Rudraksh Kapil

Backpropamine Implementation

Skills: C++, Python, AI, OpenGL, Reinforcement learning, Research, Version control, Visualization, Technical documentation, LaTeX.

Description: This is my term project for the Data Science course at NIT Warangal that I completed in third year. The task was to reimplement any research paper in the domain of machine learning, so my group members and I decided on this one.

The goal of the project was to use neuromodulated plastic weights in recurrent neural networks to improve the performance of agents on reinforcement learning tasks like maze navigation and cue reward association. You can find the final presentation we delivered as well as the videos of the tasks after training here.

Cue-reward association task