Timezone: »

Addressing Sample Complexity in Visual Tasks Using HER and Hallucinatory GANs
Himanshu Sahni · Toby Buckley · Pieter Abbeel · Ilya Kuzovkin

Tue Dec 10 10:45 AM -- 12:45 PM (PST) @ East Exhibition Hall B + C #208

Reinforcement Learning (RL) algorithms typically require millions of environment interactions to learn successful policies in sparse reward settings. Hindsight Experience Replay (HER) was introduced as a technique to increase sample efficiency by reimagining unsuccessful trajectories as successful ones by altering the originally intended goals. However, it cannot be directly applied to visual environments where goal states are often characterized by the presence of distinct visual features. In this work, we show how visual trajectories can be hallucinated to appear successful by altering agent observations using a generative model trained on relatively few snapshots of the goal. We then use this model in combination with HER to train RL agents in visual settings. We validate our approach on 3D navigation tasks and a simulated robotics application and show marked improvement over baselines derived from previous work.

Author Information

Himanshu Sahni (Georgia Institute of Technology)
Toby Buckley (Offworld Inc.)
Pieter Abbeel (University of California, Berkley & OpenAI)
Ilya Kuzovkin (OffWorld)

Machine Learning Architect @ OffWorld Inc, Pasadena, CA | Neuroscience and Machine Learning PhD student @ University of Tartu, Estonia

More from the Same Authors