Deep Learning Course (NYU, Spring 2020) - Week 11 – Practicum: Prediction and Policy learning Under Uncertainty (PPUU)

Course website: http://bit.ly/DLSP20-web Playlist: http://bit.ly/pDL-YouTube Speaker: Alfredo Canziani Week 11: http://bit.ly/DLSP20-11 0:00:00 – Week 11 – Practicum PRACTICUM: http://bit.ly/DLSP20-11-3 This practicum proposed effective policy learning for driving in dense traffic. We trained multiple policies by unrolling a learned model of the real world dynamics by optimizing different cost functions. The idea is to minimize the uncertainty in the model’s prediction by introducing a cost term that represents the model’s divergence from the states it is trained on. 0:01:03 – Introduction to the problem and Deterministic predictive decoder 0:29:39 – Variational predictive model and Action insensitivity 0:53:52 – Training the agent with different strategies and evaluating them