Improved fuzzy sparse multi-class least squares support vector machine
What is it about?
The classification problem is a key area of research in machine learning. The Least Squares Support Vector Machine (LSSVM) is an important classifier that is commonly used to solve classification problems. this paper proposes an improved fuzzy sparse multi-class least squares support vector machine (IF-S-M-LSSVM) that maximises the synergy between the sparse algorithm and the fuzzy membership degree by comprehensively analysing the modelling characteristics of the multi-class LSSVM model and the distribution characteristics of the data. Firstly, the new model uses a simple and effective sparse algorithm based on the intra-class center distance, which can delete training sample points to different degrees by adjusting the sparse ratio. Secondly, to address the impact of sample noise on determining the optimal hyperplane, fuzzy membership degree based on sample density is added, allowing different samples to play different roles in determining the optimal hyperplane. The experimental results on UCI benchmark datasets demonstrate that IF-S-M-LSSVM outperforms several other multi-class support vector machines in terms of classification accuracy and training speed. This new method offers an efficient solution to data classification and discrimination problems without requiring complex theory and large workload. This idea can serve as a valuable reference for dealing with big data pattern recognition problems in the future.
Why is it important?
Its widespread use stems from its replacement of the inequality constraint in the Support Vector Machine (SVM) with the equality constraint, which transforms the convex quadratic programming (QP) problem of SVM into the solution of linear equations.