Panagiota Fatourou
Professor at the Department of Computer Science of the University of Crete, Greece
Abstract: Parallel and Distributed Data Series Processing
Processing of large collections of real-world data series is nowadays one of the most challenging and critical problems for a wide range of diverse application domains, including finance, seismology and other earth sciences, astrophysics, neuroscience, engineering, etc. Due to the unprecedented growth in size that data series collections experience nowadays, traditional, serial-execution data series indexing technologies are rendered inadequate. Thus, one of the most pressing issues in data series processing is achieving enhanced performance and high scalability. This tutorial focuses on major techniques for building distributed and concurrent data series indexing solutions, which are designed to inherently take advantage of modern hardware, in order to accelerate data series processing times for both on-disk and in-memory data. In particular, we will study a collection of data series indices that utilize the entire computational power of modern clusters (multiple nodes, multi-core and SIMD architecture of each node, as well as Graphics Processing Units (GPUs)) to tackle the performance and scalability goals in data series processing.