Analysis and follow up of Blanchard et al. (2017), Machine Learning with Adversaries: Byzantine-Resilient Gradient Descent

This project revolves around Blanchard, El Mhamdi, Guerraoui & Steiner' founding paper, Machine Learning with Advesaries: Byzantine Tolerant Gradient Descent (2017). The first part, which is not treated in this notebook, consists in a thourough analysis of the article. In the second part—which is the subject of this notebook—we design and implementat a follow-up of the paper, based on Byzantine workers suspension.

Both parts have been presented to Prof. El Mahdi El Mhamdi, co-author during an oral defence.