- Link to the paper:
- https://ieeexplore.ieee.org/document/8258617
- Please cite:
- Mousavi, Seyed Muhammad Hossein, S. Younes MiriNezhad, and Atiye Mirmoini. "A new support vector finder method, based on triangular calculations and K-means clustering." 2017 9th International Conference on Information and Knowledge Technology (IKT). IEEE, 2017.
A New Support Vector Finder Method, Based on Triangular Calculations and K-means Clustering
Finding and determining best Support vector samples, play and important role in the accuracy and efficiency of classification process. Many support vector methods have been proposed, which each one has its pros and cons. In this paper, a new support vector finder method, based on triangle, has been presented, which finds support vectors based on triangular calculations, like calculating triangle angles, area and defining threshold for them. According to those thresholds, Support vectors for each class will be defined. At the end of the whole process, K-means clustering method takes place on the remaining samples. Note that K-means could happen before the main process. After finding support vectors, the result will be classified by classification algorithms like SVM, Least Squares and Linear discriminant analysis algorithms, in the binary mode, and the acquired results will be compared with original data. The acquired results are satisfactory, precise and comparable with the best support vector finder methods.
- DOI:
- https://ieeexplore.ieee.org/document/8258617
- Please cite:
- Mousavi, Seyed Muhammad Hossein, S. Younes MiriNezhad, and Atiye Mirmoini. "A new support vector finder method, based on triangular calculations and K-means clustering." 2017 9th International Conference on Information and Knowledge Technology (IKT). IEEE, 2017.