Unverified50% confidenceFactExact time
The paper derives explicit projected and excess-objective tails, extreme quantiles, and, for a fixed batch size b, the factor b^{1-t}.
1
Sources
50%
Confidence
Long-term
Relevance
10/19/2026
First Seen
Sources
Related Entities
Related Claims
UnverifiedThe limiting log marginal likelihood derived under the HTMP assumption is a convex function with respect to the heavy-tail degree parameter.67% similarUnverifiedThe paper derives generalization error upper bounds separately for the preliminary estimator and the constrained estimator; these bounds are expressed in terms of complexity measures such as the covering number or VC dimension of the family of measurable sets, and do not depend on the specific structure of the response space.66% similarUnverifiedThe paper establishes non-asymptotic, high-probability upper bounds for the convergence of empirical occupation measures to the invariant measure, under the contraction assumption in kernel-induced geometry and the standard spectral gap assumption.66% similarUnverifiedThe simulations validate the batch-size factor b^{1-t}, indicating that increasing the batch size can mitigate the effect of heavy-tailed gradients.65% similarUnverifiedThe paper decomposes the estimation difficulty into: the degree to which the true mappings deviate from a common non-decreasing shape (characterized by the sup-distance τ and the total variation V), and the scarcity n of labels per problem.65% similar
Cite This Claim
Stable URI
https://kongchang.com/claim/1064982API
curl https://kongchang.com/api/v1/knowledge/claims/1064982MCP
get_claim(id=1064982)