Supervised Isomap with Dissimilarity Measures in Embedding Learning

This paper proposes a supervised version of the Isomap algorithm by incorporating class label information into a dissimilarity matrix in a financial analysis setting and shows that the accuracy of the proposed approach is comparable to the state-of-the-art Support Vector Machines and Relevance Vector Machines despite the fewer dimensions used resulting from embedding learning.

Abstract

In this paper we propose a supervised version of the Isomap algorithm by incorporating class label information into a dissimilarity matrix in a financial analysis setting. On the credible assumption that corporates financial status lie on a low dimensional manifold, nonlinear dimensionality reduction based on manifold learning techniques has strong potential for bankruptcy analysis in financial applications. We apply the method to a real data set of distressed and healthy companies for proper geometric tunning of similarity cases. We show that the accuracy of the proposed approach is comparable to the state-of-the-art Support Vector Machines (SVM) and Relevance Vector Machines (RVM) despite the fewer dimensions used resulting from embedding learning.