Genetic Algorithms
Also known as GA
Overview
Genetic algorithms evolve populations of candidate solutions through selection, mutation, and crossover, making them useful when objectives are constrained, discrete, noisy, or difficult to optimize with gradients. Research spans problem-specific representations and operators for tasks such as vehicle routing, feature selection, scheduling, and neural or circuit design, as well as hybrids with reinforcement learning, Bayesian optimization, and learned models. Runtime analysis on benchmark functions studies how population size, diversity, mutation, and parameter adaptation affect convergence, while practical work evaluates solution quality, feasibility, and computation cost on domain benchmarks. A central challenge is preserving exploration and diversity while scaling to large problems and ensuring gains transfer beyond the instances used for tuning.
Latest papers
49 papers2 papers in the last four weeks, down 60% on the four weeks before. 0.0% of all new papers.