Probabilistic Focal Search: accelerating bounded-suboptimal search via lower bounds
Source: arXiv cs.AI RSS政策與監管
arXiv:2609.10584v1. Bounded-suboptimal search aims to find solutions within an error factor of w while cutting search cost. Focal search (FS) uses a heuristic to choose among FOCAL nodes that satisfy the threshold w·f_min, but its deterministic policy can leave f_min unchanged across expansions. The authors propose Probabilistic Focal Search (PFS), which follows the FS guidance with probability p and otherwise expands the node with the lowest f in OPEN.