AI Paper Atlas

Research map / Reinforcement learning / Visual and representation RL

Adjacency Constraint for Efficient Hierarchical Reinforcement Learning

Tianren Zhang, Shangqi Guo, Tian Tan, Xiaolin Hu, Feng Chen

TPAMI 2022 · Reinforcement learning › Visual and representation RL · approach: Subgoal generation for hierarchical RL · 73 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 k-step adjacency constraint to goal-conditioned HRL (vs Vanilla)HRAC · Discrete tasks and Ant tasksepisode reward / success rateImproved
use learned adjacency instead of oracle adjacency (ablation)HRAC · Ablation taskslearning curve performanceNo clear effect
use adjacency matrix-based training instead of NoAdj trajectory-pair training (ablation)HRAC · Ablation taskslearning curve performanceImproved
use differentiable adjacency loss instead of negative-reward penalty (ablation)HRAC · Ablation taskslearning curve performanceImproved
increase balancing coefficient eta (ablation)HRAC · Ablation taskslearning curve performanceImproved
add adjacency constraint vs HIRO under additive state noise sigma=0.1HRAC · Stochastic Ant Gather/Mazelearning curve performanceImproved
use HIRO reward shaping with sparse rewardsHIRO · Key-Chest and Ant Maze Sparseepisode rewardHurt
add hindsight experience replay to low levelHRL-HER · Locomotion taskssuccess rateHurt
use hierarchy (HIRO) vs flat DDPGHIRO · FetchPush and FetchPickAndPlacesuccess rateImproved

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