ralogic

Start porting RAL to Pytorch/Tensorflow/MXNet, probably Pytorch. MVP would be nice, context for why it’s harder is ok, do awesome

environment setup

*.need conda and pip3 and MDAnalysis conda install tensorflow matplotlib scipy
pip3 install encodermap

smoke test: MNest

1. stock images implemented in tensorflow

high level questions

*. utility of combining activators and repressors

activation functions

tanh, relu. sigmoid, leaky relu, softmax, softplus

1. understand functions
2. multidimensional: -CNN convolution (5x5 -> 3x3 etc) -RNN recurrence (loops back output from previous predictions in time) -turning variable size inputs into a fixed size structure

further reading

Paperpile Archive

Learning from Data MOOC

Activation Functions

Tsetlin Machines

Tensorflow Playground

The Nielsen Textbook

Autoencoders Explained

Encodermap