AI Meetup Series with Rocky Duan

Sep 7, 2017 · San Francisco, United States of America

We are thrilled to announce that on September 7th we are hosting Rocky Duan!

Brief BioRocky Duan is a PhD student at UC Berkeley with Pieter Abbeel.  He works on reinforcement learning, robotics, unsupervised learning, and meta learning.

Talk Title: Meta Learning and Control

Talk Abstract: Traditionally AI researchers design machine learning algorithms.  This has led to impressive results in supervised learning (e.g. image classification, speech recognition, machine translation) and reinforcement learning (e.g. Atari, Go, Dota 2, robot control).  However, these algorithms still learn substantially slower than humans. Meta learning enables machines themselves to design machine learning algorithms, which can be tailored towards domain-specific application scenarios, achieving significantly faster learning than human-designed, general-purpose learning algorithms. In this talk we will explore meta learning frameworks for reinforcement learning and imitation learning, two leading paradigms for learning control policies that can not only sense, but also act, in their environment.

Main references for the talk are:  RL^2 and One-Shot Imitation Learning.

Agenda:

6:00 pm to - 6:30 pm   Check-in, Networking

6:30 pm to -  7:15 pm  Rocky's talk, Q&A

7:15 pm   Networking

See you all,

Event organizers

Are you organizing AI Meetup Series with Rocky Duan?

Claim the event and start manage its content.

I am the organizer
Social
Rating

based on 0 reviews