My second quarter at Stanford I did a research rotation with Dr. Eric Darve, researching pruning methods for Deep Neural Networks in an effort to make them more computationally efficient. I joined my research with the class project for CS230 with Andrew Ng. My research didn’t quite unfold in time for the class project, but I got some great experience in wrangling with deep learning. I ended up writing my report as a tutorial for helping others retrain their own images using the yad2k Keras implementation of Yolo v2. I retrained it on the KITTI benchmark driving set. While my current research is focused more on developing algorithms for robotic control through principled optimization, I think deep learning and machine learning have many promising areas of application and look forward to using it again in the future!

Final Report: Exploring Iterative Pruning in Deep Convolutional Neural Networks