Approximation Theory and Machine Learning
Purdue University, September 29 - 30, 2018
Talks to take place in the Mathematical Sciences Building (MATH) 175 - Note this is a change in venue. Lunch and Poster/Reception will be in LAWSON COMMONS.
Machine learning has made a tremendous impact on the world through products such as automatic face and object recognition in computer vision, self-driving cars, automatic language translation, and a variety of other products. That said, much of the research on machine learning tends to focus on successful algorithms for specific machine learning tasks. The broader theoretical picture of when and why machine learning algorithms are successful tends to be discussed, instead, in a number of related domains and disciplines spanning applied mathematics, computer science theory, and statistics. This conference will highlight the importance of approximation theory as it is used in mathematics in existing and future machine learning and data science problems.
Invited Speakers
- Misha Belkin, Ohio State University
- An interpolation perspective on modern machine learning, abstract, video, slides (pdf)
- David Bindel, Cornell University
- Linear Algebra Support for Scalable Kernel Methods, abstract, video, slides (pdf)
- Paul Constantine, University of Colorado, Boulder
- Approximation Theoretic Advice for Supervised Learning, abstract, video, slides (pdf)
- Sanmi Koyejo, University of Illinois
- How effective is your classifier?, abstract, video, slides (pdf)
- Sven Leyffer, Argonne National Labs
- Optimization for Machine Learning, abstract, video, slides (pdf)
- Mauro Maggioni, Johns Hopkins University
- David Stewart, Professor of Mathematics, University of Iowa
- Approximation by Ridge functions with weights in a specified set, abstract, video, slides (pdf)
- Yusu Wang, Ohio State University
- Graph reconstruction via discrete Morse Theory, abstract, video, slides (pdf)
- Stefan Wild, Argonne National Lab
Organizers
- Greg Buzzard, Department of Mathematics, Purdue University
- David Gleich, Department of Computer Science, Purdue University
- Guang Lin, Department of Mathematics, Purdue University
Questions about registration can be addressed to Anna Hook (hook6@purdue.edu).
Conference funded by IMA conference grant, Purdue College of Science, Purdue Department of Mathematics and Purdue Department of Computer Science.