going to suddendly help neural nets (famously robust to noise already) better handle noise? In any case, if we really want to learn relations there are symbolic learning techniques that can do that very cheaply from very few examples. Are we reinventing the wheel, only with more artificial neurons, now? And what's the point of that?

AstralStorm 8y ago
The problem is that neither AI can differentiate between something new and noise. Even probabilistic models have this problem to a lesser degree. So the idea is indeed to have an ANN learn a kind of a semantic network - then the problem is still pruning it and clustering reliably as well as detecting the new.

darawk 8y ago

> Then there’s the opacity problem. Once a deep-learning system has been trained, it’s not always clear how it’s making its decisions. This is no less true of human decision making.

otabdeveloper1 8y ago

What do you mean 'limits'?? There are no limits to deep learning! Singularity here we come! (That was sarcasm. Sad that a disclaimer is needed nowadays.)