How machines search for good answers when they cannot compute them directly. Climb hills, escape traps with controlled randomness, then evolve a solution.
3 units, about 3 hours. Free.
Implement hill climbing, watch it trap in a local optimum, and fix it with random restarts.
2Implement simulated annealing and explain why accepting bad moves early helps find the global peak.
3Implement selection, crossover and mutation, and evolve a target phrase from random letters.