|Numerical methods for machine learning|
|Time：||Wed/Fri 13:00-14:50, 2016-06-29 ~ 2016-07-29|
|Instructor：||Da Kuang [UCLA]|
|Place：||Conference Room 3, Floor 2, Jin Chun Yuan West Building|
The time will be changed to Wed/Fri 10:00-12:00 from July 8. The place will be conference room 4 on Wednesday and conference room 3 on Friday.
This course introduces selected machine learning methods that are largely based on numerical linear algebra. We will particularly focus on unsupervised learning methods in classical literature and recent research. Emphasis will also be given to practical implementation issues. Students will implement existing algorithms and write blog articles about their findings as part of the assessment, and work on projects with the potential to become research papers.
The lectures will be designed for entry-level graduate students in applied mathematics and computer science. Note that numerical optimization techniques, an important component in machine learning, will not be covered.
Students can use programming languages of their choice (e.g. Python, Matlab, C++). Students are responsible for learning the programming languages by themselves. Those who are not familiar with a language for scientific computing should learn the language before the lecture series starts.
Trefethen and Bau, Numerical linear algebra, SIAM 1997.
Demmel, Applied numerical linear algebra, SIAM 1997.
Trevor Hastie, Robert Tibshirani, and Jerome Friedman, Elements of statistical learning. (free online)
Leskovec, Rajaraman, and Ullman, Mining massive datasets. (free online)
Bengio, Learning deep architectures for AI, FTML 2009. (free online)