AI Paper Atlas

Research map / Reinforcement learning / Visual and representation RL

Iteratively Learn Diverse Strategies with State Distance Information

Wei Fu, Weihua Du, Jingwei Li, Sunli Chen, Jingzhao Zhang, Yi Wu

NeurIPS 2023 · Reinforcement learning › Visual and representation RL · approach: Hierarchical skills and priors · 1 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
iterative learning (ITR) instead of population-based training (PBT), N_L=4policy diversity search · 2-D navigationlandmarks discovered: 2.0 -> 3.5Improved
iterative learning (ITR) instead of population-based training (PBT), N_L=5policy diversity search · 2-D navigationlandmarks discovered: 2.2 -> 4.5Improved
fix Lagrange multiplier instead of two-timescale GDA (ablation)SIPO-WD · GRF 3v1distinct strategies: 3.0 -> 1.0Hurt
replace state-distance intrinsic reward with action cross-entropy (ablation)SIPO-WD · GRF cornerdistinct strategies: 3.0 -> 1.7Hurt
replace state-distance intrinsic reward with action cross-entropy (ablation)SIPO-WD · GRF CAdistinct strategies: 3.0 -> 2.3Hurt
replace state-distance intrinsic reward with action cross-entropy (ablation)SIPO-WD · GRF 3v1distinct strategies: 3.0 -> 2.7Hurt
use filtering-based optimization instead of Lagrangian constraint (ablation)SIPO-WD · GRF 3v1distinct strategies: 3.0 -> 1.3Hurt
use filtering-based optimization instead of Lagrangian constraint (ablation)SIPO-WD · GRF cornerdistinct strategies: 3.0 -> 1.0Hurt
train M policies simultaneously (PBT) instead of iteratively (ablation)SIPO-WD · GRF 3v1distinct strategies: 3.0 -> 2.7Hurt
SIPO-WD instead of RSPOSIPO · Humanoiddiversity: 0.53 -> 0.71Improved
SIPO-RBF instead of RSPOSIPO · Humanoiddiversity: 0.53 -> 0.53No clear effect
SIPO-RBF instead of RSPOSIPO · SMAC 2m_vs_1zkNN state entropy: 0.032 -> 0.038Improved

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