AI 용어집
인공지능 완전 사전
Variable Neighborhood Search (VNS)
Metaheuristic based on the systematic exploration of different neighborhood structures to escape local optima and find globally optimal solutions.
Shaking
Random perturbation phase in VNS using a neighborhood structure to generate a starting solution far from the current local optimum.
Local Search
Iterative improvement phase applied after shaking to converge toward a local optimum in the neighborhood of the perturbed solution.
Variable Neighborhood Descent (VND)
Deterministic variant of VNS sequentially exploring different neighborhood structures until no further improvement is possible.
Reduced VNS
Simplified variant of VNS applying local search directly to the current solution without an intermediate shaking phase.
General VNS
Extended version of VNS incorporating advanced neighborhood change strategies and exploration-exploitation balancing mechanisms.
Skewed VNS
VNS variant introducing a bias to favor solutions distant from the reference solution, useful for avoiding premature convergence.
Neighborhood Change
Mechanism determining when and how to switch between different neighborhood structures based on improvement or stagnation criteria.
Perturbation Control
Adaptive strategy controlling the intensity of the shaking phase based on the quality of found solutions and the number of iterations without improvement.
Neighborhood Sequence
Predefined or dynamic order of exploring different neighborhood structures influencing the convergence and diversification of the search.
Variable Neighborhood Decomposition Search (VNDS)
Hybridization of VNS with decomposition techniques solving sub-problems on variable parts of the solution.
Multi-start VNS
Approach executing VNS from multiple different initial solutions to increase the probability of finding the global optimum.
Parallel VNS
Parallel implementation of VNS simultaneously exploiting multiple neighborhood structures or executing independent searches in parallel.
Hybrid VNS
Combination of VNS with other metaheuristics such as simulated annealing, genetic algorithms, or tabu search to improve performance.
Adaptive VNS
VNS variant dynamically adapting parameters and neighborhood structures based on performance history during the search.
Multi-objective VNS
Extension of VNS to multi-objective optimization problems managing a set of Pareto-optimal solutions with specific diversification mechanisms.
Dynamic Neighborhood
Approach where neighborhood structures evolve dynamically during the search according to the characteristics of the explored solution landscape.