IMSc Webinar

An Insight to Fuzzy-Based Learning Algorithms For Clustering Microarray Datasets

Esha Kashyap

Pondicherry University

Microarray experiments monitor the expression of genes over a set of samples or experimental conditions. Clustering approaches focused on this genomic information can determine sub-population of patients with analogous prognostic characteristics. These datasets have an intricate high-dimensional data structure associated with noise and imprecise information. The aim of this research is to develop an efficient clustering algorithm for patient stratification. This presentation provides an insight to microarray data clustering. It further proposes and presents a robust fuzzy-based clustering algorithm, Geodesic Fuzzy Clustering with Local Information (GEOCLUS). The proposed approach embeds a microarray dataset on a Riemannian manifold of constant curvature while retaining trends and patterns. GEOCLUS is built on the Fuzzy C-Medoids algorithm and factors into the local non-linear interactions between the features of the dataset. GEOCLUS searches for treatment groups; it partitions cancer samples with distinct histopathological and molecular characteristics. High-dimensional microarray datasets are studied. The experimental studies documents the robustness of the proposed approach.


zoom.us/j/99448264327

Meeting ID: 994 4826 4327
Passcode: 280743