Unverified50% confidenceFactExact time
The paper focuses on the nonlinear SGD setting in which the gradient noise has a regularly varying tail index t ∈ (1,2).
1
Sources
50%
Confidence
Long-term
Relevance
10/19/2026
First Seen
Sources
Related Entities
Related Claims
UnverifiedSGD引入的随机噪声是一种隐式正则化来源，使优化轨迹倾向于平坦极小值而非尖锐极小值72% similarUnverifiedUnder the homoscedasticity assumption, noise variance is constant, which is the standard assumption of classical linear regression; under heteroscedasticity, variance changes with input.70% similarUnverifiedThe paper introduces the strongly adapted sub-ψ condition to characterize the concentration properties of noise sequences under adapted filtering.69% similarUnverifiedThere is a Tweedie equation connection between the classical denoiser and the score: the optimal denoiser can be written as D(x) = (1/σ²)(x + σ² ∇_x log p_σ(x)).67% similarUnverifiedDenoised mean difference is a variant of the linear probing technique that eliminates noise unrelated to the target concept by averaging or orthogonalizing results across multiple control conditions.66% similar
Cite This Claim
Stable URI
https://kongchang.com/claim/1064974API
curl https://kongchang.com/api/v1/knowledge/claims/1064974MCP
get_claim(id=1064974)