Improved Kernel Density Estimation Self-organizing Incremental Neural Network to Perform Big Data Analysis

Plenty of data are generated continuously due to the progress in the field of network technology. Additionally, some data contain substantial noise, while other data vary their properties in according to various real time scenarios. Owing to these factors, analyzing big data is difficult. To address these problems, an adaptive kernel density estimation self-organizing neural network (AKDESOINN) has been proposed. This approach is based on the kernel density estimation self-organizing incremental neural network (KDESOINN), which is an extension of the self-organizing incremental neural network (SOINN). An SOINN can study the distribution using the input data online, while KDESOINN can estimate the probability density function based on this information. The AKDESOINN can adapt itself to the changing data properties by estimating the probability density function. Further, the experimental results depict that AKDESOINN succeeds in maintaining the performance of KDESOINN, while depicting an ability to adapt to the changing data.