AI Paper Atlas

Research map / Reinforcement learning / Visual and representation RL

Addressing Sample Complexity in Visual Tasks Using HER and Hallucinatory GANs

Himanshu Sahni, Toby Buckley, Pieter Abbeel, Ilya Kuzovkin

NeurIPS 2019 · Reinforcement learning › Visual and representation RL · approach: Robust visual encoders for RL · 5 citations (OpenAlex)

Read the paperCheck your idea against 137,217 papers

AI-extracted. The measured effects below were read by AI from the paper's full text (no summary yet). Numbers and claims can be wrong or missing context, so check the paper before you cite or build on it.

Measured effects reported in the paper

What was changedWhereMeasuredEffect
Add HALGAN goal hallucinations to HER replayDDQN · TurtleBot pebble collectionlearning onsetImproved
Add HALGAN goal hallucinations to HER replayDDPG · MiniWorld Navigatetask completionImproved
Add HALGAN goal hallucinations to HER replayDDPG · MiniWorld Pick&Placetask completionImproved
Use naive HER without visual goal hallucinationDDQN · TurtleBot pebble collectionlearning onset: 100K stepsImproved
Use naive HER without visual goal hallucinationDDPG · MiniWorld Navigatetask completionHurt
Use naive HER without visual goal hallucinationDDPG · MiniWorld Pick&Placetask completionHurt
Use VAE-based reward reassignment with HER instead of hallucinating goalsDDPG · MiniWorld Navigatetask completionHurt
Use VAE-based reward reassignment with HER instead of hallucinating goalsDDPG · MiniWorld Pick&Placetask completionHurt
Use dense rewards from random goal imagesRIG- · MiniWorld Navigatetask completionMixed
Use dense rewards from random goal imagesRIG- · MiniWorld Pick&Placetask completionMixed
Reduce HALGAN training snapshots to 1000 images (ablation)HALGAN+HER · TurtleBot pebble collectionlearning speed: 1000 imagesHurt
Add an L2 penalty to the generator output to encourage minimal image edits (ablation)HALGAN · HALGAN goal hallucinationhallucination minimalityImproved

Closest papers by meaning

Writing a paper in this area? scime draft is a free LaTeX editor with AI included, real-time co-authors and a check of your references and venue rules before you submit. It also imports OpenAI Prism projects.

Open scime draft

Get the next scime Dispatch and a monthly field update

Free. We send a confirmation link first; every email has a one-click unsubscribe. Privacy