Event: November 14, Math Colloquium: Difference-of-convex Learning: optimization with nonconvex sparsity functions.

Event Date: November 14
Location: 126 Clements Hall
Time: 3:30 PM

Miju Ahn Department of Engineering Management, Information, and Systems (EMIS), Lyle School of Engineering, SMU

Sparse learning involves building a robust and efficient model by incorporating sprasity functions in the optimization formulation to select significant variables to serve in the model. The first part of this talk emphasizes on a bi-criteria Lagrangian formulation. In the second part, we introduce a new system of constraints called affine sparsity constraints (ASCs) defined by a discrete sparsity function.