Support Vector Machines for Longitudinal Analysis
Kristiaan Pelckmans, Hong-Li Zeng · 2015
This paper considers the problem of learning a classifier from observed longitudinal data. Here, each data-point takes the form of a single time-series. Assuming that each such series comes with a binary label, the problem of learning to score this label of a new time-series is considered. Hereto, the notion of margin underlying the classical Support Vector Machine (SVM) is extended to LSVM for functional data. LSVM is also a convex optimization problem and its dual form is derived. Empirical results for specified cases with significance tests indicate the efficacy of the LSVM for analyzing such longitudinal data.