Recently, the book titled "Tensor Computation for Data Analysis",authored by Associate Professor YipengLIU, doctoral students JianiLIUand ZhenLONG, along with ProfessorCe ZHUfrom the School of Information and Communication Engineering at UESTC, has been officially published by Springer. This two-year effort discusses a series of machine learning methods that extend tensor computation, providing a detailed exploration of the fundamentals of tensor computation. The book comprehensively and hierarchically introduces various applications of tensor computation methods in data analysis, serving as an introductory book for tensor computation in the field of data analysis.
Associate ProfessorYipengLIUhas been dedicated to theoretical, methodological, and applied work related to tensor data analysis. Grounded in tensor computation, the focus has been on tensor decomposition and low-rank optimization, extending to technologies such as signal recovery, image enhancement, spectrum sensing, collaborative filtering, tensor regression, lightweight neural networks, adversarial attacks, and anomaly detection. This approach effectively avoids the data structure loss caused by matrixization of high-dimensional arrays in traditional data processing and addresses the computational challenges of the "curse of dimensionality" in large-scale data processing. ProfessorLIUhas published over 80 international journal and conference papers, leading two projects supported by the National Natural Science Foundation of China and one international/regional research cooperation project in Sichuan Province. He is an IEEE Senior Member, editorial board member of IEEE Signal Processing Letters, and chief guest editor of Signal Processing: Image Communication (Elsevier).
Writing Monographsto Promote Tensor Computation
In May 2019, YipengLIU's tutorial presentation at ISCAS 2019 garnered attention and positive reviews from fellow scholars. Subsequently, he received an invitation from the Director of Springer Publishing Group to write a monograph on tensor data analysis.
During the research on tensor data analysis, the team realized the lack of an introductory book for senior undergraduates and graduate students in this direction. While there were some excellent review materials in computational mathematics and certain application topics, there was a gap in systematically summarizing the applications of tensor computation in the field of data analysis, especially incorporating the latest developments of the past decade. Moreover, existing materials often lacked systematic application examples.
Building on the team's nearly six years of expertise in tensor data analysis, they expanded their efforts to provide a comprehensive overview.While summarizing the latest research achievements, the book aims to lower the requirements for background knowledge and technical reserves, making it beneficial for every reader interested in tensors. Beyond serving as a professional reference, the team hopes the book becomes the first introductory book for data analysis using tensor computation. To achieve this, the structure and content of the book have been meticulously designed. The initial two foundational chapters help readers grasp the basics of tensor computation by extending fundamental concepts from linear algebra. Each introduction of a new optimization method starts with an explanation in matrix form before transitioning to the tensor computation system. Every technical application chapter is accompanied by numerical experiments with code, facilitating reader understanding and practical application.
