Open problem: Tensor decompositions: Algorithms up to the uniqueness threshold?

Aditya Bhaskara, Moses Charikar, Ankur Moitra, Aravindan Vijayaraghavan

Research output: Contribution to journalConference articlepeer-review

Abstract

Experts state that tensor decompositions are unique in a way which not the same as standard matrix decompositions. Tensor decompositions have a broad range of applications in fields as diverse as psychometrics, chemometrics, medicine, signal processing and data mining. It is more reasonable to assume that the factors are linearly independent than that they are completely uncorrelated with each other in many applications. One of the related algorithms plays a key role in many applications in statistics and machine learning. The canonical setting is when samples are given from a high-dimensional mixture model. The concerned algorithm has applications to phylogenetic reconstruction, learning mixtures of Gaussians, independent component analysis, multi-view models and learning mixtures of product distributions.

Original languageEnglish (US)
Pages (from-to)1280-1282
Number of pages3
JournalJournal of Machine Learning Research
Volume35
StatePublished - Jan 1 2014
Event27th Conference on Learning Theory, COLT 2014 - Barcelona, Spain
Duration: Jun 13 2014Jun 15 2014

ASJC Scopus subject areas

  • Software
  • Control and Systems Engineering
  • Statistics and Probability
  • Artificial Intelligence

Fingerprint Dive into the research topics of 'Open problem: Tensor decompositions: Algorithms up to the uniqueness threshold?'. Together they form a unique fingerprint.

Cite this