PI: Jeff Krichmar
Institution: University of California, Irvine
Neuromodulation for learning and acting of the reinforcement learning algorithms:
We are trying to apply the mechanisms in neuromodulation to reinforcement learning and use the signals from neuromodulation
to modulate the learning and acting of the reinforcement learning algorithms.
Motion Decomposition in Dynamic Scenes: This project focuses on learning continuous egomotion representation from monocular video and pose information. The trained model infer ego motion parameters and object velocities for unseen videos.