The Neutral Point Method for Kernel-Based Combination of Disjoint Training Data in Multi-modal Pattern Recognition

Abstract

Abstract. Multiple modalities present potential difficulties for kernel-based pattern recognition in consequence of the lack of inter-modal kernel measures. This is particularly apparent when training sets for the differing modalities are disjoint. Thus, while it is always possible to consider the problem at the classifier fusion level, it is conceptually preferable to approach the matter from a kernel-based perspective. By interpreting the aggregate of disjoint training sets as an entire data set with missing inter-modality measurements to be filled in by appropriately chosen substitutes, we arrive at a novel kernel-based technique, the neutral-point method. On further theoretical analysis, it transpires that the method is, in structural terms, a kernel-based analog of the well-known sum rule combination scheme. We therefore expect the method to exhibit similar error-canceling behavior, and thus constitute a robust and conservative strategy for the treatment of kernel-based multi-modal data.

The Relationship Between Kernel and Classifier Fusion in Kernel-Based Multi-Modal Pattern Recognition: An Experimental Study

“…If a new object maps into a point strictly at the discriminant hyperplane (3), it cannot be attributed to any one of the two classes. All these points are called in [8]. neutral points ( )…”

“…The aim of this paper is an experimental study of the neutral point method as a methodological bridge between kernel fusion and classifier fusion in multi-modal supervised two-class pattern recognition . ( ) 1 y ω = ± In Sections 2 and 3, we briefly outline the main principles of combining kernels and kernel-based modalityspecific classifiers, considered in our recent publication [8]. Then, in Sections 4 and 5, we discuss the results of experimental comparison of these mutually complementary principles as applied to simulated dependent and independent modalities.…”

“…In our recent paper [8], we theoretically studied the relationship between kernel fusion and classifier fusion as 1-4244-0973-X/07/$25.00 ©2007 IEEE constituting two distinct approaches to multimodal pattern recognition at, respectively, the sensor and classifier levels. To achieve this in fully general terms, we specified a novel "neutral point" method for kernel-based combination of disjoint multi-modal training data.…”

A Kernel-Based Framework for Medical Big-Data Analytics

“…However, methods developed by the author (the 'neutral point substitution' method [17][18][19]) render this tractable. (Neutral point substitution involves the unbiased missing value substitution of a placeholder mathematical object in multi-modal SVM combination).…”

A Modified Neutral Point Method for Kernel-Based Fusion of Pattern-Recognition Modalities with Incomplete Data Sets

“…For the latter case, it was proposed in [8] to adopt the neutral point substitution (NPS) method originally developed in [7] as a means of kernel-based combining of disjoint multi-modal training data, i.e., when only one feature is known for each object. Important advantages of the NPS method are that it is implicitly incorporated into the SVM training framework and it is free from the necessity of inventing any heuristic surrogates for replacing the missing data.…”