Exploratory data analysis requires fundamental techniques to understand, describe and eventually extract value from large amounts of data. One of such techniques is data clustering. Modern applications of clustering require the processing of large volumes of data and pose computational challenges to state-of-the-art approaches. At EURECOM, we tackle these issues by developing novel approaches to scale clustering methods without compromising in clustering accuracy. We do so by combining ideas form graph theory and linear algebra with our expertise in distributed systems. 

Here is a list of research topics in this area: 

  • Scalable clustering algorithms

  • Large-scale spectral clustering

  • Online clustering algorithms


  • Y. Han and M. Filippone. Mini-batch spectral clustering, 2016. arXiv:1607.02024. [ bib | pdf | http ]

  • M. Filippone, F. Camastra, F. Masulli, and S. Rovetta. A survey of kernel and spectral methods for clustering. Pattern Recognition, 41(1):176-190, January 2008. [ bib | pdf | http ]


Syndicate content

Data Science