User-Defined Tensor Data Analysis

FasTensor is a parallel data programming model developed for big data applications, offering a user-friendly interface for expressing data analysis operations from various domains. It simplifies data management tasks and achieves significant speedup over Spark and other peer systems, making it an ideal tool for scientists and data engineers.

Format: Paperback / softback
Length: 101 pages
Publication date: 30 September 2021
Publisher: Springer Nature Switzerland AG

FasTensor is a potent parallel data programming model designed specifically for big data applications. This comprehensive book offers a user-friendly guide for installing and utilizing FasTensor, empowering users to effortlessly express a wide range of data analysis operations, originating from neural networks, scientific computing, or traditional database management systems (DBMS). By eliminating the underlying and time-consuming tasks of data management, such as partitioning, communication, and parallel execution, FasTensor liberates users from the complexities involved. This SpringerBrief provides a comprehensive overview of the state-of-the-art in parallel data programming models, highlighting the motivation behind the development of FasTensor. It showcases the FasTensor application programming interface (API) through numerous examples and two real-world use cases from cutting-edge scientific applications. FasTensor demonstrates remarkable speedup capabilities, outperforming Spark and other peer systems in executing large-scale data analysis operations. It simplifies the programming process for data analysis tasks at a massive scale on supercomputers, making it highly productive and efficient. A complete reference of FasTensor includes its theoretical foundations, C++ implementation, and practical applications in domains such as physical and geosciences, where scientists analyze extensive datasets. Data engineers, software developers, and data scientists who utilize Spark or TensorFlow for data analysis, including training deep neural networks, will also find this SpringerBrief invaluable as a valuable reference tool.

Weight: 191g
Dimension: 235 x 155 (mm)
ISBN-13: 9783030707491
Edition number: 1st ed. 2021