Robust Meta-learning for Mixed Linear Regression with Small Batches

This project contains the code for the paper accepted at NeurIPS 2020 with the above title. The file RPCA.py contains an implementation of the algorithm and the simulations done in the paper. This project also contains the code of its preceeding ICML 2020 paper where we provide the base code for an implementation of our end-to-end meta learning algorithm for a mixture of linear regression in the file meta_learning_utils.py. As a interesting case, in the file meta_learning_sine.py we simulate this algorithm on a mixture of sine wave reconstruction problem which is a generalization of linear regression under orthonormal polynomial feturization.