Saved in:
| Main Author: | |
|---|---|
| Format: | Preprint |
| Published: |
2026
|
| Subjects: | |
| Online Access: | https://arxiv.org/abs/2604.14345 |
| Tags: |
Add Tag
No Tags, Be the first to tag this record!
|
Table of Contents:
- As search depth increases in autonomous reasoning and embodied planning, candidate action spaces expand exponentially, often exhausting computational budgets. While heuristic pruning is a critical countermeasure, existing approaches lack formal safety guarantees when guided by surrogate evaluators such as Large Language Models (LLMs), which exhibit systematic biases. We formulate node expansion as a localized Best-Arm Identification (BAI) problem under bounded bias $L$ and derive a sample complexity upper bound of $\mathcal{O}((Δ-4L)^{-2})$, identifying $Δ> 4L$ as the regime where safe elimination is feasible. We further establish an information-theoretic lower bound of $Ω((Δ-2L)^{-2})$ that characterizes the structural limits of biased exploration. Motivated by these results, we propose PAC-MCTS, a bias-aware pruning framework that dynamically adapts confidence bounds during search. Experiments on Blocksworld and ALFWorld demonstrate that PAC-MCTS consistently improves robustness and search efficiency over strong pruning baselines, achieving up to 78\% fewer API evaluations and over 3$\times$ higher sample efficiency under strict compute budgets. Ablation studies further validate the predicted degradation behavior as evaluator bias increases.