# Hill Climbing **Domain:** Optimization / Search / Artificial Intelligence **Doc Type:** Canonical Method Node **Maturity:** Foundational ## Definition **Hill climbing** is a local search method that repeatedly moves from a current solution to a neighboring solution with a better objective value. ## Technical Context Because the method uses local improvement, it can stop at a [[wiki/Local Optimum|local optimum]] even when a better solution exists elsewhere in the [[wiki/Optimization Landscape|optimization landscape]]. Gradient methods are one important continuous case, but hill climbing is the broader search concept. ## Historical Context [[wiki/Geoffrey Hinton|Geoffrey Hinton's]] 1977 thesis treated entrapment at local maxima as the usual objection to hill climbing and argued that the objection did not apply inside the particular feasible space he constructed. ## See Also [[wiki/Gradient Descent|Gradient Descent]], [[wiki/Optimization|Optimization]], [[wiki/Relaxation Labeling|Relaxation Labeling]]