We gratefully acknowledge support from
the Simons Foundation and member institutions.
Full-text links:


Current browse context:


Change to browse by:

References & Citations

DBLP - CS Bibliography


(what is this?)
CiteULike logo BibSonomy logo Mendeley logo del.icio.us logo Digg logo Reddit logo ScienceWISE logo

Computer Science > Machine Learning

Title: Inductive supervised quantum learning

Abstract: In supervised learning, an inductive learning algorithm extracts general rules from observed training instances, then the rules are applied to test instances. We show that this splitting of training and application arises naturally, in the classical setting, from a simple independence requirement with a physical interpretation of being non-signalling. Thus, two seemingly different definitions of inductive learning happen to coincide. This follows from the properties of classical information that break down in the quantum setup. We prove a quantum de Finetti theorem for quantum channels, which shows that in the quantum case, the equivalence holds in the asymptotic setting, that is, for large number of test instances. This reveals a natural analogy between classical learning protocols and their quantum counterparts, justifying a similar treatment, and allowing to inquire about standard elements in computational learning theory, such as structural risk minimization and sample complexity.
Comments: 6+10 pages
Subjects: Machine Learning (cs.LG); Quantum Physics (quant-ph); Machine Learning (stat.ML)
Journal reference: Phys. Rev. Lett. 118, 190503 (2017)
DOI: 10.1103/PhysRevLett.118.190503
Cite as: arXiv:1605.07541 [cs.LG]
  (or arXiv:1605.07541v2 [cs.LG] for this version)

Submission history

From: Peter Wittek [view email]
[v1] Tue, 24 May 2016 16:56:46 GMT (91kb)
[v2] Sat, 13 May 2017 10:48:23 GMT (58kb)

Link back to: arXiv, form interface, contact.