Maximizing Expected Gain in Supervised Discrete Bayesian Classification When Fusing Binary Valued Features
Abstract
In this paper, previously reported work is extended for fusing binary valued features. In general when mining discrete data to train supervised discrete Bayesian classifiers, it is often of interest to determine the best threshold setting for maximizing performance. In this work, we utilize a discrete Bayesian classification model, a gain function, to determine the best threshold setting for a given number of binary valued training data under each class. Results are demonstrated for simulated data by plotting the expected gain versus threshold settings for different numbers of training data. In general, it is shown that the expected gain reaches a maximum at a certain threshold. Further, this maximum point varies with the overall quantization of the data. Additional results are also shown for a different gain function on the decision variable, that are used to extend previously reported results.
Document Details
- Document Type
- Technical Report
- Publication Date
- Jul 01, 2009
- Accession Number
- ADA533682
Entities
People
- Peter K. Willet
- Robert S. Lynch Jr.
Organizations
- Naval Undersea Warfare Center