Fast Sparse Classification for Generalized Linear and Additive Models

TL;DR

This paper introduces fast, scalable methods for sparse generalized linear and additive models, significantly improving computational efficiency while maintaining interpretability and accuracy on large, complex datasets.

Contribution

The authors develop novel algorithms using surrogate cuts and priority queues for rapid feature screening in sparse models, including an alternative exponential loss for logistic regression.

Findings

Algorithms are 2 to 5 times faster than previous methods.

Models achieve accuracy comparable to black box models.

Techniques handle thousands of features and observations efficiently.

Abstract

We present fast classification techniques for sparse generalized linear and additive models. These techniques can handle thousands of features and thousands of observations in minutes, even in the presence of many highly correlated features. For fast sparse logistic regression, our computational speed-up over other best-subset search techniques owes to linear and quadratic surrogate cuts for the logistic loss that allow us to efficiently screen features for elimination, as well as use of a priority queue that favors a more uniform exploration of features. As an alternative to the logistic loss, we propose the exponential loss, which permits an analytical solution to the line search at each iteration. Our algorithms are generally 2 to 5 times faster than previous approaches. They produce interpretable models that have accuracy comparable to black box models on challenging datasets.