Abstract

Similarity measurement processes are a core part of most machine learning algorithms. Traditional approachesfocus on either taxonomic or thematic thinking. Psychological research suggests that a combination of bothis needed to model human-like similarity perception adequately. Such a combination is called a SimilarityDual Process Model (DPM). This paper describes how to construct DPMs as a linear combination of existingmeasures of similarity and distance. We use generalisation functions to convert distance into similarity. DPMsare similar to kernel functions. Thus, they can be integrated into any machine learning algorithm that useskernel functions.Clearly, not all DPMs that can be formulated work equally well. Therefore we test classificationperformance in a real-world task: the detection of pedestrians in images. We assume that DPMs are onlyviable if they yield better classifiers than their constituting parts. In our experiments, we found DPM kernelsthat matched the performance of conventional ones for our data set. Eventually, we provide a construction kitto build such kernels to encourage further experiments in other application domains of machine learning.

Reference

Eidenberger, H., Klauninger, B., & Unger, M. (2016). Machine Learning with Dual Process Models. In Proceedings of the 5th International Conference on Pattern Recognition Applications and Methods. 5th International Conference on Pattern Recognition Applications and Methods, Rom, EU. https://doi.org/10.5220/0005655901480153