Cis400/401 Final Project Report

The rise of ‘big data’ and large-scale machine learning has created an increasing need for distributed optimization.dual descent and the alternating Most of the current literature has focused on coordinate descent, a prominent distributed optimization technique, due to its simplicity and effectiveness. We focus on implementing two other optimization techniques distributed dual descent and the alternating direction method of multipliers (ADMM), and implement distributed versions of both for quadratic functions, linear class SVMs, and logistic regression. Since dual descent and ADMM are algorithms that use multiple nodes and communicate between them, they lend themselves to some networking techniques. In particular, we utilized the User Datagram Protocol (UDP) to facilitate this nodal and communication structure, a novel use of UDP in this area of research.Ultimately, we provide a package that easily allows for optimization problems with Dual Descent or ADMM over the provided families of functions, and hope that it can be extended with other functions and/or optimization techniques in the future.