News for hackers. Stuff that matters.
Re:Visual Introduction to Self Supervised Learning by alleycat5000 on Friday February 28, 2020 @02:18PM (#22445466)
Some really interesting work lately on "contrastive" learning, where the accuracy is really getting on par with supervised learning, e.g. https://arxiv.org/abs/2002.05709
Re:Visual Introduction to Self Supervised Learning by amitness on Saturday February 29, 2020 @11:44AM (#22451432)
Yup, this paper is really interesting. I'm reading it currently and will be posting an illustrated post soon on it.
Re:Visual Introduction to Self Supervised Learning by amitness on Friday March 06, 2020 @11:30AM (#22504862)
Here is the illustrated post on SimCLR: https://amitness.com/2020/03/illustrated-simclr/
Re:Visual Introduction to Self Supervised Learning by jjoonathan on Friday February 28, 2020 @02:48PM (#22445715)
For those of us out of the loop, could you summarize the idea of contrastive learning as a whole?
Re:Visual Introduction to Self Supervised Learning by yazr on Friday February 28, 2020 @03:32PM (#22446063)
A fully illustrated article [1]

And Lilian Weng blog on self-supervision [2]

.. CPC is .. translating a generative modeling problem to a classification problem... uses cross-entropy loss to measure how well the model can classify the “future” representation amongst a set of unrelated “negative” samples...

[1] https://ankeshanand.com/blog/2020/01/26/contrative-self-supe...

[2] https://lilianweng.github.io/lil-log/2019/11/10/self-supervi...

Re:Visual Introduction to Self Supervised Learning by jjoonathan on Friday February 28, 2020 @03:41PM (#22446155)
Thanks!
Re:Visual Introduction to Self Supervised Learning by Voloskaya on Friday February 28, 2020 @05:41PM (#22447093)
(one variant of) The task is: Given a crop of an image (or a short audio snippet etc.), can you find the matching crop that also comes from the same image from a set containing a lot of negative samples (crops of others, unrelated images)?

To succeed, the encoder needs to be able to extract the underlying, useful information (called slow features) contained in the patch and discard the noise as this will make the retrieval process much easier.

This yields an encoder that gives pretty good representations of your inputs and you can then finetune some additional layers on top of it for your final task.