At IEEE/SICE International Symposium on System Integrations (SII 2021) held in online, I presented following two topics.
“Behavioral Cloning from Observation with Bi-Directional Dynamics Model”
“Towards Deep Robot Learning with Optimizer Applicable to Non-Stationary Problems”
The first study is a joint research with an intern student from Technical University of Munich (TUM). It focuses on architectural improvements to achieve the high performance of imitation learning. The second research if for improving the SGD optimization method, which is essential for deep learning, for robot learning (i.e. adaptation to non-stationary problems).