A Deep Conceptual Incremental learning Based High Dimensional Data Clustering model- A Deep Learning Approach

Abstract

                    Clustering is one of the important and vital tasks in the natural language processing which is becoming familiar in the data based application domains. Performance of machine learning clustering models depends on the quality of data or learning representation. Machine learning algorithm plays an important role in high dimensional data clustering. However the algorithms are suffering with low accuracy and distribution of the data points. Deep learning approaches are currently explored for solving these challenges and for better data representation for clustering. In this paper, a novel deep learning approach named as Deep Conceptual Incremental Clustering has been proposed for analysing unstructured and high dimensional data. It implies the autoencoder model.  It is effective in transform learning from high dimensional to low dimensional feature space to extract the concept specific features on the distribution of the data. It determines the salient features on ensuring the minimum reconstruction error. Architecture is composed of multiple layer to sparse representation and to eliminate the over fitting. Further all parameters are fine tuned with respect to certain criterion which considered as Loss function and encoder function. The encoder function is used to map the data points into latent representations. Finally it is helpful to find a better initialization of the parameters. Extensive experiments have been conducted on real datasets to compare proposed model with several state-of-the-art approaches. The experimental results show that Deep Conceptual Incremental clustering can achieve both effectiveness and good scalability on high dimensional data.