================================================================================================== Lecture |Practice (Monday 11:30-12:50 at 106/2 Monday 11:30-13:00 Feb.Mar. Apr. | Apr.May. at Room 303/1 24 03 10 17 24 31 07 14 21 28 | 28* 05 12 19 ---------------------------------------------------------------------------------------------- Lee DoHwan o o o o o - o o o * | * o o o o Prokopenko Sergey o o o o - o o * | * - o - - ikhtar Andrey o o o - o o o * | * - o o - Dzmitry Samsonau o o o o o o * | * o o o o ============================================================================================== 01) Introduction to Evolutionary Computation: chromosome, gene, crossover, mutation, population, generation, etc. Also toy problem called "All-One." 02) Lucky dog designed by chromosome, and their exploration in a gridworld 03) Slide-show: Could Artificial Intelligence be intelligent? 04) Slide-show: Evolutionary Computation 05) NP-hard Problem: Knapsack Problem, Traveling Salesperson Problem (TSP), etc 06) Determing weight configulation of feed forward neural network by GA instead of backpropagation 07) Geme-theory by GA: Iterated Prisonner's Dillema 08) Sorting-network by GA - seeking an algorithm with minimum comparison. Is 60 minimum with 16 items? 09) Multi-objective GA: when we have multiple fitness criteria 10) Multi-modal GA: when we have multiple solutions -> XXX ----- [Practice] 02) Random-hillcriming & mutation-hill-criming on all-one, 20-D function minimization and lucky dog also IPD preparation: display 20 games of A (random) vs B (random, always-one, always-zero, tit-for-tat) 03) IPD: simulation of the game + Hillis' GA for sorting net: double helix => diploydy pressure can be seen? 04) Neural network: N-even-parity with weight configuration being by hillclimbing