A new smooth method based on rotated hyperbola for support vector machine in classification
Abstract
One smooth rotated hyperbola for support vector machine (RHSSVM) is proposed. Based on the approximation property of the hyperbola to its asymptotic lines, we build this model. Theoretical analysis shows that RHSSVM has the least error on approximating the plus function compared with three other typical models. Experimental results show that RHSSVM reduces the consuming time and efficiently deal with large scale and high dimensional programs.
- Publication:
-
Journal of Physics Conference Series
- Pub Date:
- September 2018
- DOI:
- 10.1088/1742-6596/1074/1/012129
- Bibcode:
- 2018JPhCS1074a2129W