The HSIC Bottleneck: Deep Learning without Back-Propagation · HackerPapers discussions

The HSIC Bottleneck: Deep Learning without Back-Propagation — HN submission

Tue, 30 Jan 2024 10:04:49 GMT

The HSIC Bottleneck: Deep Learning Without Back-Propagation

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

The HSIC Bottleneck: Deep Learning without Back-Propagation — mentioned by andreyk

Fri, 16 Aug 2019 05:07:17 GMT

It's good to be wary of hype, but this comment seems overly down on where we are by focusing on AI today doing 'curve fitting' ; sure, supervised learning is at it's core that, but when it achieves things like image captioning or super accurate translation calling it just curve fitting is reductionist and downplays how powerful these techniques are. Furthermore, the notion that neural networks can only be used as part of supervised learning (and in particular with backpropagation) is totally off. Just to cite a few examples: http://proceedings.mlr.press/v48/taylor16.pdf…

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

The HSIC Bottleneck: Deep Learning without Back-Propagation — HN submission

Thu, 15 Aug 2019 08:26:33 GMT

The HSIC Bottleneck: Deep Learning Without Back-Propagation

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

The HSIC Bottleneck: Deep Learning without Back-Propagation — mentioned by marmaduke

Wed, 14 Aug 2019 22:07:07 GMT

Mentions a so called HSIC bottleneck, maximizing mutual information between hidden states and labels, minimizing mutual dependence between input and hidden states (?) PDF here https://arxiv.org/pdf/1908.01580v1.pdf Another place I recently saw Reproducing Kernel Hilbert Spaces was in Stein Variational Gradient Descent (PyMC3 implements it). It seems like a clever basis set but have to admit I don't understand what's going on.

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

The HSIC Bottleneck: Deep Learning without Back-Propagation — HN submission

Mon, 12 Aug 2019 15:30:38 GMT

The HSIC Bottleneck: Deep Learning Without Back-Propagation

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