Deep learning can’t be trusted, brain modelling pioneer says
spectrum.ieee.org
spectrum.ieee.org
All that to say, it's easy to find flaws with it, but it has been game changing. I'm naturally skeptical about someone who has just published a book on something hes been working on for 50 years that he believes is "better" than the current (admittedly fad) technology. I'd like to see some real examples about the specific kinds of problems his tech solves, and how it solves them better than other methods. If he's saying it's better than deep learning on things deep learning is good at (e.g. can it beat a neural network model at classifying imagenet?) then I call BS absent a compelling demo. If it's better at some other thing, let's hear what that is.
I think the progress was the result of competitions that kept everyone blindfolded to test datasets. No one could exaggerate the performance of their methods, allowing deep learning to emerge as a winning solution. I think deep learning has fallen short of the hype because researchers are exaggerating their results, and what is needed to propel progress is a renewed focus (with funding) for competitions.
Advances in DL have been mostly empirical not theoretical from what I have seen. It's less about being able to "envisage" and more about empirical testing.
I may be reading too deeply into your question and you don't actually mean that theoretical. For something lighter, I'd suggest studying some of the common model architectures in the area you're interested in (like you can look at the pytorch lightning or torchvision or detectron2 models or etc source code) and understanding why things are the way they are.
[0] https://topml.rice.edu/ (this is an annual conference, can look at last years talks and references therein, search for topml, etc)
You will be very disappointed if you expect that there is some grand theory of Deep Learning.
My suggestion is to follow Arxiv Sanity: http://www.arxiv-sanity.com/top
We have these tools that solve specific problems in specific contexts yet all these companies are on the hype marketing train. Are these investors not doing due diligence or am I missing something?
Look, everyone will jump on the ART bandwagon if it ever actually beats deep learning. Machine Learning is a very empirical field! But all the people throwing rocks at Deep Learning have failed to propose anything that actually works better.
I think the latest IEEE articles have been a little intellectually lightweight, perhaps we're just simply not their audience. Still, in the style of IEEE papers, I would like to see references for the claims made.
> But all the people throwing rocks at Deep Learning have failed to propose anything that actually works better
I was thinking something similar. DL for sure has problems, but I am yet to see something better. ART doesn't just need to be better, it needs to be better in _at least_ one of: accuracy, training time, recall time, amount of training data, quality of training data.
Also nothing is perfect.
And even severely imperfect things can be reasonable good solution/improve a situation.
So don't base a curt ruling on it.
Be very careful about using it when it can put lives at risk (e.g. self driving).
EDIT:And make the people selling the system responsible for it's failure!! You can't expect a person you tell to not touch the wheel for hours to then be able to react asap if something goes wrong, that's not how humans work. Hence putting the responsibility on the human which happens to sit behind the wheel is completely unreasonable.END EDIT
Anyway my main problem with DL is that its hard to evaluate how trust-able it is if applied to a very diverse real world situations (i.e. self driving). You can evaluate it on data sets but data sets are extremely unlikely to get anywhere close to representing a diverse reality, and while each edge case not represented is very rare due to the diversity of reality there is a endless number of them so sooner or later they will be hit. And as they are edge case not represented by your data or evaluation sets you can't relay reason well what will happen if they are it. Furthermore due to the nature of data collection it's not rare for independent data sets to unexpectedly share subtle biases.
A good example is a Tesla car getting confused because in front of it was a truck which had traffic lights loaded on the back. A extremely rare/unusual situation, which happened anyway.
So you have to layer different systems, preferable also some fallback emergency systems not dependent on DL.
Let's see the performance of ART on some benchmark tasks, then we can talk.
Here's Geoff Hinton on benchmarks (again; I cite that all the time):
GH: One big challenge the community faces is that if you want to get a paper published in machine learning now it's got to have a table in it, with all these different data sets across the top, and all these different methods along the side, and your method has to look like the best one. If it doesn’t look like that, it’s hard to get published. I don't think that's encouraging people to think about radically new ideas.
Now if you send in a paper that has a radically new idea, there's no chance in hell it will get accepted, because it's going to get some junior reviewer who doesn't understand it. Or it’s going to get a senior reviewer who's trying to review too many papers and doesn't understand it first time round and assumes it must be nonsense. Anything that makes the brain hurt is not going to get accepted. And I think that's really bad.
What we should be going for, particularly in the basic science conferences, is radically new ideas. Because we know a radically new idea in the long run is going to be much more influential than a tiny improvement. That's I think the main downside of the fact that we've got this inversion now, where you've got a few senior guys and a gazillion young guys.
https://www.wired.com/story/googles-ai-guru-computers-think-...
And here's a recent paper noting yet another way machine learning benchmarks are borked:
https://arxiv.org/abs/2003.08907
P.S. And, please leave that aggressive attitude out of research discussions ("bullshit walks"?). Science is not the constant comparison of dick measurements. That's perhaps the point of sports, but anything that stifles debate and inhibits the development of new ideas is anti-science.
Just last year there was a breakthrough in decades old problem of protein-folding. The year before that we figured out that large language models have astonishing few shot learning abilities. CLIP introduced strong 0-shot capabilities to image recognition. Progress in image generation from descriptions has been nothing but astonishing with Dall-E and GLIDE from OpenAI.
Not to mention breakthroughs in game theoretical problems like Counterfactual Regret Minimization and more recent application of Deep Learning to make CFR practical in actual games. Including continuous progress in the AlphaGo linage to AlphaZero, MuZero, and last year EfficientZero and Player of Games.
To make an analogy it's as if we could sort ever larger lists because we keep sorting them with ever bigger computers... but the only sorting algorithm we know is bubblesort.
You bring up large language models. Language models are about as old as computational linguistics [6]. Large language models are... large. There's nothing that's new about them. "Attention" is an anthropomorphically misnamed engineering tweak of the kind that was rightly lambasted by Drew McDermot fourty years ago [7]. Self-play and reinforcement Learning are now more than 60 years old [8].
As to "few shot" learning, that is said for a language model previously trained on a copy of the entire web! In what sense is that "few shot"? It is truly shocking to hear researchers make such obviously false claims with a straight face. And that they are rewarded with publication of their work is nothing less than a scandal.
Deep learning research is going nowhere. It is stuck in a rut. It is dead. It has given up the spirit. It is an ex-research field. This research field is a goner. It is dead, jim. And I'm sure it will be decades before people start to say it out loud (the current generation of reserachers will leave the dirty job to their students), but that doesn't change a thing.
______________
[1] Except in identifying new weaknesses of deep learning systems, as for example Overinterpretation in the paper I cite above, adversarial examples [2], shortcut learning [3], "the elephant in the room" [4], being "right for the wrong reasons" [5] etc etc etc etc.
[2] https://arxiv.org/abs/1312.6199
[3] https://www.nature.com/articles/s42256-020-00257-z
[4] https://arxiv.org/abs/1808.03305
[5] https://aclanthology.org/P19-1334.pdf
Which of course are very welcome. The only way to address the limitations of current approaches is to become aware of them and try to understand them. Unfortunately, it is only a tiny, tiny minority in the field who does that, while the majority is happy to spout nonsense like "[a billion plus] few shot" learning etc.
[6] https://www.semanticscholar.org/paper/Speech-and-language-pr...
[7] https://www.semanticscholar.org/paper/Artificial-intelligenc...
[8] Arhtur Samuel's checkers player beat a human champion in 1961. Donald Michie build MENACE, a reinforcement learning algorithm implemented in matchboxes in 1960:
https://rodneybrooks.com/forai-machine-learning-explained/
Ctrl+F for "Machine learning started with games" to find the relevant section.
But you seem well read, so I don't think there is a point in trying to convince you.
It's also interesting coming from someone researching Inductive Logic Programming that provides no results at all.
From my perspective, I don't care for theoretical advances. I care about results and if scale provides results, I'm happy.
For people who are just reading this exchange take a look here:
https://pbs.twimg.com/media/FHHPjU0VIAUrSMA?format=jpg&name=...
It's 16 computer generated images from the description below each image. Having capability like that is just very useful and fun.
I wish you also luck in finding a "cat paying checkers in style of Salvador Dali" or a "psychedelic painting of a hamster dragon" on the internet.
There is a lot of stuff on the web, but not everything.
Like I say you can sort ever larger lists with ever bigger computers and bubblesort. But you can go a lot further if you use the brain and come up with a better sorting algorithm.
But I think you would be really surprised to find out how much humans have achieved by using their brains rather than ever bigger computers. Here's a small example that happened to be on my mind recently:
https://en.wikipedia.org/wiki/Copernican_Revolution
That is an example of the difference between expensive toys like Dall-E and world-changing scientific work. Though I appreciate that the difference can sometimes appear somewhat mudddled and a machine that can generate a "hamster dragon" may seem as an important scientific breakthrough. Depends on what you're more interested in: hamster dragons, or understanding how the world works. Oh well.
As to this:
>> It's also interesting coming from someone researching Inductive Logic Programming that provides no results at all.
That's an obvious -and odious- attempt at trolling me, but unlike yours, my HN profile is linked to my real-world identity and so I am forced to respond with politenes and to represent my field with dignity. Therefore, I will point out a famous achievement of ILP with which you are sadly unfamiliar and that I expect you to immediately discount on the grounds that you didn't know about it, so it can't have been all that important. I'm linking to an article in the popular press that will be more accessible. It doesn't go into depth over the machine learning approach used but it's ILP; I link to the scholarly article immediately after:
Robot Makes Scientific Discovery All by Itself
For the first time, a robotic system has made a novel scientific discovery with virtually no human intellectual input. Scientists designed “Adam” to carry out the entire scientific process on its own: formulating hypotheses, designing and running experiments, analyzing data, and deciding which experiments to run next.
https://www.wired.com/2009/04/robotscientist/
And the scholarly article:
Functional genomic hypothesis generation and experimentation by a robot scientist
There isn't really that much difference. DeepMind took some 3 years to max out CASP protein-folding benchmark running from 1994 on a problem known since ~1960. Protein folding is exactly the type of world-changing scientific work. But it is just a tip of the iceberg. DL is being successfully applied to PDE solving or electron density predictions and more.
The psychedelic hamster dragon is an example of the fact that DL is capable of 0-shot generalization. It is a significant scientific observation in itself.
And with regard to your example. We are discussing whether deep learning has staled. I give you mostly examples from last year like EfficientZero (30 Oct) ~500x improvement in sample efficiency over 2013 DQN, some literally not older than a month: GLIDE (20 Dec), Player of Games (6 Dec).
You are giving me example from 2004 (2009?).
I'm not trying to troll you. I just think you are biased in you assessment of significant results and what it means to be stalled.
Nope. Not at all. That's a misunderstanding of the goal of science which is to explain how the world works. AlphaFold, like all neural network models is a predictive model but has no explanatory power. It can predict the structure of proteins from sequences but it cannot explain why, or how, proteins fold. Scientists can still not explain how proteins fold, certainly not by interacting with AlphaFold.
Why is that a problem? You said you care about results. I linked to the wikipedia article about the Copernican Revolution. I recommend you read the wikipedia article on epicycles [1]. To summarise, for hundreds of years, from Hipparchus of Nicaea, through Claudius Ptolemy and all the way up to Copernicus, astronomers used systems of epicycles (circles-upon-circles) to explain the apparent movements of the planets. Epicycle-based models matched observations very well and predicted the movement of the planets very accuratey because it is always possible to fit a smooth curve with arbitrary precision given a sufficient number of epicycles. However, even Copernicus' model that had the sun in the center of the universe, unlike many earlier models, could not explain why the planets moved the way they did. A true explanation only became available when Isaac Newton developed his laws of universal gravitation.
Newton's theory (itself later superseded by Einstein's theory of general relativity) provided a true explanation of observable phenomena and a way to calculate better models and obtain more accurate predictions than were ever possible before. That is the power of explaining and understanding the world and simply modelling an incomprehensible set of observations, which is all that neural networks can do, doesn't even come close.
>> I'm not trying to troll you. I just think you are biased in you assessment of significant results and what it means to be stalled.
ILP is a tiny field with maybe a few dozen people that consistently publish on the subject. By comparison, there are many tens of thousands of papers published on deep learning every year, the deep learning conferences receive many thousands of submissions and research is funded with many millions of dollars by governments, militaries, and the largest of large technology corporatations, the Googles and Facebooks etc, and training state-of-the-art models requires many terrabytes of data, many petaflops of computing power, to the extent that training such models is now only feasible for the aforementioned large corporations. Normally, I'd complain that you are comparing apples to oranges. How can a few people with meager resources be expected to outdo DeepMind and OpenAI?
Yet the algorithms that I study run on a student laptop, are trained in a weakly supervised manner, take seconds to train, and generalise robustly from single examples without any sort of pre-training whatsoever (for instance, the approach I study, Meta-Interpetive Learning, comfortably and routinely learns recursive prorgams with arbitrary structure from a single example, without an example of the "base-case"). These are capabilities unheard of in neural networks that must be trained on hundreds of millions of examples [2] of actual programs (rather than examples of inputs and outputs, as in ILP) in order to generate programs and cannot solve programming tasks, nor generate solutions, unlike the examples tasks and solutions in their training sets; while in the case of ILP every programming task solved is, by definition unseen (hence, "weakly supervised" because examples are programming tasks without examples of the solutions).
That is what it means for a field to be stalled. When a single graduate student with a seven-year old laptop can solve problems that the dominant approach cannot even attetmpt (in this case, learning from a single example, without pretraining and with weak supervision).
____________
[1] https://en.wikipedia.org/wiki/Deferent_and_epicycle
[2] https://arxiv.org/abs/1703.07469v1
See Section 4.4. Hyperparameters and Training: "Each minibatch was re-sampled so the model saw 256 million random programs and 1024 million random I/O examples during training.".
Many complex problems do not appear to be explainable in the way planet motions are explainable. The fact that many aspects of the universe are explainable in simple laws is rather extraordinary in itself. In reality, there may be just a lot of irreducible complexity. Maybe success of Deep Learning teaches us that certain things do not fit into framework of science that you assume to be the only correct framework. Reality does not care about what we would want to be true.
Maybe invention of AlphaFold will lead to better understanding of protein folding the same way that invention of steam engine lead to the laws of thermodynamics. Historically, science often lags behind progress of technology. Time will judge.
I also thought that the lack of resources will be your answer. In 2004-2009 neural networks were in quite similar position to ILP. However, NN showed significant potential as evaluated by the broad community at the time and so resources allocated to NN started to grow exponentially. ILP and symbolic methods more broadly do not show similarly convincing potential to the broad community. I'm sure OpenAI, DeepMind, Google, Micosoft etc. would not hesitate to dump equivalent resources they do into many other projects if they saw potential. DeepMind, Microsoft and Google especially hire people with very broad range of expertise.
What may be true is that the world is starting to reach limits of the resources that can be dedicated to deep learning. Exponential growth in resource use can not go forever unless deep learning will start feeding back into available resources and we are certainly not there, not yet.
> Yet the algorithms that I study run on a student laptop, are trained in a weakly supervised manner, take seconds to train, and generalise robustly from single examples without any sort of pre-training whatsoever
Show me that applied to: https://github.com/fchollet/ARC
It's not impossible but I'm concerned that neural networks will only be used as an excuse to not have to understand anything anymore (not just in biology, in all of the sciences). That would be a terrible outcome- a dumbing down of science and an eventual loss of our ability to understand the world. As you say, time will be the judge of that.
>> Show me that applied to: https://github.com/fchollet/ARC
I had a look at ARC back when François Chollet's paper came out that presented it. I was interested because a) it's the kind of problem that ILP eats for breakfast and b) Chollet's name is well-known and it would be a chance to get our work noticed by people who normally wouldn't notice it.
However.
ARC is proposed as a benchmark that is hard for current big-data approaches and that would need elements of intelligent reasoning to solve (if I understand correctly, that's why you brought it up yourself?). Such benchmarks have been proposed before, in particular the Bongard problems [1], in machine vision, and the Winograd schemas [2], in NLP. As with ARC, the "defense" of such datasets against dumb, no-reasoning, big-data approaches is that there are few examples of each problem task. And yet, in both Bongard problems and Winograd schemas, neural nets have now achieved high accuracy. How did they do it? They did it by cheating: instead of training on the original, and few, Bongard problems, for example, people created generators for similar problems that could generate many thousands of problems [3]. That way, the neural nets folks had their big data and they could train their big networks. Same for Winograd schemas.
... And the same thing has already happened, at a preliminary stage, with ARC:
https://arxiv.org/abs/2011.09860
In the paper I link above, the authors use a data augmentation technique that consists of rotations and colour transformations of problem tasks. It should be obvious that this adds no useful information that can help any attempt at solving ARC problems with reasoning, and that it only serves to help a neural net better overfit to the training tasks. And yet, the system in the paper achieves good results in a small selection of ARC tasks. Admittedly, that is a very small selection (only 10x10 grids and not on any of the held-out test set that only François Chollet has access to) but the important point is that it is possible to make progress in solving ARC without actually showing any reasoning ability, despite the claims of the Kolev et al. paper above; and despite Chollet's intent for ARC to avoid exactly that.
This is pretty standard in deep learning and the reason why I gripe about benchmarks in my earlier comment. Deep learning systems consistently beat benchmarks designed to test abilities that deep learning systems don't have. They do this by cheating, enabled by weaknesses in said benchmarks. They cheat by overfitting to surface statistical regularities of the data (see earlier cited papers) thus learning "shortcuts" around the intended difficulties of the benchmarks. The trained systems are useless against any other dataset and their performance degrades precipitously once exposed to real-world data, but that doesn't matter, because a team that trains such a system to a new SOTA result will get to publish a paper and claim its academic brownie points, and it's very difficult to argue with tis "results" without beating the benchmarks yourself, because of the publishing climate described in Hinton's quote, above.
Of course, if you want to beat the benchmarks yourself, you have two options. One way is to also cheat and throw a bunch of data at the problem without attempting to understand it or really solve it. The other way is to do it the "right" way, to try to design a system that really demonstrates the abilities the benchmark is supposed to be measuring. Which can sometimes be done, but it takes a lot of time and effort. In the case of ARC, that means learning core priors.
If you've read the Chollet paper that introduced ARC, there's a section on "core priors", that the author says a system must possess before it can solve the ARC tasks. Those core priors must be learned (or worse, coded in by hand). So anyone who decides to solve ARC the way its creator intended will need to spend a great deal of time teaching their system core priors, even before they can start on the actual problems. Meanwhile, some big, 30-person team at a large tech corp with a few milllion dollars' budget will be throwing a gigaflop of compute and a few terrabytes of data on the problem and "solving" it without having to learn any "core priors", by finding some way around the dearth of training data, just as was done with Bongard problems and Winograd schemas. Then anyone who has published any earlier results obtained "the hard way" will be left looking like a loser. Why would I want to subject myself to such a humiliation and drag the reputation of my field through the mud?
So, no, I'm not "showing" you anything. Sorry. If you're interested in the results of ILP and MIL systems, or my own work, check out the ILP and MIL literature. Ask me if you want and I can give you some pointers.
__________________
[1] https://en.wikipedia.org/wiki/Bongard_problem
[2] https://en.wikipedia.org/wiki/Winograd_schema_challenge
[3] https://link.springer.com/chapter/10.1007%2F978-3-319-44781-...
See Section 2.1 "The Dataset": "For each class of each problem we generate 20000 training images. We also generate an additional 10000 images per class per problem as a testing set."
P.S. This paper compares GPT-2,3 to Louise (the system I created for my PhD) and also to humans and Magic Haskeller (an Inductive Functional Programming system, learning programs in Haskell) on some programming tasks in P3 (a Turing complete language with only 7 instructions, kind of a precursor of Brainfuck). I'm not affiliated with the authors:
https://proceedings.neurips.cc/paper/2021/hash/0cd6a652ed1f7...
Louise does alright, outperforming all three other systems and humans on the "very high complexity" tasks, but GPT-3 outperforms it on simpler tasks. In truth, Louise can do better than that, especially with a newer version, not available to the authors, and that has improved one-shot learning capabilities. On the other hand, the comparison is unequal: I was able to find the programming tasks and their solutions in a github repository predating the publication of GPT-3, so GPT-3 most likely ingested and memorised them, while it probably also benefited by the many examples of similar programming tasks in Brainfuck, that can be found online. Louise, as usual, learned in true one-shot fashion, without having seen examples of the tasks or their solutions before.
I agree, to a degree, with your assessment of the problem of cheating. Currently community is applying two fixes to that: 1) require few-shot learning ability 2) require good performance on many different few-shot learning benchmarks
My hobby is future forecasting and I was thinking for some time about creating a benchmark for AGI based on Bongard problems, ARC and Montezuma's Revenge. I did have a question about ARC that managed to surprise community: https://www.metaculus.com/questions/3762/will-early-2020-ai-... I have one about Montezuma's Revenge where I require some reasonable score after 30 min of game play, no specific to Montezuma's Revenge engineering allowed, not even retrieving rewards from internal game memory: https://www.metaculus.com/questions/5460/ai-rapidly-learning... I'm aware of the paper you mentioned around ARC and my fix to the benchmark would be to require few-shot generalization on task-level e.g. tested system, if it is a learning system, must have access only to couple of tasks for training before start of evaluation. The same with regard to Bongard problems. Requiring that all 3 benchmarks are solvable by one unified system IMO would be a reasonable test for general, visual intelligence.
DeepMind recently decided to include comparison of Gopher performance on Massive Multitask Language Understanding benchmarks vs. forecasters prediction. Forecasting communities seems to have gained some clout in AI circles (OpenAI is also cooperating with Metaculus on AI forecasting). Few-shot ARC forecast from Metaculus could give you an external estimation of the difficulty that you could use against reviewer 2.
Also, despite what you think about large language models memorizing everything few-shot SuperGlue is still standing: https://www.metaculus.com/questions/4932/when-will-ai-achiev... My intuitive evaluation of GPT-3 light text reasoning skills based on playing around with it matches few-shot SuperGlue score of 72 vs. 90 for avg. human.
> Meanwhile, some big, 30-person team at a large tech corp with a few milllion dollars' budget will be throwing a gigaflop of compute and a few terrabytes of data on the problem and "solving" it without having to learn any "core priors", by finding some way around the dearth of training data, just as was done with Bongard problems and Winograd schemas.
Compute is cheap and it reliably gets cheaper. Models like GPT-3 even with cost of tens of millions of dollars are economically justifiable. It's quite possible that OpenAI is currently turning profit on GPT-3. Real market performance is the ultimate benchmark of whether "performance degrades precipitously once exposed to real-world data".
Yep, you caught me. ILP is not magick. You actually have to do some work to train ILP algorithms.
To clarify, there is no "engineering fine-tuning" required. ILP approaches rely on background knowledge, which can be thought of as a library of sub-routines from which target programs are composed. ILP is unique in that the representation of background knowledge is the same as that of target programs (they're all logic programs) so they can learn their own background knowledge incrementally. So the effort to solve ARC would include training an ILP system to learn core priors from some initial set of low-level primitives. Once learned, core priors would be included in the background knowledge needed to actually solve ARC tasks (along with other stuff like rotations, translations etc that don't have to be learned because they're straightforward to hand-code). Clearly, that would take time and work. Deep learning approaches simply replace time and effort with data and compute. But there is always a cost to be paid. You can't have learning for free.
Learning core priors is not going to be simple. First of all, we don't even know exactly what they look like. Chollet gives a list, which I think is not exhaustive, but there's no dataset with examples that can be used in training. So there's a lot of work to be done, even just to figure out how to learn the core priors. Or you could do like on Kaggle and solve a sub-set of tasks with some geometric manipulations and a brute-force search. But that, too, would be cheating and it wouldn't generalise well anyway.
>> Requiring that all 3 benchmarks are solvable by one unified system IMO would be a reasonable test for general, visual intelligence.
I don't think that addresses the real issue with benchmarks. Machine learning benchmarks don't work because we have no idea what we want them to test, and that in turn because we don't have a theory of intelligence and artificial intelligence. Benchmarks like the Bongard problems, or ARC, now, are just educated guesses about the kind of behaviour an intelligent system should exhibit. Because they are just guesses, we're not really going anywhere and every few years someone comes up with a new benchmark, claims that this one is better than earlier ones, and then watches in horror as some dumb machine rips it to shreds. This will continue for ever until someone comes up with a plausible theory of what, exactly, it is that we are trying to test. Chollet actually tried that, but I think wrapping it all up in yet another new benchmark removes the focus from the theoretical work and focuses it on racing for SOTA again, thus exposing his work to the same old big-data shennanigans.
For example, 72% accuracy on a language understanding task should indicate pretty strong language understanding ability already _if_ the benchmark was actually testing language understanding. 72% accuracy would certainly beat a dog, or a chimp, it would probably beat a toddler and may even beat an educated adult who is not a native speaker of the test language. So it's clear that testing "few" or "any" shot makes no difference because no language understanding ability is being tested anyway. The benchmarks are fundamentally broken. They won't get any better by tweaking them. You can't make a thermometer measure luminosity no matter how much you tweak it.
Anyway pre-training, as I noted earlier, removes all legitimacy from any claims of "few shot" learning. First we need to undersand what is being learned during pre-training and if that has anything to do with some element of intelligent, then we can decide what "few shot" performance tells us.
>> Real market performance is the ultimate benchmark of whether "performance degrades precipitously once exposed to real-world data".
That sounds like you're saying that a neural network model's performance has something to do with how well it does as a commercial product. You can't mean that, so please explain. What's the market got to do with anything?
P.S. Interesting stuff on Metaculus.
For all we know, there are multiple ways to converge to a trainable input recognition system. And if ART is bringing comprehensibility to the table, it may look even less like what actual brains do than modern deep learning algorithms; nothing about what I know regarding how brains operate suggests to me that they are comprehensible as anything other than "a pattern that works."
Most of you have probably not heard of Stephen Grossberg. He is in my opinion one of the greatest neuroscientists of this generation. Maybe the greatest theoretical neuroscientist ever.
Neuroscience is swimming (barely) in a ton of data, but very little in the way of a unifying framework. Steve's body of work over 60 years provides that. ART is only one part of it. ART has some way to go before proving itself as undeniably useful as deep-learning on the application side of things. But it is very compelling as a model for how our (mammalian) brains might be operating.
One fundamental idea, which is worth considering in applications is how almost all of deep-learning is founded on the idea of an "error" (A mis-match between expectation and the prediction) and how this flows through the system. Backprop is obviously the most popularly known method for this flow. ART is fundamentally different in that it works with the idea of a match rather than a mis-match. This foundational even philosophical difference is interesting because if allows for learning to happen all the time even while perceiving is simultaneously happening. It also allows for learning to be unsupervised. This philosophical departure point is actually very interesting: If you rely entirely on the idea of an error and throw out the rest, can you even hope to build a conscious system? Steve btw also has what I think is the most elegant and mechanistically precise explanation of consciousness, which builds on ART. Even if you are skeptical of the current promise of ART as a replacement for deep-learning, his explanation of how ART-like systems enable consciousness (and different conscious percepts) are alone worth the price of admission.
I’ve written more about Stephen Grossberg and his book here: https://saigaddam.medium.com/the-greatest-neuroscientist-you...
Obviously back-propagation based NN have become very popular (even got a rebrand as Deep Learning), how has ART changed in the last 25 years, is there any interesting results using ART instead of DL?
The Wikipedia article is light on details [1], the 'types' described look basic and preliminary.
That's true of traditional models as well. The difference is that when there's a screw up it might be a little easier to determine the cause. That is certain useful, but if there are, on average, fewer screw ups with DL then the net benefit would seem to favor DL.
What I'd really like is to have direct access to a working ART model and dataset for it... I wasn't able to find one after a bit if searching though. (I am however on mobile at the moment which makes that a bit more cumbersome)
Without that, this sounds interesting but isn't something I can really evaluate beyond "researcher claims his research is better than status quo"
The performance problem with deep learning is well known. Great results up to 90%+ of the time, totally bogus a few percent of the time. Great for ad targeting. Automatic driving, not so much. If somebody got past that, it's important.
Please don't explain me why your model is better.
_show me_ your model is better with results.
I see a lot of: "this model is explainable and has no forgetting modes". But I cannot care about an explainable model that does not work.
There are many classic explainable models, how do they compare to art in real life applications?
"The problem with today’s AI, he says, is that it tries to imitate the results of brain processing instead of probing the mechanisms that give rise to the results."
Today's AI works really well for many applications (computer vision has leaped in quality). But of course I did not expect perfection from ImageNet trained models. I also did not expect gpt3 to be agi. But are they good (especially when compared with what we had not long ago)? Hell yeah
Also, the advancements in protein folding are superb.
I understand the author is not dismissing the current advancement. But if you claim your model is better you ought to compare it with the state of the art
Extremely high reviews on Amazon: https://www.amazon.com/Conscious-Mind-Resonant-Brain-Makes/p...
Anecdotal example: a few months ago I bought an electric shaver that was very inexpensive and had thousands of 5-star reviews. It failed two months later, and when I went to leave the negative review, I found that the product no longer existed, but another identical-looking product with a different name from a different(?) company was the best-reviewed cheap electric shaver.
Edit: not claiming or implying anything about the author, book or the reviews. Just pointing the curve doesn't look legit to me.
I don't know about this ART thing though, never heard of it.
Sure the SCII AI eventually became very strong but it’s also somewhat immersion breaking to see all the odd stuff it ends up doing. And that’s with a well tuned game, in something under active development you end up with a host of other issues.
> It has successfully been used to win at starcraft and dota now right?
They didn't manage to beat the best in either game, but they did get close to top players in both games.
I mean, playing well is obviously not a goal for making the AI for a single-player mode or a bot for a multi-player game; the goal there is to play in an entertaining way, and the optimal system probably is one that is simple, easy to test, and is beatable in a manner that's fun for the players. Probably the only niche (very narrow!) where you would actually want near-optimal AI is for making a "sparring partner" for training e-sports players.
People working on AI systems that are going to be actually used in games shipped to gamers is a separate community with (IMHO) almost no overlap.
Hard to assess vague claims without seeing something concrete.
Literally nothing in that repository makes an argument that this method is better than deep learning in any aspect whatsoever, they don't even attempt such a comparison.
Regarding a direct comparison, it may be difficult to do such a comparison by just running some code on data sets, which the one group of algorithms has been heavily optimized for with excessive funding, while the other has been not. Even more so, as such a review would have a difficult time to asses any of the critical features, like chances of memory loss. Comparing technologies isn't as easy as running them side by side.