Emergence of Invariance and Disentanglement in Deep Representations · HackerPapers discussions

Emergence of Invariance and Disentanglement in Deep Representations — HN submission

Sun, 07 Jan 2018 13:12:33 GMT

Emergence of Invariance and Disentangling in Deep Representations

3 points; 0 discussion replies on HN. Paper and source history. A submission is not an endorsement.

Emergence of Invariance and Disentanglement in Deep Representations — mentioned by AlexCoventry

Sat, 30 Dec 2017 19:02:21 GMT

Emergence of Invariance and Disentangling in Deep Representations https://arxiv.org/abs/1706.01350 "we show that in a deep neural network invariance to nuisance factors is equivalent to information minimality of the learned representation, and that stacking layers and injecting noise during training naturally bias the network towards learning invariant representations. We then show that, in order to avoid memorization, we need to limit the quantity of information stored in the weights, which leads to a novel usage of the Information Bottleneck Lagrangian on the weights as a learning criterion"

Paper and discussion history. A mention is not an endorsement.

Emergence of Invariance and Disentanglement in Deep Representations — mentioned by AlexCoventry

Wed, 08 Nov 2017 03:37:06 GMT

I'm a bit surprised that Soatto's and Tishby's papers aren't on the reading list (https://stats385.github.io/readings). I think they have some of the most interesting theories, at the moment, about why Deep Learning works. https://arxiv.org/abs/1706.01350 https://openreview.net/pdf?id=ry_WPG-A-

Paper and discussion history. A mention is not an endorsement.

Emergence of Invariance and Disentanglement in Deep Representations — mentioned by backpropaganda

Wed, 02 Aug 2017 10:29:46 GMT

The kind of math that would be necessary to understand papers like [1][2][3][4]. [1]: https://arxiv.org/abs/1703.04933 [2]: https://arxiv.org/abs/1701.07875 [3]: https://arxiv.org/abs/1706.01350 [4]: https://arxiv.org/abs/1512.04860

Paper and discussion history. A mention is not an endorsement.

Emergence of Invariance and Disentanglement in Deep Representations — HN submission

Fri, 16 Jun 2017 00:41:36 GMT

On the Emergence of Invariance and Disentangling in Deep Representations

3 points; 0 discussion replies on HN. Paper and source history. A submission is not an endorsement.