AI Paper Atlas

Research map / Multi-agent systems and games / Human-AI and goal-conditioned

Learning to Predict Without Looking Ahead: World Models Without Forward Prediction

Daniel Freeman, David Ha, Luke Metz

NeurIPS 2019 · Multi-agent systems and games › Human-AI and goal-conditioned · approach: Intrinsically motivated exploration · 20 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 observational dropout with peek probability p=10%Joint policy and world-model training on swing-up cart-pole · Swing-up cart-polecumulative score: higher than ∼ 500Improved
Use observational dropout with peek probability p=5%Joint policy and world-model training on swing-up cart-pole · Swing-up cart-poletask success rate: solves the task half of the timeMixed
Train the world model jointly with the policy using observational dropout rather than a forward-predictive lossSwing-up cart-pole transfer, 120-unit models · Swing-up cart-pole transfercumulative scoreImproved
Increase the forward-predictive model width by 10xForward-predictive cart-pole baseline · Swing-up cart-pole transfercumulative score: 10xNo clear effect
Use a convolutional instead of a fully connected world modelObservational-dropout grid-world agent · Grid worldreward: 10650 parameters; 1201 learnable parametersImproved
Use a world model trained at lower peek probabilitiesLinear policy using frozen world-model features · Car Racingscore: around 40%Improved
Reduce peek probability too close to zeroJoint policy and world-model training on Car Racing · Car RacingscoreHurt
Remove VAE noise by using the encoder's predicted mean as the latent observation (ablation)Car Racing agent · Car RacingscoreImproved
Replace the convolutional world model with a fully connected world model (ablation)Observational-dropout grid-world agent · Grid worldreward: 10650 parameters; 1201 learnable parametersHurt
Train with a peek probability close to p=50%Convolutional grid-world world model · Grid worldnext-frame prediction correlation: p=50%Improved

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