Structured Learning via Logistic Regression

2014/07/03 by Justin Domke, Domke, Justin · 2 citations
Computer Science · Mathematics · #Face and Expression Recognition #Machine Learning and Algorithms #Neural Networks and Applications #cs.LG #stat.ML

paper · pdf · doi:10.48550/arxiv.1407.0754

Advances in Neural Information Processing Systems 2013

arxiv created 2014/07/03 · arxiv updated 2014/07/04

Abstract

A successful approach to structured learning is to write the learning objective as a joint function of linear parameters and inference messages, and iterate between updates to each. This paper observes that if the inference problem is "smoothed" through the addition of entropy terms, for fixed messages, the learning objective reduces to a traditional (non-structured) logistic regression problem with respect to parameters. In these logistic regression problems, each training example has a bias term determined by the current set of messages. Based on this insight, the structured energy function can be extended from linear factors to any function class where an "oracle" exists to minimize a logistic loss.

Cited by

• Heatmap Regression without Soft-Argmax for Facial Landmark Detection
• Integrated inference and learning of neural factors in structural support vector machines

Related

• Long-tail learning via logit adjustment
• An Entropy-Based Model for Hierarchical Learning
• Structured Prediction in Online Learning
• Learning Approximate Inference Networks for Structured Prediction
• Structured Prediction by Conditional Risk Minimization