Title: Deep Learning to Learn (keynote about the state of the art in Reinforcement Learning)
About Pieter Abbeel
As founder of Covariant, Director of the Berkeley Robot Learning Lab and Co-Director of the Berkeley Artificial Intelligence (BAIR) Lab, Pieter Abbeel pioneers at the front-end of teachable AI Robotics. His work is often featured in the press, including the New York Times, Wall Street Journal and Tech Review. We are delighted to welcome him at the TU Delft for a talk about “Deep Learning to Learn” for the special occasion of opening our AI for Retail (AIR) Lab Delft.
“Reinforcement learning and imitation learning have seen success in many domains, including autonomous helicopter flight, Atari, simulated locomotion, Go and robotic manipulation. However, the amount of data required for these methods remains very high. In contrast, humans can pick up new skills far more quickly. To do so, humans might rely on a better learning algorithm or on a better prior (potentially learned from past experience), and likely on both. In this talk I will describe some recent work on meta-learning for action, where agents learn the imitation/reinforcement learning algorithms and learn the prior. This has enabled acquiring new skills from just a single demonstration or just a few trials. While originally designed for imitation and RL, our work is more generally applicable and also advanced the state of the art in image recognition from a small number of examples.”
About AIRLab Delft
The AI for Retail (AIR) Lab Delft is a joint TUDelft-Ahold Delhaize industry lab consisting of a robotics research program and test site focused on developing state-of-the-art innovations in the retail industry. AIRLab Delft will further drive innovations for daily business while building more knowledge of the intersection between retail, AI and robotics. This second AIRLab location comprises a robotics research programme and test site for developing state-of-the-art innovations in the retail industry. More info can be found here.
Prof. dr. ir. M. (Martijn) Wisse
Director – TU Delft Robotics Institute