Automated-Conference-Scheduler
Artificial Intelligence - Assignment 1
A Conference scheduler that uses a combination of greedy hill climbing and simulated annealing to find the best schedule. Also tried enforced hill climbing, hill climbing with uniform cost search, hill climbing with Iterative Deepening Depth First Search.
class descriptions : session : (set of papers going on parallely) int *papers int numPapers track : (set of sessions happening in a single time slot) session *sessions int numSessions conference : (time slot) track *tracks int numTracks (previously parallelTracks) int numSessions (previously sessionsInTrack) int numPapers (previously papersInSession)
additions :
-
probabilistic move to bad neighbours
-
restart after improvement goes below threshold
-
for analysis - plot score vs time
-
enforced hill climbing
-
simulated enealing
in3.txt
Observations :
better to boil at lower temperature, increase neighbors, increase break count for ghc 545109 - if(Temperature > 2) Temperature = 200 - itr / 4; else Temperature = 800 / itr; and neighbors = 250, count = 1000 545570- if(Temperature > 2) Temperature = 100 - itr / 5; else Temperature = 500 / itr; and neighbors = 250, count = 1000
maxima seen = 56136