xjiajiahao / projection-free-online-methods

Dependencies

1. Operating System: Linux
2. MATLAB R2019a or above

How to run

Online Covariance Matrix Estimation

% conduct experiment in the stochastic setting and plot figures
test_CME_stoch;
plot_CME_stoch;

% conduct experiment in the adversarial setting and plot figures
test_CME_adv;
plot_CME_adv;

Online Collaborative Filtering

Step 1: compile the ./Utils/proj_l1.cpp file from the terminal

mex -largeArrayDims ./Utils/proj_l1.cpp -lm -ldl

Step 2: change the current directory to ./Utils, and then execute the following script in MATLAB to generate the requried datasets

build_CF_datasets;

Step 3: change the current directory back to ./, and then execute the following scripts to conduct experiments

% conduct experiment on MovieLen 100K dataset and plot figures
test_CF_MovieLens100K;
plot_CF('MovieLens100K');

% conduct experiment on Jester 1 dataset and plot figures
test_CF_Jester1;
plot_CF('Jester1');

% conduct experiment on MovieLen 1M dataset and plot figures
test_CF_MovieLens1M;
plot_CF('MovieLens1M');

% conduct experiment on Jester 3 dataset and plot figures
test_CF_Jester3;
plot_CF('Jester3');

Online Binary Classification

Step 1: change the current directory to ./Utils, and then execute the following script in MATLAB to generate the requried dataset

build_eeg_dataset;

Step 2: change the current directory back to ./, and then execute the following scripts to conduct experiments

% online binary classification with smooth cost functions
test_SVM_smooth;
plot_SVM_smooth;

% online binary classification with l1 regularized cost functions
test_SVM_l1;
plot_SVM_l1;