An Integration of Extreme Learning Machine for Classification of Big Data

Classification is an important task in data mining field. As the time of big data is coming, the traditional methods of classification cannot satisfy the requirements of real-time processing and storage for big data. This study firstly applies a machine learning technique called extreme learning machine (ELM) to classify for big data. Performances of ELM for big data are evaluated by using big data. The experimental results show that classification method based on ELM outperforms other methods based on BP neural networks and support vector machine (SVM). Secondly, the possibility of paralleling ELM based on MapReduce is analyzed and a paralleling ELM is designed.