Note: A MATLAB implementation, which includes reproducibility of the figures in the paper can be found here.
This code is a Python implementation of our ICASSP 2025 paper by the name Fast Sparse Learning from Streaming Data with LASSO. The proposed method is very easy to implement and we provide an example script how to run. We remark that the proposed method converges only for uncorrelated features.
Run the juptyer notebook example_code.m - specify the system settings in the top cells for the desired synthetic data (e.g., noise, length, sparsity).
module "proposed_method" contains "online_lasso" - the fn where the proposed method is implemented.
util/generate_data.m - a fn that generates data.
util/bar_plot - a fn that makes bar plots to visualize feature seleciton.