Code for the paper "Active Learning with Weak Supervision for Gaussian Processes". Part of the code was inspired by the github repository of the paper "Yarin Gal, Riashat Islam, Zoubin Ghahramani. "Deep Bayesian Active Learning". ICML 2017. https://arxiv.org/pdf/1703.02910.pdf".


### To reproduce the experiments in "Active Learning with Weak Supervision for Gaussian Processes":

Sine curve experiments, q=0.2: python -m experiment.sine_regression.sine_regression_experiment_fs_3_cost_function --nb_experiments 15 --nb_acquisitions 40 --nb_queries 1 --q 0.2 --fs 3.0  --save_dir "experiment/sine_regression/res/" --fit --interpolate

Sine curve experiments, q=2.0: python -m experiment.sine_regression.sine_regression_experiment_fs_3_cost_function --nb_experiments 15 --nb_acquisitions 40 --nb_queries 1 --q 2.0 --fs 3.0 --save_dir "experiment/sine_regression/res/" --fit 

Sine curve experiments, non-uniform data pool, w=3: python -m experiment.sine_regression.sine_regression_experiment_effect_data_non_uniform_input_ind --nb_experiments 15 --nb_acquisitions 40 --nb_queries 1 --q 1.0 --fs 3.0 --save_dir "experiment/sine_regression/res/" --fit

Sine curve experiments, non-uniform data pool, w=7: python -m experiment.sine_regression.sine_regression_experiment_effect_data_non_uniform_input_ind --nb_experiments 15 --nb_acquisitions 40 --nb_queries 1 --q 1.0 --fs 7.0 --save_dir "experiment/sine_regression/res/" --fit

UCI data sets, concrete compressive strength (data needs to be downloaded): python -m experiment.uci.concrete_experiment --nb_experiments 15 --nb_acquisitions 80 --nb_queries 1 --save_dir "experiment/uci/res/" --data_dir "experiment/uci/data/" --fit --fit_hyp --interpolate

UCI data sets experiment, superconductivity (data needs to be downloaded): python -m experiment.uci.superconductivity_experiment --nb_experiments 15 --nb_acquisitions 80 --nb_queries 1 --save_dir "experiment/uci/res/" --data_dir "experiment/uci/data/" --fit --fit_hyp

Toy class experiments, version 1: python -m experiment.toy_class.toy_class_experiment --nb_experiments 15 --nb_acquisitions 25 --nb_queries 1 --save_dir "experiment/toy_class/res/" --version v1 --fit

Toy class experiments, version 2: python -m experiment.toy_class.toy_class_experiment --nb_experiments 15 --nb_acquisitions 25 --nb_queries 1 --save_dir "experiment/toy_class/res/" --version v2 --fit

Toy class experiments, version 3: python -m experiment.toy_class.toy_class_experiment --nb_experiments 15 --nb_acquisitions 25 --nb_queries 1 --save_dir "experiment/toy_class/res/" --version v3 --fit