Policy-Gradient-Network-Arduino

• Arduino/C++ implementation of Deep Policy Gradient Network with Reward-To-Go policy
• Custom built deep learning library for embedded controllers with floating point support
• Forward pass and backpropagation handled by means of custom matrix manipulation libraries written in C++
• Machine learning implementation close to popular libraries
• Built as final project for ENM 531: Data Driven Modelling and Probabilistic Scientific Computing course at UPenn
• Read more about it here
• Watch demo here.