Why train when you can optimize?

Re:Why train when you can optimize? by sicp-enjoyer on Saturday April 30, 2022 @02:08AM (#31213777)

I'm surprised to see almost no discussion of fourier series in that paper, considering fourier series is all about representing signals as linear combinations of sinusoidal functions.

[ Parent | on HN ]

Re:Why train when you can optimize? by PartiallyTyped on Saturday April 30, 2022 @02:22AM (#31213830)

You may be interested in [1] where they go to a great extend to show that the convolution operation that we consider in DL is the dual of fourier series [2].

[1] https://geometricdeeplearning.com

[2] https://arxiv.org/pdf/2104.13478.pdf page 27 (23 if you count book pages).

[ Parent | on HN ]

Re:Why train when you can optimize? by amelius on Saturday April 30, 2022 @08:00AM (#31215357)

Is convolution in DL not implemented with the FFT as the underlying workhorse?

[ Parent | on HN ]

Re:Why train when you can optimize? by PartiallyTyped on Saturday April 30, 2022 @09:41AM (#31216013)

Probably no since FFT is slower and less parallelizable than products.

[ Parent | on HN ]

Re:Why train when you can optimize? by anoy8888 on Saturday April 30, 2022 @08:58AM (#31215707)

I like your article and went to your home page to find more good articles and I like what I saw. Thank you for sharing. The only thing is that I am reading from my phone and the site is not very mobile friendly

[ Parent | on HN ]