AI Paper Atlas

Research map / Reinforcement learning / Visual and representation RL

Weakly-Supervised Reinforcement Learning for Controllable Behavior

Lisa Lee, Ben Eysenbach, Ruslan Salakhutdinov, Shixiang (Shane) Gu, Chelsea Finn

NeurIPS 2020 · Reinforcement learning › Visual and representation RL · approach: Latent-space world models · 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
Use WSC goal generation in disentangled latent space instead of VAE-space goalsSkewFit · Sawyer visual goal taskssuccess vs stepsImproved
Add auxiliary factor-identity prediction loss to SkewFit (Skewfit+pred) (ablation)SkewFit · Sawyer visual goal taskssuccess vs stepsHurt
Use disentangled distance for reward only, goals sampled in VAE space (SkewFit+DR) (ablation)SkewFit · Push n=3success vs stepsMixed
Replace WSC goal generation by VAE-space goals (ablation)WSC · Sawyer taskssuccess vs stepsHurt
Increase number of weak labels N from 128 to 4096 (ablation)WSC disentangled model · PushLights n=3 obj1_xPearson correlation: 0.44 -> 0.93Improved
Use 1024 pairwise labels (ablation)WSC disentangled model · all domainsPearson correlation: 0.8 or higherImproved
Increase number of weak labels (ablation)WSC · visual goal-conditioned RLtask performance: 1024 labels sufficientImproved
WSC with only 128 labels vs SkewFit on DoorLights (ablation)WSC · DoorLightstask performance: 128 labelsImproved
Drop weak labels for task-irrelevant factors (hand position, lighting) (ablation)WSC · PushLightstask performanceNo clear effect
Use weak supervision instead of unsupervised VAE for disentanglement (ablation)WSC · PushLights n=3 obj1_xPearson correlation: 0.02 -> 0.86Improved
Corrupt 5% of weak labels (ablation)WSC · PushLights n=1Pearson correlation (All): 0.797Mixed
Raise label noise from 5% to 10% (ablation)WSC · PushLights n=1Pearson correlation (All): 0.797 -> 0.638Hurt

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