Unverified50% confidenceFactExact time
This work proposes the first algorithm that learns k arbitrary positive linear combinations of ReLUs in fully polynomial time without strong assumptions (with respect to the Gaussian distribution and L2 loss, both sample and time complexity are poly(d, k, 1/ε)).
1
Sources
50%
Confidence
Long-term
Relevance
10/26/2026
First Seen
Sources
Related Entities
Related Claims
UnverifiedAll previous algorithms for this problem require superpolynomial time in k or 1/ε, or rely on strong well-conditioned assumptions.71% similarUnverifiedPreviously, the best known algorithms for the fixed-point problem of ℓ∞ contraction maps were exponential-time algorithms based on the simplex method, or polynomial-time algorithms for special structures.69% similarUnverifiedThe paper "A Quasipolynomial Algorithm for ℓ∞ Contractions" proposes an algorithm that can compute an ε-approximate fixed point of any non-expansive map under the d-dimensional ℓ∞ norm in quasipolynomial time.69% similarUnverifiedThe paper uses quantitative geometric estimates of regular polynomial systems to transform the original geometric testing problem into a finite-dimensional, structurally explicit auxiliary optimization problem.68% similarUnverifiedFor a model to truly generalize modular arithmetic tasks, it needs to implicitly discover periodic representations such as the discrete Fourier transform, rather than memorizing the answer for every pair of inputs.68% similar
Cite This Claim
Stable URI
https://kongchang.com/claim/1098110API
curl https://kongchang.com/api/v1/knowledge/claims/1098110MCP
get_claim(id=1098110)