Magnet seminars are usually held in room B21 on Thursdays, 11am. Check below for upcoming seminars and potential changes of schedule/location. You may also import the Magnet seminars public feed into your favorite calendar app. For further information, contact Aurélien or Michaël.
Thu, October 8, 2020
Where? Virtual (see Zoom link)
Predictive analysis of networked data is a fast-growing research area whose application domains include document networks, online social networks, and biological networks. Inference and clustering on large-scale networked data are especially interesting when, given a limited budget of queries, we aim to obtain the similarities of the most informative node pairs, and accurately predict the remaining ones. Our work on these problems consists of two main research lines.
We investigate active learning by pairwise similarity over the leaves of input trees originating from hierarchical clustering procedures. We provide a full characterization of the number of queries needed to achieve the exact reconstruction of the tree cut. Our algorithms come with theoretical guarantees and, more importantly, lend themselves to linear-time implementations in the relevant parameters of the problem.
We study correlation clustering as an active learning problem: each similarity score can be learned by making a query, and the goal is to minimize both the disagreements and the total number of queries. We describe a simple and fast active learning algorithm, which provably achieves an almost optimal trade-off between queries and clustering error, and we test it on different datasets.
Zoom link: univ-lille-fr.zoom.us/j/95660806602
Thursday, October 8, 2020 - 11:00 to 12:00
Virtual (see Zoom link)