Top all-time · #305
Jürgen Schmidhuber
Personal blog · people.idsia.chaimath
Jürgen Schmidhuber's personal blog at people.idsia.ch covers ai and math. Since 2014, 164 different Hacker News readers have linked it 255 times, most often “Annotated History of Modern AI and Deep Learning (2022)”. No feed was found, so read it on the site itself.
Sometimes recommended · 8 of 40 sampled comments recommend it?
Why HN links it
HN readers link Schmidhuber's personal site for his deep learning history and credit-assignment claims, and for essays on novelty, curiosity and beauty. Commenters call the site a rabbit hole and a balanced history.
- Recommended for · 8 comments
- Called 'useful, accurate', 'more balanced', and 'particularly interesting'; readers told to read it or check it out.
Summarised by Claude from a sample of 40 HN comments that link here. Each comment is marked recommending, critical or neutral (citing without judging); how this works.
Most-linked posts?
- Annotated History of Modern AI and Deep Learning (2022)10 people · 11 mentions · first linked 2022-12-24
- Formal Theory of Creativity and Fun and Intrinsic Motivation (2010)8 people · 10 mentions · first linked 2015-04-08
- Critique of Paper by “Deep Learning Conspiracy”8 people · 8 mentions · first linked 2015-06-30
- Juergen Schmidhuber's home page7 people · 7 mentions · first linked 2015-05-08
- Scientific Integrity, the 2021 Turing Lecture, and the 2018 Turing Award6 people · 8 mentions · first linked 2021-09-24
- Turing Oversold6 people · 6 mentions · first linked 2021-09-14
- Jürgen Schmidhuber's Home Page6 people · 6 mentions · first linked 2021-04-15
- A Nobel Prize for Plagiarism5 people · 5 mentions · first linked 2024-12-08
- 1991: NN learns to program NN fast. 2021: NN learns to program Transformers fast4 people · 5 mentions · first linked 2021-03-27
- Gödel Machine4 people · 5 mentions · first linked 2018-01-16
What Hacker News says?
Wow, they look quite artistic. Jürgen Schmidhuber has done extensive study of so-called low complexity art, see http://people.idsia.ch/~juergen/beauty.html
Comment linking /~juergen/beauty.html
on “Show HN: Primitive Pictures”Submitted Turing Oversold? as a story · 354 points
HN discussionDirect link with the pdf version: http://people.idsia.ch/~juergen/DeepLearning2July2014.pdf Unfortunately, a good PR is definitely important for prizes.
Comment linking /~juergen/DeepLearning2July2014.pdf
on “Yann LeCun, Geoffrey Hinton and Yoshua Bengio win Turing Award”Submitted 1931: Kurt Gödel shows limits of math, logic, computing, AI as a story · 347 points
HN discussionI encourage downvoters to read the article below for a perspective on how LeCun, Hinton and Bengio may have unfairly excluded a wealth of prior work in their version of deep learning history: http://people.idsia.ch/~juergen/deep-learning-conspiracy.htm... Note for people put off by the link: The conspiracy refers to the title of a paper by the Turing award winners, not Schmidhuber's.
Comment linking /~juergen/deep-learning-conspiracy.html
on “Yann LeCun, Geoffrey Hinton and Yoshua Bengio win Turing Award”Submitted Who Invented Backpropagation? as a story · 200 points
HN discussionFor a more balanced view of the history of deep learning see Schmidhuber‘s review http://people.idsia.ch/~juergen/deep-learning-overview.html and http://people.idsia.ch/~juergen/who-invented-backpropagation... .
Comment linking /~juergen/who-invented-backpropagation.html
on “Yann LeCun, Geoffrey Hinton and Yoshua Bengio win Turing Award”Submitted LeCun's 2022 paper on autonomous machine intelligence does not cite prior work as a story · 164 points
HN discussionJürgen Schmidhuber introduced curiosity/boredom mechanisms as a way to improve learning in reinforcement learning environment: https://people.idsia.ch/~juergen/curiositysab/curiositysab.h... This mechanism can be formalized. > Zero reinforcement should be given in case of perfect matches, high reinforcement should be given in case of `near-misses', and low reinforcement again should be given in case of strong mismat…
Comment linking /~juergen/curiositysab/curiositysab.html
on “Human”Submitted Deep Learning: Our Miraculous Year 1990-1991 as a story · 156 points
HN discussionHe was done a great deal of injustice when he was passed over for the Turing award that was given to Hinton, Bengio, LeCun. Then there is this from his blog -- Dec 2024: Sadly, the Nobel Prize in Physics 2024 for Hopfield & Hinton is a Nobel Prize for plagiarism. They republished methodologies developed in Ukraine and Japan by Ivakhnenko and Amari in the 1960s & 1970s, as well as other techniques, without citing the…
Comment linking /~juergen/physics-nobel-2024-plagiarism.html
on “Animals Made from 13 Circles (2016)”Submitted Who invented deep residual learning? as a story · 114 points
HN discussionwow, that's beautiful - the whole site https://people.idsia.ch/~juergen/ is an amazing rabbit hole im gonna lose myself in.
Comment linking /~juergen/
on “Animals Made from 13 Circles (2016)”Submitted Critique of Paper by “Deep Learning Conspiracy” as a story · 81 points
HN discussionSee also work from Schmidhuber in the mid/late 1990s https://people.idsia.ch/~juergen/locoart/node12.html
Comment linking /~juergen/locoart/node12.html
on “Animals Made from 13 Circles (2016)”Submitted Critique of 2018 Turing Award for Drs. Bengio and Hinton and LeCun as a story · 50 points
HN discussionhttps://people.idsia.ch/~juergen/1991-unnormalized-linear-tr...
Comment linking /~juergen/1991-unnormalized-linear-transformer.html
on “Google Titans architecture, helping AI have long-term memory”Submitted Composing Music with LSTM Recurrent Networks – Blues Improvisation (2002) as a story · 41 points
HN discussionI've seen this idea used in machine learning as well. There was a paper [1] whose goal was a binary classification of the center pixel in a region of interest. Interestingly, their results improved when they applied a foveal blurring surrounding the pixel to be classified. [1]: http://people.idsia.ch/~juergen/nips2012.pdf
Comment linking /~juergen/nips2012.pdf
on “A Canvas Made of Pixels”Submitted Jürgen Schmidhuber – Can All-Purpose Robots Fuel a Comeback? as a story · 36 points
HN discussionHis view aligns both with a leading neuroscience explanation of the brain (predictive coding [1]) and with Active Inference / the Free Energy principle [2] from optimal control theory. A similar theory of intelligence, called H-JEPA (hierarchical joint embedding predictive architecture) [3] is also put forward by Yann LeCun, a major AI pioneer. Another AI pioneer, Jürgen Schmidhuber, subsequently criticized LeCun's…
Comment linking /~juergen/lecun-rehash-1990-2022.html
on “An LLM is a lossy encyclopedia”Submitted Turing Awardees Republished Key Methods and Ideas Without Credit as a story · 32 points
HN discussion> Methodologically, how do we create agents that aren't just good at several tasks, but make up their own tasks It's a good question, it has been asked a few times, and there are some answers[1][2] already, with the most general being to endow the agent with intrinsic motivation defined as an information-theoretic objective to maximize some definition of surprise . Then the agent in question will develop a general c…
Comment linking /~juergen/interest.html
on “DeepMind: A Generalist Agent”Submitted Jürgen Schmidhuber's Home Page as a story · 28 points
HN discussionTuring would be a better example, perhaps. See https://people.idsia.ch/~juergen/turing-oversold.html
Comment linking /~juergen/turing-oversold.html
on “Geniuses of the past were aristocratically tutored”Submitted The most cited neural networks all build on work done in my labs as a story · 23 points
HN discussionAh yes, Perceptron. Had a couple typos.. sorry, was on phone. The bio-inspiration was via Frank Rosenblatt, who is referred to in that article tho yeah, the history is over in his article: https://en.wikipedia.org/wiki/Frank_Rosenblatt#Perceptron "Rosenblatt was best known for the Perceptron, an electronic device which was constructed in accordance with biological principles and showed an ability to learn. He develo…
Comment linking /~juergen/deep-learning-overview.html
on “Understand how transformers work by demystifying the math behind them”Submitted A Nobel Prize for Plagiarism as a story · 19 points
HN discussionWhat do you mean, the development of the "Perception"? Do you mean the Perceptron? In that case, Backprop was invented way later than the Perceptron (see https://people.idsia.ch/~juergen/who-invented-backpropagatio... ). I don't see any information in your linked Wikipedia article that supports a bio-inspired origin. In fact, researchers have been wondering whether an equivalent to Backprop might be found in biologi…
Comment linking /~juergen/who-invented-backpropagation.html
on “Understand how transformers work by demystifying the math behind them”Submitted Who invented convolutional neural networks? as a story · 5 points
HN discussion
