Bin-Packing with GA + Wisdom of Crowds

Overview

A research paper and implementation pairing a genetic algorithm with a Wisdom-of-Crowds aggregation step and a greedy refinement pass to solve the NP-complete bin-packing problem. The GA evolves candidate packings via tournament selection, crossover, and mutation; WoC aggregates the most common bin assignments across independent runs; the greedy fixer guarantees valid, capacity-respecting bins.

Highlights

• GA with tournament selection, single-point crossover, and decaying mutation rate.
• Wisdom-of-Crowds aggregation over 10 GA runs to bias toward consistently-fit bin groupings.
• Greedy refinement (first-fit decreasing) to repair overpacked bins post-aggregation.
• Benchmarked GA vs GA+WoC across 10–50 item instances; WoC wins on optimality at higher item counts.