Author

Name: Wenjun Wu

GT username: wwu94

Project Info

Implementation Platform: MATLAB 2016b and ABAGAIL Java Package ([https://github.com/pushkar/ABAGAIL)]

Wilt Dataset:

~/wwu94/wilt.mat

This project experimented with four random search algorithms: randomized hill climbing, simulated annealing, genetic algorithm and MIMIC.

Neural Network Weight Optimization

Randomized Hill Climbing (RHC)

​ execute Neural Network\hill\nn_usinghill.m

​ to print out result: NN\hill\print.m

​ change 'MaxIter' to the number of iterations desired

​ results are stored in .mat files

Genetic Algorithms

​ execute Neural Network\ga\nn_usingga.m

​ to print out result: NN\hill\print.m

​ change 'MaxIter' to the number of iterations desired

​ results are stored in .mat files

Simulated Annealing

​ execute Neural Network\sa\nn_usingsa.m

​ to print out result: NN\hill\print.m

​ results are stored in .mat files

Optimization Problems

Knapsack Problem

​ execute ABAGAIL\src\opt\test\TravelingSalesmanTest.java

​ change N to be the desired number of items

​ enter the desired iterations

Traveling Salesman Problem

​ execute ABAGAIL\src\opt\test\KnapsackTest.java

​ change N to be the desired number of items

​ enter the desired iterations

Count Ones Problem

​ execute ABAGAIL\src\opt\test\CountOnesTest.java

​ change N to be the desired number of items

​ enter the desired iterations