Best-First Heuristic Search for Multicore Machines

Abstract

To harness modern multicore processors, it is imperative to develop parallel versions of fundamental algorithms. In this paper, we compare different approaches to parallel best-first search in a shared-memory setting. We present a new method, PBNF, that uses abstraction to partition the state space and to detect duplicate states without requiring frequent locking. PBNF allows speculative expansions when necessary to keep threads busy. We identify and fix potential livelock conditions in our approach, proving its correctness using temporal logic. Our approach is general, allowing it to extend easily to suboptimal and anytime heuristic search. In an empirical comparison on STRIPS planning, grid pathfinding, and sliding tile puzzle problems using 8-core machines, we show that A*, weighted A* and Anytime weighted A* implemented using PBNF yield faster search than improved versions of previous parallel search proposals.

Improving lung cancer prognosis assessment by incorporating synthetic minority oversampling technique and score fusion method

“…Thus, wrapper approaches are generally considered to produce better feature subsets. 22 In this study, a wrapper approach based on classifier subset evaluator and BestFirst search method 23 was used for feature reduction.…”

Scalable and optimal planning based on Pregel

“…31 Some of these studies took a multi-thread approach on a multi-core machine. 32 Later, they extended the A* algorithm into a shared-memory parallel environment with a hash function which assigned each state to a unique process.…”

“…This idea was later developed into the shared‐memory parallel planning method . Some of these studies took a multi‐thread approach on a multi‐core machine . Later, they extended the A* algorithm into a shared‐memory parallel environment with a hash function which assigned each state to a unique process. Some researchers put their emphasis on exploiting distributed memory computing clusters .…”

Distributed Heuristic Forward Search for Multi-agent Planning

“…Parallel planning methods (Vrakas, Refanidis, & Vlahavas, 2001;Kishimoto, Fukunaga, & Botea, 2009;Burns, Lemons, Ruml, & Zhou, 2010) aim to speed-up the solution of centralized planning problems given access to a distributed computing environment, such as a large cluster. Parallel planning can be performed in ma-strips by applying existing approaches to the underlying strips problem (i.e., ignoring agent identities).…”