Deep-Learning-in-Julia

A hands-on approach with Julia implementing deep neural networks from scratch with a functional API.

This repo is an academic exercise on imlpementing a deep neural network from scratch in the Julia language with API similar to the Keras Functional API ... <3 Keras. The goal of the project is to have the ability to build a basic neural network from a short list of customizable layers, activation functions, and optimization algorithms with code profile that is as small as possible with no dependencies. Hopefully there will be some sort of wiki associated with this repo to walk through parts of the implementation.

Project Under Construction

This is a personal project that just happens to be public. Like most of my personal projects, it's not done. Please stay on the line while I transfer you to a more polished project.

Current Status:

• Ability to build network model structure
• Dense & Input layers
• Basic activation functions (linear, relu, sigmoid, softmax)
• Forward propagation
• Loss functions & Evaluation
• Activation & Loss function derivatives
• Back-propagation algorithm & training
• Dropout Layers
• Other things I haven't thought of yet
• Correctly organize modules into a Julia package