News for hackers. Stuff that matters.

A comment on Neural Networks Are Essentially Polynomial Regression.

Re:Neural Networks Are Essentially Polynomial Regression by thermonot on Friday August 30, 2019 @01:41PM (#20840744)
Important point:

> Our work so far has primarily been on general feedforward NNs. Our investigations have not yet involved much on specialized networks such as convolutional NNs (CNNs) for image classification, recurrent NNs (RNNs) for text processing, and so on. Though we intend to adapt our ideas to these frameworks, we view them as separate, orthogonal issues.

[ Parent | on HN ]
Re:Neural Networks Are Essentially Polynomial Regression by ta0xdeadbeef on Friday August 30, 2019 @03:35PM (#20841825)
>Though we intend to adapt our ideas to these frameworks, we view them as separate, orthogonal issues.

The authors should read about dynamical systems. I don't understand why the physics/controls folks, DSP folks, and ML folks don't read each other's papers. They're using the same math and using the same concepts. The latter two groups seem to be working together more often however.

[ Parent | on HN ]