Performance of Extreme Learning Machine

Classification is a critical aspect of machine learning with wide-ranging applications in various fields. In this article, the effectiveness of two popular classification algorithms is compared, the Extreme Learning Machine (ELM) and k-Nearest Neighbors (KNN) classifiers. ELM is a single hidden layer feedforward neural network that uses random weights for quick training and good accuracy. A nonparametric classification technique called KNN, on the other hand, gives class labels based on the training set’s k-Nearest Neighbors’ majoritarian vote. The comparison is conducted on a benchmark dataset to assess the effectiveness of these algorithms. The classifiers are assessed on five datasets, where the initial step involves pre-processing the data by normalizing the features and converting the class labels into numerical values. We then train and test the classifiers on the dataset and report the training and testing accuracy as well as the confusion matrix to evaluate their performance. Our findings suggest that ELM is a promising alternative to KNN for classification tasks especially when there are few training samples and high data dimensionality conditions. The results also highlight the importance of careful parameter tuning and model selection in achieving high performance in classification tasks.