Approximate Manifold Regularization: Scalable Algorithm and Generalization Analysis
This repository provides the code used to run the experiments of the paper "Approximate Manifold Regularization: Scalable Algorithm and Generalization Analysis". The paper has been published in IJCAI-19. The paper applied Nystom and PCG to LapRLS, borrowing the idea from FALKON. The implementation also use tricks provided in the repository (https://github.com/LCSL/FALKON_paper).
The codes are implemented in MATLAB.
- ./datasets: All datasets are available in Libsvm Data.
- ./data: Store processed data including kernel matrix and graph Laplacian.
- ./result: Store final results used in the paper.
- ./core_functions: Implementation of compared algorithms.
- ./parameter_tune: Tune parameters.
- ./utils: Some utils including constructing kernel matrix and graph Laplacian, drawing curves and optimal parameters setting.
- Download data sets into ./datasets
- Run Exp1_*.m for experiment 1.
- Run Exp2_test_labeled_curve.m for experiment 2.
- Run Exp3_test_sample_curve.m for experiment.