bra-keton Hacker News • Recommends
On “Is AI Riding a One-Trick Pony?”
Prof. Hinton has an interesting talk about his new 'capsule' model based on psychology of shape perception: link to this video
0:00 / 0:00
#capsulenetworks#convolutionalneuralnetworks#geoffreyhinton
Geoffrey Hinton talk "What is wrong with convolutional neural nets ?"
trwappers
3.9K subscribers • 1 video on Hallway
- Recommended by 2 people: 2 on Hacker News.
- Presented by Geoffrey Hinton
- Linked in the descriptionarXiv 1710.09829Paper on openreview.net
- May be out of date. Uploaded 9 years ago, in an area where tools change fast.
- An opinion. It argues one side of a question engineers disagree on.
Advanced. Deep dive for people who know the field.
199K views9 years ago2 recommendations
AI and machine learningNeural networks and deep learningComputer vision
Geoffrey Hinton critiques convolutional neural networks and introduces his capsule networks model based on the psychology of shape perception.
Brain & Cognitive Sciences - Fall Colloquium Series Recorded December 4, 2014 Talk given at MIT. Geoffrey Hinton talks about his capsules project. Talks about the papers found here: arxiv.org/abs/1710.09829 and here: openreview.net/pdf?id=HJWLfGWRbWhy it's here
2 people recommend thisWhat people say. Recommended as a foundational talk explaining the limitations of CNNs and introducing capsule networks.
dr_zoidbergon Hacker News • Recommends
On “A New Twist on Neural Networks”
Hinton explains the concept of capsules in this video: link to this video Which is a lot better than reading someone tell you about this new idea that is called "capsules" but doesn't go into detail. The only thing is that, when this presentation was given, it seems they hadn't worked much more than MNIST (so the new thing now would be the toys-recognition net). Better source, with date: link (December 2014, for the lazy).
uoaeion Hacker News • Mentions
On “BadNets: Identifying vulnerabilities in the machine learning model supply chain”
I have a hypothesis that these errors in interpretation stem mainly from the max pooling layers that occurs in nearly all CNNs today. Simply cutting out a large fraction (at least 75%!) of the information passing through each pooling layer is making the models quite brittle. That, and obviously it would be a big help to be able to interpret meaningful symbols (e.g. words) from the images rather than just recognize a certain arbitrary class of pattern activations. That is pretty hard, of course, but IMO essential for the upcoming self-driving car revolution, if it is to succeed. G. Hinton has some ideas though they aren't very fleshed out yet.1 1 link to this video
giardinion Hacker News • Mentions
On “Ask HN: What's your favorite Machine Learning topic?”
Hinton's "capsule networks": link link to this video link They sound a bit like "grandmother neurons/cells": link I would very much appreciate an ELI5 of capsules and some prognostications about how capsules may affect the state of the art in neural networks and the current flood of papers and graduate students in the pipeline of ML.
nabla9on Hacker News • Mentions
On “Deciphering China’s AI Dream”
…l begin adopted and refined. It's just alternating layers of affine transformations and nonlinearity with lots of tricks and improved routing. There needs to be fundamental theoretical breakthrough in every decade or so, and not just refinements and new applications. Hinton made his "What is wrong with convolutional neural nets" speech years ago " link to this video I believe that Hinton's capsule networks type attempts are what we need more. They may not deliver immediately impressive results because fundamental research involves lots of setbacks.…