neupy

neupy copied to clipboard

Feature Request: Implementation of SET training method

1

Could Spare Evolutionary Training be implemented as a training method? From reading up on it, from my understanding it is somewhat similar to Adam, but I likely am grossly oversimplifying...

DPR-Sanchez

Thought I should include this material here in case a new base activation function layer would be useful to add out of the box: https://mc.ai/comparison-of-activation-functions-for-deep-learning-initial-winner-ftswish/

DPR-Sanchez

For example: http://neupy.com/modules/generated/neupy.layers.Convolution.html#neupy.layers.Convolution

itdxer

itdxer

itdxer

itdxer

Current implementation doesn't work for cases when classifier has single output with sigmoid activation. Problem is only with negative classes.

itdxer

Will there be GPU support for competitive models?

5

I know this is a probably a difficult task. But since we are switching to Tensorflow, maybe we could add something like https://github.com/cgorman/tensorflow-som in the future?

lejinghu

3D convolution and pooling

1

itdxer

itdxer