This repository contains the codes for MOMPCA. A preprint of the corresponding paper, is available at,
D. Paul, S. Chakraborty, S. Das. Robust Principal Component Analysis: A Median of Means Approach. arXiv preprint arXiv:2011.06461. https://arxiv.org/abs/2011.06461.
The code is written in R programming language. The arguments for MoMPCA function is as follows:
INPUT:
- X: the n*p matrix, whose rows denote the data points.
- L: No. of partitions of the data.
- d: The number of principal components required.
- eta: The learning rate in the projected gradient descent.
- tmax: The maximum number of iterations.
- verbose: If 1 prints the iteration numbers. If 0, no printing. default=0.
OUTPUT:
- V: The p*d matrix, whose columns contain the first d principal components.
- mu: The componentwise median of the data.
- Y: The n*d matrix which is the low-dimensioanl representation of the data.