Stochastic Optimization via Forward Slice

• Activity Number: 322
• Type: Contributed
• Date/Time: Tuesday, August 11, 2015 : 8:30 AM to 10:20 AM
• Sponsor: Section on Statistical Computing
• Author(s): Bob Salim* and Lurdes Inoue
• Companies: and University of Washington
• Keywords: Optimization ; Algorithms

Abstract

Optimization consists of maximizing or minimizing a real-valued objective function. In many problems, the objective function may not yield closed-form solutions. Over many decades, optimization methods, both deterministic and stochastic, have been developed to provide solutions to these problems. However, some common limitations of thesemethods are the sensitivity to the initial value and that often current methods only find a local (non-global) extremum. In this talk, we propose an alternative stochastic optimization method which we call "Forward Slice". The method's performance relative to available optimization methods, including conjugate gradient, Nelder-Mead, and simulated annealing, was assessed via simulations using multimodal objective functions. Based on the simulations, the forward slice method correctly selects the highest maximum more frequently than alternative methods.