Output details
13 - Electrical and Electronic Engineering, Metallurgy and Materials
Imperial College London : A - Electrical and Electronic engineering
Article title
A Geometric Approach to Low-Rank Matrix Completion
Type
D - Journal article
Title of journal
IEEE Transactions on Information Theory
Article number
-
Volume number
58
Issue number
1
First page of article
237
ISSN of journal
0018-9448
Year of publication
2012
URL
-
Number of additional authors
2
Additional information
The low-rank matrix completion problem has wide applications in online recommendation and machine learning. The common approaches to solve this problem include the convex relaxation and greedy algorithms, both of which require the incoherence conditions. The major novelty of this paper is the proposed geometric approach that for the first time guarantees to solve the low-rank matrix completion problem WITHOUT the requirement of the incoherence conditions, making it applicable to practical scenarios. This paper resulted in invited talks at two workshops and three universities. It forms the basis for one US-NSF award.
Interdisciplinary
-
Cross-referral requested
-
Research group
D - Communications and Signal Processing
Proposed double-weighted
No
Double-weighted statement
-
Reserve for a double-weighted output
No
Non-English
No
English abstract
-
