On Low Rank Recovery Problems Using Quadratic Envelope Regularization
(2024) In Master’s Theses in Mathematical Sciences MATM03 20232Mathematics (Faculty of Sciences)
Centre for Mathematical Sciences
- Abstract
- Many optimization models from practical problems are too bad to be worked by standard optimization techniques. Here the bad properties include nonconvexity and high discontinuity, such as the problem named low-rank recovery. The traditional approach, nuclear norm minimization, can solve the low-rank recovery but may contain bias. Algorithms like the FBS and the ADMM are highly effective in convex optimization but may encounter difficulties or fail when applied to nonconvex problems. Inspired by the lower semi-continuous convex envelope, we develop an unbiased approach with the quadratic envelope as a regularizer for the low-rank recovery. By adjusting the parameter value of the regularizer for the different problems, it can be a powerful... (More)
- Many optimization models from practical problems are too bad to be worked by standard optimization techniques. Here the bad properties include nonconvexity and high discontinuity, such as the problem named low-rank recovery. The traditional approach, nuclear norm minimization, can solve the low-rank recovery but may contain bias. Algorithms like the FBS and the ADMM are highly effective in convex optimization but may encounter difficulties or fail when applied to nonconvex problems. Inspired by the lower semi-continuous convex envelope, we develop an unbiased approach with the quadratic envelope as a regularizer for the low-rank recovery. By adjusting the parameter value of the regularizer for the different problems, it can be a powerful tool and offer an increased probability for algorithms to converge to the global minima. (Less)
Please use this url to cite or link to this publication:
http://lup.lub.lu.se/student-papers/record/9144464
- author
- Liu, Qian LU
- supervisor
- organization
- course
- MATM03 20232
- year
- 2024
- type
- H2 - Master's Degree (Two Years)
- subject
- keywords
- Optimization, Non-convex Optimization, Low-rank Recovery, Convex Envelope, Quadratic Envelope
- publication/series
- Master’s Theses in Mathematical Sciences
- report number
- LUNFMA-3145-2024
- ISSN
- 1404-6342
- other publication id
- 2024:E2
- language
- English
- id
- 9144464
- date added to LUP
- 2025-07-01 08:28:16
- date last changed
- 2025-07-01 08:28:16
@misc{9144464, abstract = {{Many optimization models from practical problems are too bad to be worked by standard optimization techniques. Here the bad properties include nonconvexity and high discontinuity, such as the problem named low-rank recovery. The traditional approach, nuclear norm minimization, can solve the low-rank recovery but may contain bias. Algorithms like the FBS and the ADMM are highly effective in convex optimization but may encounter difficulties or fail when applied to nonconvex problems. Inspired by the lower semi-continuous convex envelope, we develop an unbiased approach with the quadratic envelope as a regularizer for the low-rank recovery. By adjusting the parameter value of the regularizer for the different problems, it can be a powerful tool and offer an increased probability for algorithms to converge to the global minima.}}, author = {{Liu, Qian}}, issn = {{1404-6342}}, language = {{eng}}, note = {{Student Paper}}, series = {{Master’s Theses in Mathematical Sciences}}, title = {{On Low Rank Recovery Problems Using Quadratic Envelope Regularization}}, year = {{2024}}, }