Ημερολόγιο Εκδηλώσεων

Περιγραφή:
Ομιλητής: Βασίλειος Σαμολαδάς, Σχολή ΗΜΜΥ | http://www.softnet.tuc.gr/~vsam/

Σύνοψη Ομιλίας:
Modern distributed information systems often require monitoring complex continuous queries on high-volume, rapid distributed data streams. As data volumes increase, centralizing the data in order to process it can become practically infeasible. In such settings, approximate query monitoring offers a practically viable strategy by summarizing the distributed streams at (or, near) the source, and only paying the cost of transmitting succinct summaries, such as histograms, sketches, or more generally, high-dimensional vectors.

This talk presents research done in recent years on this problem, on algorithms for distributed query monitoring that have a strong geometric flavor. The discussion touches several other areas of computer science, such as data analytics, optimization and the interplay between theory and practice.