GitHub - hjelp200121/02456ActiveLearning: Active Learning project for 02456

Branches Tags

Name		Name	Last commit message	Last commit date
Latest commit History 50 Commits
asy/cluster_margin		asy/cluster_margin
.gitignore		.gitignore
AL_intro.py		AL_intro.py
accuracies.py		accuracies.py
cluster_margin.py		cluster_margin.py
cluster_margin_optimize.py		cluster_margin_optimize.py
cluster_margin_parameters.db		cluster_margin_parameters.db
committee.py		committee.py
committee_hard_parameters.db		committee_hard_parameters.db
committee_hard_parameters2.db		committee_hard_parameters2.db
committee_optimize.py		committee_optimize.py
committee_soft_parameters.db		committee_soft_parameters.db
committee_soft_parameters2.db		committee_soft_parameters2.db
main.ipynb		main.ipynb
margin.py		margin.py
margin_optimize.py		margin_optimize.py
margin_parameters.db		margin_parameters.db
model.py		model.py
readme.txt		readme.txt
train_epochs.py		train_epochs.py
uniform_random.py		uniform_random.py

Repository files navigation

Running the code in the ipynb produces results on the MNIST dataset with num_runs of all the different methods used in the project and takes ~3*num_runs hours on the hpc.

The main file we have used is the accuracies.py with num_runs = 10 for all the data, which we have run on batch mode on the hpc.
If you want to really test the data we recommend running the accuracies.py file for each of the different methods and then plotting it using the same file.