Multiclass Classification Using Random Forest Classifier

Abstract

A multiclass classification using Random Forest Classifier is proposed in this paper. The Random forest classifier is commonly used for solving the multiclass classification tasks in machine learning. The Random forest classifier predicts a valid classification results with minimum training time when compared to other classifier algorithms. For pattern classification tasks on an image dataset are performed in order to evaluate the performance of the proposed classifier. First, we consider some sample datasets of flower species and birds. Second, we train the object class models by using datasets to develop the random forest classifier for image classification. Third, we test the object class models and find the accuracy .In this we use Haralick,Hu moments to extract global features to quantify an image. In the experiments we test our method using FLOWERS-17,CUB_200_2011 dataset and compare the results with other methods like logistic regression, k-nearest neighbour, support vector machine.