GitHub - efancher/cs234_work_final

Branches Tags

Name		Name	Last commit message	Last commit date
Latest commit History 19 Commits
poster		poster
project_update		project_update
Agent.jl		Agent.jl
ChainMDP.jl		ChainMDP.jl
ChainMDP2.jl		ChainMDP2.jl
Final Project - Maximum Reward Path.ipynb		Final Project - Maximum Reward Path.ipynb
Final Project CS238.ipynb		Final Project CS238.ipynb
Final_Project_Parallel_Chains.ipynb		Final_Project_Parallel_Chains.ipynb
MaximumRewardPath.jl		MaximumRewardPath.jl
ParallelChains.jl		ParallelChains.jl
Readme.txt		Readme.txt
RunBP.jl		RunBP.jl
RunMR.jl		RunMR.jl
RunPC.jl		RunPC.jl
Untitled.ipynb		Untitled.ipynb
final_report.log		final_report.log
mr_chart.R		mr_chart.R
mr_simulation_1.csv		mr_simulation_1.csv
mr_simulation_2.csv		mr_simulation_2.csv
mr_simulation_2_runs_36_40.csv		mr_simulation_2_runs_36_40.csv
mr_simulation_2_runs_41_45.csv		mr_simulation_2_runs_41_45.csv
mr_simulation_2_runs_46_46.csv		mr_simulation_2_runs_46_46.csv
mr_simulation_2_runs_47_50.csv		mr_simulation_2_runs_47_50.csv
mr_simulation_2_runs_51_55.csv		mr_simulation_2_runs_51_55.csv
mr_simulation_2_runs_56_60.csv		mr_simulation_2_runs_56_60.csv
mr_simulation_all3_1run.csv.bkup		mr_simulation_all3_1run.csv.bkup
mr_simulation_all3_1run.csv.bkup2		mr_simulation_all3_1run.csv.bkup2
pc_chart.R		pc_chart.R
pc_simulation.csv		pc_simulation.csv
pc_simulation_10000.csv		pc_simulation_10000.csv
pc_simulation_rerun.csv		pc_simulation_rerun.csv
pc_simulation_rerun2.csv		pc_simulation_rerun2.csv
pc_simulation_rerun_timing.csv		pc_simulation_rerun_timing.csv
pc_simulation_upto1000.csv		pc_simulation_upto1000.csv
seed_chain_simulation.csv		seed_chain_simulation.csv
seed_regret.csv		seed_regret.csv
seed_regret.csv.tmp		seed_regret.csv.tmp
th_regret.csv		th_regret.csv
ucb_regret.csv		ucb_regret.csv

Repository files navigation

This code is for my final project for CS234.
I did a replication review of
Dimakopoulou, M., Osband, I., and Roy, B. V.
Scalable coordinated exploration in concurrent
reinforcement learning. CoRR, abs/1805.08948, 2018.
URL http://arxiv.org/abs/1805.08948.

Bipolar chain was done mostly in notebook:
Final Project CS238.ipynb
with MDP code in Agent.jl

Parallel Chain was implemented in RunPC.jl, with MDP code in ParallelChains.jl

Maximum reward path was implemented in RunMR.jl, with MDP code in MaximumRewardPath.jl

Final report is in project_update/final_report.tex (or .pdf, if you prefer)