Abstract
Projection-free optimization via different variants of the Frank-Wolfe (FW), a.k.a. Conditional Gradient method has become one of the cornerstones in optimization for machine learning since in many cases the linear minimization oracle is much cheaper to implement than projections and some sparsity needs to be preserved. In a number of applications, e.g. Poisson inverse problems or quantum state tomography, the loss is given by a self-concordant (SC) function having unbounded curvature, implying absence of theoretical guarantees for the existing FW methods. We use the theory of SC functions to provide a new adaptive step size for FW methods and prove global convergence rate O(1/k) after k iterations. If the problem admits a stronger local linear minimization oracle, we construct a novel FW method with linear convergence rate for SC functions.
Original language | English |
---|---|
Title of host publication | Proceedings of the 37th International Conference on Machine Learning |
Subtitle of host publication | International Conference on Machine Learning, 13-18 July 2020, Virtual |
Editors | Hal Daumé III, Aarti Singh |
Publisher | Proceedings of Machine Learning Research |
Pages | 2814-2824 |
Number of pages | 11 |
Volume | 119 |
Publication status | Published - 2019 |
Event | 25th Americas Conference on Information Systems of the Association-for-Information-Systems( AMCIS) - Cancun, Cancun, Mexico Duration: 15 Aug 2019 → 17 Aug 2019 |
Conference
Conference | 25th Americas Conference on Information Systems of the Association-for-Information-Systems( AMCIS) |
---|---|
Country/Territory | Mexico |
City | Cancun |
Period | 15/08/19 → 17/08/19 |
Keywords
- OPTIMIZATION
- CONVERGENCE
- COMPLEXITY