A hierarchical multiclass support vector machine incorporated with holistic triple learning units

Abstract

This paper proposes a new hierarchical learning structure, namely the holistic triple learning (HTL), for extending the binary support vector machine (SVM) to multi-classification problems. For an N-class problem, a HTL constructs a decision tree up to a depth of dN=3e þ 1. A leaf node of the decision tree is allowed to be placed with a holistic triple learning unit whose generalisation abilities are assessed and approved. Meanwhile, the remaining nodes in the decision tree each accommodate a standard binary SVM classifier. The holistic triple classifier is a regression model trained on three classes, whose training algorithm is originated from a recently proposed implementation technique, namely the least-squares support vector machine (LS-SVM). A major novelty with the holistic triple classifier is the reduced number of support vectors in the solution. For the resultant HTL-SVM, an upper bound of the generalisation error can be obtained. The time complexity of training the HTL-SVM is analysed, and is shown to be comparable to that of training the oneversus-one (1-vs.-1) SVM, particularly on small-scale datasets. Empirical studies show that the proposed HTL-SVM achieves competitive classification accuracy with a reduced number of support vectors compared to the popular 1-vs-1 alternative.

Human continuous activity recognition based on energy-efficient schemes considering cloud security technology

“…What is more, the classification time is shorter. Inspired by the classification structure of partial binary tree, Xia et al , combined the classification structure of decision tree and regression technique to solve the problem of multi‐class classification. This paper will adopt the kind of classification structure of decision tree and regression technique.…”

“…Holistic triple learning classification node can divide the test samples into three sub‐categories, and the implementation technique for the regression model is benefited from the training algorithm of the least‐squares SVM (LS‐SVM), which is assured to have a sparse solution . As the implementation technique of an HTL classification node is described in , there is no need to describe here in detail. Except for HTL classification node, the rest of the nodes are the standard binary SVM classifier.…”

“…At present, many researches have proposed many methods to reduce energy consumption for activity recognition system, such as through reducing the sample rate of data acquisition, avoiding the extraction of frequency domain features, and adjusting the slide window for data pre‐processing , as well as by improving the recognition algorithm to reduce the time complexity of the algorithm ; while reduces the working hours of the activity recognition system by automatically adjusting the data acquisition, as well as the activation period and dormant period of recognition algorithm, so as to reduce energy consumption.…”