Chidchanok Lursinsap, Chulalongkorn University, Thailand Title: Fast supervised learning using versatile RBF

Abstract: There are several machine learning applications concern the huge amount of timely data.  Currently, learning new data using a already trained supervised neural network must involve the previously trained data.  The learning time obviously depends upon the amount of data and the data storage can be overflowed in a very short time.  In this study, a very fast 1-pass-throw-away learning algorithm based on a hyper-ellipsoidal function that can be translated and rotated to cover the data set during learning process is introduced.  The function depends upon the distribution of the data set.  In addition, we present versatile elliptic basis function (VEBF) neural network with one hidden layer.  The hidden layer is adaptively divided into sub-hidden layers according to the number of classes of the training data set.  Each sub-hidden layer can be scaled by incrementing a new node to learn new samples during training process.  The learning time is O(n), where n is the number of data.  The network can independently learn any new incoming datum without involving the previously learned data.  There is no need to store all the data in order to mix with the new incoming data during the learning process.

Bio Sketch: Chidchanok Lursinsap received his B.Eng. (honors) from Chulalongkorn University, Thailand in 1978, M.S. and Ph.D. from University of Illinois at Urbana-Champaign, USA in 1982 and 1986, respectively.  He was a lecturer at Department of Computer Engineering, Chulalongkorn University in 1979.  In 1986, he was a visiting assistant professor at Department of Computer Science, University of Illinois at Urbana-Champaign.  From 1987 to 1996, he worked at The Center for Advanced Computer Studies, University of Louisiana at Lafayette, USA as an assistant and associate professor.  After that, he came back to Thailand to establish a Ph.D. program in computer science at Chulalongkorn University and became a full professor.  His major research interests include neural learning and its applications to other science and engineering areas.