nicholasbaard Goto Github PK

followers: 6.0 following: 6.0 repos: 9.0 gists: 0.0

Name: Nic Baard

Type: User

Bio: RL Enthusiast! Exploring Reinforcement Learning! I'm inspired by DeepMind's work. Building on research to contribute to positive AI for humanity.

Location: South Africa

Nic Baard's Projects

cnns-for-image-classification

A Pytorch implementation of Convolutional Neural Networks, using LeNet5 for Image Classification

dqn

A replication of the Deep Q-Network Algorithm as seen in "Human-level control through deep reinforcement learning" - Mnih et. al. 2015

dynamic-programming

A python implementation of dynamic programming methods to solve the Bellman Equation.

linear-regression

A simple linear regression implementation in python from first principles using gradient descent.

logistic-regression

Implementation of Logistic Regression from first principles in python. For binary classification.

model-free-learning

An implementation of model-free learning techniques including SARSA, Q-Learning and SARSA-Lambda

A python implementation of the multi-armed bandit problem using reinforcement learning. The repo contains implementations of the epsilon greedy, optimistic initialization and upper confidence bound (UCB) methods.

multi-layered-perceptron

A python implementation of a feed forward neural network from first principles. The network is trained via gradient descent and backpropagation on the MNIST dataset.