Neural Networks

A Testspace for Neural Networks and Machine Learning Projects in Python

Projects:

• A Pong playing AI using Policy Gradient Reinforcement Learning (Src).
  • Pong game from OpenAI's Gym package
  • Full writeup can be found here
• Classifies Breast Cancer tumors as malignant or benign (Src)
  • Single-Layer Perceptron Network
  • Trained on the Breast Cancer Wisconsin Data Set
  • Full writeup can be found here
• Classifies handwriten digits (0-9) (Src)
  • Deep Convolutional Network
  • Trained on the MNIST dataset
  • Simplified Graph of the Network
  • I implemented this network into an android app
• Classifies handwriten digits (0-9) (Src)
  • Single-Layer Perceptron Network
  • Trained on the MNIST dataset
• Can model basic monotonic functions (Src)
  • Single-Layer Perceptron Network
  • Implemented from scratch as a proof of concept of neural networks.

Coded in Python 3.5 with:

• Google's TensorFlow
• NumPy

Useful References:

• Deep Learning by Ian Goodfellow, et al.
  • Comprehensive description of deep learning.
  • Also includes a review of prerequisite mathematics (Linear Algebra, Information Theory, etc).
• Neural Networks and Deep Learning by Michael Nielson
  • Introduces and constructs neural networks from first principles.
• Essence of Linear Algebra by Grant Sanderson
  • Video series visually explaining Linear Algebra.
• Gradient Descent Derivation by Chris McCormick
  • Explains and derives graident descent using calculus.
• Discrete Mathematics and its Applications by Kenneth H. Rosen