Efficient methods for large scale nonsmooth DC optimisation and applications

Project Title:

Efficient methods for large scale nonsmooth DC optimisation and applications

Supervisor(s):

A/Prof Adil Baghirov and Dr. Sona Taheri

Contact person and email address:

A/Prof Adil Baghirov, a.bagirov@federation.edu.au

A brief description of the project:

Many real-life problems can be formulated as unconstrained or constrained nonsmooth optimisation problems where the objective and/or constraint functions are represented as a difference of two convex (DC) functions. Such problems include clustering, supervised data classification, regression analysis and clusterwise regression problems in machine learning. Most of these problems contain a large number of variables and/or constraints. Furthermore, these functions are nonsmooth. This proposal aims to develop efficient methods for solving large scale nonsmooth DC optimisation problems and apply them to design real-time and accurate algorithms for solving large scale regression and data classification problems in machine learning.