A0351
Title: Two-step clustering: A new method in the sequential deep clustering approach
Authors: Claudia Rampichini - University of Rome La Sapienza (Italy) [presenting]
Abstract: The proposed method combines the use of two different clustering methods: fuzzy k-means and k-means. Membership degree values are used to identify units with an unclear assignment, and the crisp method is applied to this reduced dataset. A unit has an unclear assignment when membership degrees are close to each other. Knowing that one of the main problems of traditional clustering methods is related to the handling of high dimensional data, this proposal is combined with the use of a neural network according to the sequential deep clustering approach. Deep neural networks enhance clustering performance by reducing input data complexity, followed by clustering on the reduced dataset. An autoencoder neural network is utilized, and the two-step clustering method is applied to its results. As such, it is possible to obtain good results even with high-dimensional data such as images. Empirical studies demonstrate performance enhancements over individual k-means and fuzzy k-means methods, highlighting the effectiveness of neural networks in clustering tasks.
