We introduce Universum learning for multiclass problems and propose a novel formulation for multiclass universum SVM (MU-SVM). We also propose an analytic span bound for model selection with almost 2-4x faster computation times than standard resampling techniques. We empirically demonstrate the efficacy of the proposed MUSVM formulation on several real world datasets achieving > 20% improvement in test accuracies compared to multi-class SVM.
@article{arxiv.1808.08111,
title = {Multiclass Universum SVM},
author = {Sauptik Dhar and Vladimir Cherkassky and Mohak Shah},
journal= {arXiv preprint arXiv:1808.08111},
year = {2018}
}
Comments
33 pages. arXiv admin note: text overlap with arXiv:1609.09162