In this paper, an approach for probability-based class prediction is presented. This approach is based on a combination of a newly proposed Histogram Probability (HP) method and any classification algorithm (in this paper results for combination with Extreme Learning Machines (ELM) and Support Vector Machines (SVM) are presented). Extreme Learning Machines is a method of training a single-hidden layer neural network. The paper contains detailed description and analysis of the HP method by the example of the Iris dataset. Eight datasets, four of which represent computer vision classification problem and are derived from Caltech-256 image database, are used to compare HP method with another probability-output classifier [11, 18].
CITATION STYLE
Gritsenko, A., Eirola, E., Schupp, D., Ratner, E., & Lendasse, A. (2017). Solve classification tasks with probabilities. Statistically-modeled outputs. In Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics) (Vol. 10334 LNCS, pp. 293–305). Springer Verlag. https://doi.org/10.1007/978-3-319-59650-1_25
Mendeley helps you to discover research relevant for your work.