Synapse
⌘+K
Synapse
PulseExploreClubsResearchersJournals
Instagram
HomeClubsExplore
July 26, 2026Journal of Artificial Intelligence ResearchOpen Access

Hierarchical Reinforcement Learning with Optimal Level Synchronization Based on Flow-Based Deep Generative Model

View Full Paper
Ask AI
Bookmark
Share

Authors

JKJaeYoon KimJXJunyu XuanCLChristy Liang

Discussion

Loading...

Member takes

Overview

Randomized trial demonstrates improved policy training efficiency in hierarchical reinforcement learning, indicating enhanced performance in complex environments.

Key Points

  • This research aims to address the inefficiencies in training hierarchical reinforcement learning (HRL) policies by proposing a novel model based on flow-based deep generative techniques.
  • Proposed a new HRL model utilizing direct off-policy correction with flow-based deep generative modeling.
  • Conducted comparative experiments on benchmark environments to evaluate performance against existing models.
  • Demonstrated superior performance compared to traditional HRL methods in benchmark settings.
  • Showed reduced inefficiencies in training higher-level policies with direct off-policy corrections.

Cite This Study

Kim et al. (2026) studied this question.

synapsesocial.com/papers/6a65a2e2d3aea3239cd7626ahttps://doi.org/10.1613/jair.1.19264
View Full Paper
Ask AI
Bookmark
Share

Also Consider

Synapse has enriched 5 closely related papers on similar clinical questions. Consider them for comparative context:

  1. 1Bidirectional-Reachable Hierarchical Reinforcement Learning with Mutually Responsive Policies2024
  2. 2Hierarchical Message-Passing Policies for Multi-Agent Reinforcement Learning2025
  3. 3Flow-Based Policy for Online Reinforcement Learning2025
  4. 4Hierarchical Reinforcement Learning from Demonstration via Reachability-Based Reward Shaping2024 · 4 citations
  5. 5Flow to Control: Offline Reinforcement Learning with Lossless Primitive Discovery2023 · 2 citations