Unverified50% confidenceFactExact time
Non-polynomial losses have unbounded state requirements over histories with equal predictions, and this obstacle persists for bounded scalar logistic loss even after fixing an arbitrary finite-order jet of derivatives.
1
Sources
50%
Confidence
Long-term
Relevance
10/12/2026
First Seen
Sources
Related Entities
Related Claims
UnverifiedFor smooth convex residual losses, under the premise that both features and labels are unconstrained real numbers, a fixed-dimensional exact continuous recurrent state exists if and only if the loss function is a polynomial.80% similarUnverified香农的率失真理论证明，在给定比特率约束下，任何有损压缩都存在不可逾越的失真下界77% similarUnverifiedThe loss curve of linear regression takes the shape of a convex bowl; convexity guarantees that optimization has a unique global minimum, and gradient descent will not get trapped in local optima.76% similarUnverifiedThe logistic loss does not enjoy the property of polynomial losses whereby a finite-dimensional representation suffices for exact continuation of learning; exactly preserving learning capacity theoretically requires unbounded memory overhead.75% similarUnverifiedA reasonable prediction loss combined with 0% downstream control accuracy is highly consistent with representation collapse, meaning the encoder has fooled the self-supervised objective without learning any useful dynamic information.74% similar
Cite This Claim
Stable URI
https://kongchang.com/claim/1002202API
curl https://kongchang.com/api/v1/knowledge/claims/1002202MCP
get_claim(id=1002202)