Synapse
⌘+K
Synapse
PulseExploreClubsResearchersJournals
Instagram
HomeClubsExplore
October 13, 2025Open Access

NeSyC: A Neuro-symbolic Continual Learner For Complex Embodied Tasks In Open Domains

View Full Paper
Ask AI
Bookmark
Share

Authors

WCWonje ChoiJPJinwoo ParkSASang Hyun Ahn

Discussion

Loading...

Member takes

Overview

NeSyC enhances embodied agents' performance on complex tasks, employing neuro-symbolic methods and LLMs.

Key Points

  • NeSyC demonstrates superior efficiency in executing complex tasks within open-domain environments.
  • Experiments reveal NeSyC's effectiveness across multiple benchmarks, including Minecraft and RLBench.
  • The framework translates limited experiences into actionable knowledge through a novel contrastive validation scheme.
  • Incorporating memory-based monitoring enhances error detection and promotes knowledge refinement across domains.

Cite This Study

Choi et al. (2025) studied this question.

synapsesocial.com/papers/68ece2abd1bb2827d129712chttps://doi.org/10.48550/arxiv.2503.00870
View Full Paper
Ask AI
Bookmark
Share