Efficient and generic high-dimensional structures are essential for building scalable
and interactive content-based search systems. At IMEDIA we are notably working on new methods for reducing the complexity of similarity search algorithms in metric spaces as well as other algorithms such as hyperplane queries or clustering.
Selected publications: Multi-probe LSH [ACM2008], Efficient boosting of local features [ACM2009], Hyperplane queries with M-tree [BDA2007], Large scale similarity search applied to video copy detection [ViCopt2009] [TMA2007],
