K-means based probabilistic neural network (KPNN) for designing physical machine – classifier

Abstract

Cloud Computing necessitates virtual machines that can deploy users to a machine in a sub-optimal fashion for effective and maximum utilization of resources conserving overall energy within the specified duration. PNN is an effective probabilistic classifier which has been applied for a wide variety of computer application problems. However, for big data applications, we need either pre-processing algorithms for efficient classification with lower computing time or Mathematical tracking operators to speed up a parametric approach. This paper focuses to combine the traditional K means algorithm and a PNN to process the data obtained from Google cluster to classify them into pre-specified groups so as to implement PM classifier design to monitor the Cloud usage pattern. It is found after validation of KPNN with different data sets that KPNN works better than PNN in terms of accuracy even when the number of classes increases and turns out to be a computationally attractive tool.