This repo uses https://github.com/gkhayes/mlrose & some helper code from my first project https://github.com/jenngeorge/supervised-learning
setup
Ensure you have Python3 and conda installed.
With conda, create and activate the environment from environment.yml like conda env create -f environment.yml
Follow the prompt to activate the environment with conda activate ro_env
running optimization problem experiments
run the corresponding file in this directory in the ro_env environment like python3 run_one_max.py
a txt file containing info and performance metrics for each run will be written to results/{problem abbreviation}
plotting optimization problem experiments
I copied metrics from each experiment's txt file to make_opt_plots.py
If you wish to recreate the plots for your own experiment runs, you will need to do the same.
running neural network experiments
run the corresponding file in this directory in the ro_env environment
I recommend commenting out portions of __main__ for faster runtime
backprop: python3 run_nn_backprop.py
rhc: python3 run_nn_rhc.py
simulated annealing: python3 run_nn_sa.py
genetic algorithm: python3 run_nn_ga.py
a csv file containing the learning curve,
txt file containing stats and info, and a plot image
will be written to results/nn for each experiment run in each file.