# Ant Colony Optimization ## Synopsis From Chapter 6, *Swarm intelligence: Ants* in the book [[Grokking AI Algorithms, Second Edition]]. ## Deciding Next Visit Probability to choose path $x$ for the next visit: $ \frac{P_x^a H_x^{-b}}{\sum_{n} P_n^a H_n^{-b}} $ where - $P_n$: pheromones on path n - $H_n$: heuristic for path n - $a, b$: parameters to tune the weights of influences of $P$ and $H$. ## Updating Pheromones - Evaporation: degrading over time - Add $\frac{1}{\sum{H}}$ to visited path for each Ant that has finished a tour with total heuristic $\sum{H}$.