ABSTRACT

In this paper, the performance of various machine learning methods on pattern classification and recognition tasks are proposed. The proposed method for evaluating performance will be based on the feature representation, feature selection and setting model parameters.
The nature of the data, the methods of feature extraction and feature representation are discussed. The results of the Machine Learning algorithms on the classification task are analysed. The performance of Machine Learning methods on classifying Tamil word patterns, i.e., classification of noun and verbs are analysed.
The software WEKA (data mining tool) is used for evaluating the performance. WEKA has several machine learning algorithms like Bayes, Trees, Lazy, Rule based classifiers.