Prophet-Demon Adversarial Learning

Exploring final size predictions on contact networks using adversarial learning.

Implementation

Currently only the Prophet model is implemented.

To train the Prophet on Erdos-Renyi random graphs use: train_prophet.py

To train the Prophet on Watts-Strogatz small world networks use: train_prophet_watts.py

EpiNetSim.py simulates epidemics on contact networks using the Gillespie stochastic simulation algorithm.

Name		Name	Last commit message	Last commit date
Latest commit History 11 Commits
.spyproject/config		.spyproject/config
__pycache__		__pycache__
prophet_async_trained.tf		prophet_async_trained.tf
results		results
viz		viz
.DS_Store		.DS_Store
.gitignore		.gitignore
EpiNetApprox.py		EpiNetApprox.py
EpiNetSim.py		EpiNetSim.py
EpiTensor.py		EpiTensor.py
GumbelSoftMax.py		GumbelSoftMax.py
README.md		README.md
demon_net_reconstruction_stats_binary.csv		demon_net_reconstruction_stats_binary.csv
environment.yml		environment.yml
final_size_dist_vs_poibin-approx.png		final_size_dist_vs_poibin-approx.png
loss_by_episode.png		loss_by_episode.png
poibin.py		poibin.py
reconst_loss_by_episode.png		reconst_loss_by_episode.png
test_ER_finalsize_dist.py		test_ER_finalsize_dist.py
test_networkx.py		test_networkx.py
test_poibin.py		test_poibin.py
test_relaxed_bernoulli.py		test_relaxed_bernoulli.py
test_watts-strogatz_finalsize_dist.py		test_watts-strogatz_finalsize_dist.py
train_demon_VAE.py		train_demon_VAE.py
train_prophet.py		train_prophet.py
train_prophet_demon_async.py		train_prophet_demon_async.py
train_prophet_demon_pba.py		train_prophet_demon_pba.py
train_prophet_watts.py		train_prophet_watts.py
utils.py		utils.py