A scaling-up machine learning algorithm

With the rapid advancement of information technology, flood of digital data collected by business, government, and scientific applications need analyzing, digesting, and understanding. Scalability has become a necessity for data mining algorithms to process large data more effectively and extract insightful information from large data. In this paper a scaling up neural network learning algorithm is presented, which partitions a large data set into subsets, applies learning algorithm on each subset concurrently and then integrates the learned results. We proved that the scaling up neural network is equivalent to a neural network which adds a penalty term to the error function for controlling the bias and variance. The algorithm was evaluated using the large dataset from UCI repository.