项目作者: oorejola

项目描述 :
Fully Connected Forward Feed Neural Network
高级语言: Python
项目地址: git://github.com/oorejola/Forward-Feed-Neural-Network.git
创建时间: 2019-02-27T15:11:37Z
项目社区:https://github.com/oorejola/Forward-Feed-Neural-Network

开源协议:

下载


Forward-Feed-Neural-Network

This project is dead. Maybe I’ll touch it again in the future. Probably not.

This is a Toy Project for understanding a Full Connected Forward Feed Neural Network (FFNN) as well as implement back-propogation algorithm. I constructed a this FFNN using only matrix operations from numpy (Matrix Products, vector dotproducts, vector outer products.
It was important for me to gather a better mathematical idea of the FFNN and Back-Propogation algorithm prior to using
better /optimized tools. I.e TensorFlow, PyTorch, etc.

To update:

-Update indices for backpropogation algorithm for Stochastic Gradient Descent (SGD) for improved readability

-Completely Classify FFNN following E Fiesler’s paper “Neural Network Classification and Formalization”