Stretching the Least Squares to Embed Loss Functions Tables

Abstract

The method of least squares is extended to accommodate a class of loss functions specified in the form of function tables. Each function table is embedded into the standard quadratic loss function so that the nonlinear least squares algorithms can be adopted for loss minimization. This is an alternative to a more conventional approach which interpolates the function tables and minimizes the resulting loss function by some generic optimization algorithm. An advantage of the alternative over the conventional approach is the wider availability of the least squares programs compared to the generic optimization programs, especially on resource-constrained devices. Examples are given for its application to multiplicative utility function maximization problems.