There's this crowd on HN which is very vocal against academia. From what I've seen, the main points are that academia isn't efficient, most of the science coming out of academia is useless and that the whole system is just a waste of taxpayers money. Instead, what is often argued, all good research is done in private labs. Then pointing to SpaceX, Moderna, OpenAI, Google, etc.
And while it is very true that often the research coming out of Academia is useless, what is always neglected are the roots of the research done in private labs.
When Jürgen Schmidhuber and team published their work on Neural Nets back in 1991 it was also useless. Unless you had a supercomputer and very, very deep pockets you were not going to do anything with what came out of their lab.
But still, 30 years later here we are, standing on top of the shoulders of this useless research.
I do a lot of work that is based on academic research, aka building a proprietary sparse embedding model. My issue with academia is that they don’t bother to solve the practical issues. They tell you how to build a PPMI model, but what about hitting a database that’s 500TB to find co-occurrence numbers? This isn’t even touched so you’d then have to go and invent a bazillion of algorithms yourself to make your life easier. So while the bedrock is based on academic research and we thank them for that, scaling anything requires a lot of work in uncharted territories.
Like half of what Schmidhuber is always complaining about is that (except for LSTMs) people aren't standing on the shoulders of his research very much. They try to solve some of the same problems people have always wanted to solve, try some of the same approaches people always tend to try, and then tinker until it works. At no point do they consult Schmidhuber's decade-old papers where he tried something kind of similar but didn't get very impressive results, and hence they also do not think to cite him. Then he comes out of the woodwork to assert priority.
Shane Legg was in Schmidhuber's lab at IDSIA before being one of the founders of DeepMind, so he probably read the papers personally and knows what influenced him or not...
The closest to that that I've seen is that traditional academia approaches are too far removed from practical applications for highly applied fields like software engineering, or too slow for fast-moving fields like modern day ML (thus, all the preprints).
Practically no one is against hard science research, properly conducted. The issues are rampant fraud / p-hacking / unreproducible garbage mixed with an unhealthy dose of ideological monoculture and indoctrination, garnished with rising tuition prices while sitting on huge endowments in case of the Ivy Leagues.
It's sad that he is the only one speaking out about Hinton. This whole Hinton glorification seems like it's being pushed by an agenda. I'm not sure if he would receive this much attention if he held a different view (closer to LeCun or Ng), rather than these Effective Altruism takes on current AI.
Which work has more value: the abstract description of a catalogue of potential model architectures or their validated application trained on real data?
In the Schmidhuber case their is 20 years and a chain of countless other works in between the two.
Surely the roots, if we skip over the early preceptron work', are in backpropagation and Hinton, and the work going on at Edinburgh and elsewhere in the 80s.
Indeed I remember buying a set of three conference-papers-as-books around that time, titled Artificial Neural Networks .. proceedings of the whatever the conference was.
No doubt Schmidhuber made important contributions, but I see him pop up claiming to be the 'root' of it all every couple of years.
Modern backpropagation was first published by Seppo Linnainmaa as "reverse mode of automatic differentiation" (1970)[26] for discrete connected networks of nested differentiable functions.[27][28][29]
In 1982, Paul Werbos applied backpropagation to MLPs in the way that has become standard.
Paul Werbos did not apply backprop to MLPs as cleanly described in Hinton's paper, but rather to some kind of autoregressive non-linear parametrized functions with a much more specific application scope.
Both papers are direct applications of the chain rule applied to estimate the gradient of a multivariate function.
That's what bugs me about him. So much work has gone into today's models that calling his contributions "the root" isn't really warranted. He's always complaining that Hinton, LeCun, and Bengio get more credit than they deserve, and now he's over-claiming himself.
I believe invention of Transformers and especially Attention mechanism do have influence from past research but its not definitely only the Schmidhuber's work. Said that, if we remove the papers mentioned by Schmidhuber from history, I am quite certain that there will be no influence in the discovery of Transformers, hence his works can not be the root. He has to grow up and accept that work and equations can appear similar, looking at inverse squared law and saying Newton stole that from someone is being dishonest.
And while it is very true that often the research coming out of Academia is useless, what is always neglected are the roots of the research done in private labs.
When Jürgen Schmidhuber and team published their work on Neural Nets back in 1991 it was also useless. Unless you had a supercomputer and very, very deep pockets you were not going to do anything with what came out of their lab.
But still, 30 years later here we are, standing on top of the shoulders of this useless research.
And that's where Schmidhuber goes off the rails: publicly shaming published papers into citing you isn't good academic practice. It's bullying.
The closest to that that I've seen is that traditional academia approaches are too far removed from practical applications for highly applied fields like software engineering, or too slow for fast-moving fields like modern day ML (thus, all the preprints).
Practically no one is against hard science research, properly conducted. The issues are rampant fraud / p-hacking / unreproducible garbage mixed with an unhealthy dose of ideological monoculture and indoctrination, garnished with rising tuition prices while sitting on huge endowments in case of the Ivy Leagues.
However I often see this going from "there's issues" to discounting academia altogether and positioning private labs as a good alternative.
After all, most people in the open science collaboration which published the seminal paper kicking off the replication crisis were from academia.
In the Schmidhuber case their is 20 years and a chain of countless other works in between the two.
Indeed I remember buying a set of three conference-papers-as-books around that time, titled Artificial Neural Networks .. proceedings of the whatever the conference was.
No doubt Schmidhuber made important contributions, but I see him pop up claiming to be the 'root' of it all every couple of years.
related paragraph from Wikipedia:
Modern backpropagation was first published by Seppo Linnainmaa as "reverse mode of automatic differentiation" (1970)[26] for discrete connected networks of nested differentiable functions.[27][28][29]
In 1982, Paul Werbos applied backpropagation to MLPs in the way that has become standard.
Both papers are direct applications of the chain rule applied to estimate the gradient of a multivariate function.