engleski

Support Vector Machine Optimized by Fireworks Algorithm for Handwritten Digit Recognition

Handwritten digit recognition is an important subarea inthe object recognition research area. Support vector machines representa very successful recent binary classifier. Basic support vector machineshave to be improved in order to deal with real-world problems. The intro-duction of soft margin for outliers and misclassified samples as well askernel function for non linearly separably data leads to the hard optimiza-tion problem of selecting parameters for these two modifications. Gridsearch which is often used is rather inefficient. In this paper we proposethe use of one of the latest swarm intelligence algorithms, the fireworksalgorithm, for the support vector machine parameters tuning. We testedour approach on standard MNIST base of handwritten images and withselected set of simple features we obtained better results compared toother approaches from literature. (PDF) Support Vector Machine Optimized by Fireworks Algorithm for Handwritten Digit Recognition. Available from: https://www.researchgate.net/publication/338635 325_Support_Vector_Machine_Optimized_by_Firewor ks_Algorithm_for_Handwritten_Digit_Recognition [accessed Nov 09 2020].

Handwritten digit recognition ; Machine learning ; Support vector machine ; Optimization ; Swarm intelligence ; Fireworks algorith

nije evidentirano