Efficient Optimization of Multi-class Support Vector Machines with MSVMpack

Abstract

In the field of machine learning, multi-class support vector machines (M-SVMs) are state-of-the-art classifiers with training algorithms that amount to convex quadratic programs. However, solving these quadratic programs in practice is a complex task that typically cannot be assigned to a general purpose solver. The paper describes the main features of an efficient solver for M-SVMs, as implemented in the MSVMpack software. The latest additions to this software are also highlighted and a few numerical experiments are presented to assess its efficiency.

Machine and quantum learning for diamond-based quantum applications

“…However, one should be aware that SVC can require long training times if the datasets are large. This is because SVC contains processes that are computationally expensive and include matrix-related operations for kernel computations, optimization and gradient updating tasks [197][198][199]. It also tends to perform poorly if the datasets have a low signal-to-noise ratio that results in the overlapping of the target classes.…”