To refer to this page use:
http://arks.princeton.edu/ark:/88435/pr1kg0w
Full metadata record
DC Field | Value | Language |
---|---|---|
dc.contributor.author | Raz, Ran | - |
dc.date.accessioned | 2021-10-08T19:45:16Z | - |
dc.date.available | 2021-10-08T19:45:16Z | - |
dc.date.issued | 2017 | en_US |
dc.identifier.citation | Raz, Ran. "A Time-Space Lower Bound for a Large Class of Learning Problems." IEEE 58th Annual Symposium on Foundations of Computer Science (FOCS) (2017): pp. 732-742. doi:10.1109/FOCS.2017.73 | en_US |
dc.identifier.issn | 0272-5428 | - |
dc.identifier.uri | https://eccc.weizmann.ac.il/report/2017/020/ | - |
dc.identifier.uri | http://arks.princeton.edu/ark:/88435/pr1kg0w | - |
dc.description.abstract | We prove a general memory-samples lower bound that applies for a large class of learning problems and shows that for every problem in that class, any learning algorithm requires either a memory of quadratic size or an exponential number of samples. Our result is stated in terms of the norm of the matrix that corresponds to the learning problem. Let X, A be two finite sets. A matrix M : A × X → {-1, 1} corresponds to the following learning problem: An unknown element x ∈ X was chosen uniformly at random. A learner tries to learn x from a stream of samples, (a 1 , b 1 ), (a 2 , b 2 ) ..., where for every i, a i ∈ A is chosen uniformly at random and b i = M(ai, x). Let σ max be the largest singular value of M and note that always σ max ≤ |A| 1/2 · |X| 1/2 . We show that if σ max ≤ |A| 1/2 · |X| 1/2-ε , then any learning algorithm for the corresponding learning problem requires either a memory of size at least Ω ((εn) 2 ) or at least 2 Ω(εn) samples, where n = log 2 |X|. As a special case, this gives a new proof for the memory-samples lower bound for parity learning [14]. | en_US |
dc.format.extent | 732 - 742 | en_US |
dc.language.iso | en_US | en_US |
dc.relation.ispartof | IEEE 58th Annual Symposium on Foundations of Computer Science (FOCS) | en_US |
dc.rights | Author's manuscript | en_US |
dc.title | A Time-Space Lower Bound for a Large Class of Learning Problems | en_US |
dc.type | Conference Article | en_US |
dc.identifier.doi | 10.1109/FOCS.2017.73 | - |
pu.type.symplectic | http://www.symplectic.co.uk/publications/atom-terms/1.0/conference-proceeding | en_US |
Files in This Item:
File | Description | Size | Format | |
---|---|---|---|---|
SpaceTimeLowerBoundLearningProblems.pdf | 651.17 kB | Adobe PDF | View/Download |
Items in OAR@Princeton are protected by copyright, with all rights reserved, unless otherwise indicated.