Artificial Bee Colony: Swarm Optimization

Objective Landscape

Colony Parameters

Food sources (bees) 24

Abandonment limit 15

Cycles / second 1.5

Colony State

Best f(x,y)

—

Cycle

Scouts sent

Evaluations

x* = (0.00, 0.00)

How it works

The Artificial Bee Colony algorithm searches for the minimum of a function using three cooperating bee roles acting on a shared population of food sources (candidate solutions x = (x, y)):

fitness(x) = 1/(1+f(x))         if f(x) >= 0
           = 1+|f(x)|            if f(x) < 0

candidate: v_j = x_j + phi * (x_j - x_k_j),  phi in [-1, 1]
           (one random dimension j, one random other source k)

Employed bees — one per food source. Each perturbs its source along a random dimension toward/away from another random source, keeping the change only if fitness improves (greedy selection).
Onlooker bees — pick a source with probability proportional to its fitness (roulette-wheel selection: p_i = fitness_i / Σ fitness), then search its neighborhood the same way. This concentrates effort on promising regions.
Scout bees — if a source fails to improve for more than the abandonment limit consecutive cycles, it is discarded and replaced with a fresh random source, which lets the colony escape local minima.

This is a real metaheuristic optimizer (Karaboga, 2005), used today for antenna/sensor placement, neural-network hyperparameter tuning and job-shop scheduling — the same family of "swarm intelligence" algorithms as Particle Swarm Optimization and Ant Colony Optimization, but built entirely around this exploit/explore/abandon cycle.

Fresh food source Near abandonment Global best

⚠ Couldn't load the 3D engineThree.js failed to load from the CDN. Check your connection and reload.