The most cited neural networks all build on work done in my labs | Hacker News Reader

The most cited neural networks all build on work done in my labs

people.idsia.ch

23 pointsSmith4222 comments

Any claim Schmidhuber might have is always drowned out by how obstinate he is about demanding recognition.

He’s interrupted lectures at major conferences from other leaders in in the field to quibble about citations unrelated to the subject of the lecture - pretty much everything he does involves some whining about how the credit for his work has been stolen from him. It’s hard to take his claims seriously when he’s this pigheaded.

Can you provide more information about him?

The mental image I got purely from reading the blog/paper is quite intriguing...

His schtick is claiming he’s real “father of deep learning” and that all major work in ML has been built off of work done in his lab, and that he’s unjustly remained unrecognized while other pioneers in the field have received tons of awards and compensation from big tech companies.

When I say it’s his shtick, I mean that he basically brings it up whenever possible. When another big name in the field receives an award, he’ll write up a huge blog post about how ackshually they didn’t invent XYZ, but rather modified A and B, which is related to work C he did at his lab in 1980.

See https://people.idsia.ch//~juergen/critique-honda-prize-hinto...

https://people.idsia.ch//~juergen/naturedeepmind.html

https://people.idsia.ch//~juergen/critique-turing-award-beng...

At 1:03:00 in this video he interrupts a lecture at NIPS from Goodfellow (another leader in the field) to try and claim credit for the subject of the lecture in a sort of slimy way. https://youtube.com/watch?v=HGYYEUSm-0Q&feature=share

Honestly I’m not equipped to tell whether Schmidhuber’s claims are true since I haven’t followed the progression of ML research since the 1980s - but the impression I get of him is that he’s kind of a jackass who’s willing to spend a ton of time tearing others down in really petty ways. If there is some truth to his claims, perhaps the reason he’s unrecognized is because he’s a bit of an ass to work with?

I met him, he’s very nice in person (at least over the course of a day, say). Kind, patient and genuinely interested in other people.

So I doubt it is as simple as he’s a horrible person. But clearly he’s an outlier in terms of demanding recognition.

Tangent: While I've listened to him on podcasts, and am reading his Deep Learning textbook, this is the first time I've seen Goodfellow in person. I'm crushed to see that he is not, as I assumed, some 60 year old professor who has spent decades earning his expertise, and instead is someone young enough to pass as a college student!

A lot of context in his own post here, and the reply from Dr. Hinton is linked:

https://people.idsia.ch//~juergen/critique-honda-prize-hinto...

I stopped reading at [backpropagation] "wasn't created by "lots of different people" but by exactly one person who published first [BP1] and therefore should get the credit". Okay, so he was the first. Brilliant insight. Hugely valuable. He should get credit for that, and I'm very sympathetic if that hasn't happened (I don't follow other research enough to make any strong statement about if he has or hasn't received appropriate recognition). But that was 50 years ago and many equally insightful and impactful developments have gotten us to the point where backpropagation is being done in a practical way. "Exactly one person" is ridiculous.

I actually like Schmidhuber, and I don't mind his boisterous arrogance, but you have to admit there was a heavy dose of Schadenfreude there in him missing out of the Turing awarded jointly to Hinton, LeCun, and Bengio.

This is the one bit where I felt he was under-recognised. LSTMs for example are 100% his team’s work and that’s been a major thread in ANNs. Plus all the other bits...

Yeah, obnoxious or not, the Turing award is a bit too important for this kind of middle school drama.

Miron Livny does this too. He created the Condor project (a batch queue/executor system) and always interrupted talks about distributed computing to say he did it first.

Very few peoople used Condor because it was too hard to configure.

There’s a blast from the past… I actually built a headless opportunistic cluster using condor back in grad school so I could steal cpu cycles from the windows computer lab over nights and weekends.

https://www.linux.com/news/condor-building-linux-cluster-bud...

For what it's worth, even in that space, NQS was done before Condor (and based on an earlier system).

I've seen Condor used quite heavily around for long time, but originally only PVM, which was roughly contemporaneous.

yes but condor added a number of things that other folks later copied and (according to livny) didn't provide enough credit for, including (but not limited to) containers, glide-in, federation, secure communications, matchmaking (classads).

I saw him speak like 6-7 years ago and it was pretty clear that he was trolling and didn't take himself seriously.

Paired with the picture of him flexing his muscle, I think you can get a pretty clear understanding of his psychology.

As you see he always focuses one one thing. Here it is his biceps.

Time for everyone to start citing Newton and Leibniz.

This seems in jest, but who among us doesn’t cite Hipparchus when working on geospatial topics??

I'm curious, if his claim is correct, (and it seems like, from domain experts in this thread and others, he is), why has he seemingly not been recognized for his work, so consistently?

Also, assuming he is correct, what other alternative does he have for getting recognition? Everyone says this particular approach is obnoxious (making blog posts, interrupting conferences), but what non-obnoxious approaches are there to correct a deeply-entrenched mistake about NN-development credit? (I am genuinely curious here, I don't know enough about the CS research field to know if there is in fact alternative forums/channels to address something like this more politely).

I'm not sure what possessed me to do this, but:

1931: Kurt Gödel shows limits of math, logic, computing, AI - https://news.ycombinator.com/item?id=27536974 - June 2021 (296 comments)

Who Invented Backpropagation? (2014) - https://news.ycombinator.com/item?id=27127611 - May 2021 (1 comment)

Critique of 2018 Turing Award for Drs. Bengio and Hinton and LeCun - https://news.ycombinator.com/item?id=26804496 - April 2021 (20 comments)

Verifying Schmidhuber's Critique of Turing Award for Bengio and Hinton and LeCun - https://news.ycombinator.com/item?id=23649542 - June 2020 (7 comments)

Critique of 2018 Turing Award for Drs. Bengio and Hinton and LeCun - https://news.ycombinator.com/item?id=23642335 - June 2020 (1 comment)

Schmidhuber: Critique of Honda Prize for Dr. Hinton - https://news.ycombinator.com/item?id=22932838 - April 2020 (1 comment)

Deep Learning: Our Miraculous Year 1990-1991 - https://news.ycombinator.com/item?id=21166852 - Oct 2019 (34 comments)

Yann LeCun, Geoffrey Hinton and Yoshua Bengio win Turing Award - https://news.ycombinator.com/item?id=19499515 - March 2019 (146 comments)

Jürgen Schmidhuber on Consciousness (AMA on Reddit, 2015) - https://news.ycombinator.com/item?id=19241278 - Feb 2019 (3 comments)

Jürgen Schmidhuber says he’ll make machines smarter than us - https://news.ycombinator.com/item?id=17100132 - May 2018 (26 comments)

This Man Is the Godfather the AI Community Wants to Forget - https://news.ycombinator.com/item?id=17073202 - May 2018 (1 comment)

Deep Learning Conspiracy - https://news.ycombinator.com/item?id=16962938 - April 2018 (1 comment)

A Computer Scientist’s View of Life, the Universe, and Everything (1999) - https://news.ycombinator.com/item?id=14094373 - April 2017 (73 comments)

AI Pioneer Wants to Build the Renaissance Machine of the Future - https://news.ycombinator.com/item?id=13412050 - Jan 2017 (28 comments)

Jürgen Schmidhuber: The Problems of AI Consciousness Is Already Solved - https://news.ycombinator.com/item?id=13316530 - Jan 2017 (1 comment)

When A.I. Matures, It May Call Jürgen Schmidhuber ‘Dad’ - https://news.ycombinator.com/item?id=13066646 - Nov 2016 (50 comments)

When A.I. Matures, It May Call Jürgen Schmidhuber ‘Dad’ - https://news.ycombinator.com/item?id=13056794 - Nov 2016 (2 comments)

Ye LeCun replies back to Juergen Schmidhuber blog post #DeepLearningDrama - https://news.ycombinator.com/item?id=9824660 - July 2015 (3 comments)

Critique of Paper by “Deep Learning Conspiracy” - https://news.ycombinator.com/item?id=9807326 - June 2015 (33 comments)

https://news.ycombinator.com/item?id=9533749 (May 2015)

The LeCunn response is gone because Google+ is gone but I would love to hear the counter claims. I couldn't find anything googling it. If anyone has this handy, please post it.

> I'm not sure what possessed me to do this.

Because you are not only a gentleman but also a scholar, dang!

Reply on news.ycombinator.com