Breadth-First Search vs. Restarting Random Walks for Escaping Uninformed Heuristic Regions
This work addresses the problem of heuristic search inefficiency in AI planning, but it is incremental as it compares existing escape methods without introducing a new paradigm.
The paper compares breadth-first search (BrFS) and restarting random walks (RRWs) for escaping uninformed heuristic regions (UHRs) in greedy search methods, showing that EHC-RRW variants using RRWs have strong expected runtime guarantees and are evaluated on PDDL planning benchmarks.
Greedy search methods like Greedy Best-First Search (GBFS) and Enforced Hill-Climbing (EHC) often struggle when faced with Uninformed Heuristic Regions (UHRs) like heuristic local minima or plateaus. In this work, we theoretically and empirically compare two popular methods for escaping UHRs in breadth-first search (BrFS) and restarting random walks (RRWs). We first derive the expected runtime of escaping a UHR using BrFS and RRWs, based on properties of the UHR and the random walk procedure, and then use these results to identify when RRWs will be faster in expectation than BrFS. We then evaluate these methods for escaping UHRs by comparing standard EHC, which uses BrFS to escape UHRs, to variants of EHC called EHC-RRW, which use RRWs for that purpose. EHC-RRW is shown to have strong expected runtime guarantees in cases where EHC has previously been shown to be effective. We also run experiments with these approaches on PDDL planning benchmarks to better understand their relative effectiveness for escaping UHRs.