Linear Algebra for Data Science

Professors Kyunghyun Cho and Wanmo Kang have developed a new textbook on linear algebra aimed at enhancing the teaching of the subject in the context of data science and artificial intelligence. They emphasize the importance of practical concepts such as projection and singular value decomposition (SVD), which are often overlooked in traditional courses that prioritize invertibility and mathematical convenience. The textbook is structured to introduce these practical concepts early on, ensuring that students can grasp useful results and algorithms without sacrificing mathematical rigor.

The table of contents reflects a non-traditional order, placing SVD before determinants and eigendecomposition, which aligns with an optimization/variational approach to problem-solving. This method allows for a deeper understanding of linear algebra concepts through real-world applications rather than solely through algebraic derivations. Additionally, the textbook includes appendices covering related topics such as convexity, covariance matrices, and spectral decomposition, which are essential for understanding linear algebra results.

The textbook has been utilized in a course at KAIST, receiving positive feedback that has influenced its organization. The authors invite feedback and comments from readers to further improve the material. The textbook is available in both English and a mixed Korean-English version to accommodate a wider range of students.