Unverified50% confidenceFactExact time
With sigmoid activation combined with cross-entropy loss, differentiating the loss with respect to the weights simplifies to ∂L/∂w = (ŷ - y) · x.
1
Sources
50%
Confidence
Long-term
Relevance
9/11/2026
First Seen
Sources
Related Entities
Related Claims
VerifiedThe mathematical definition of entropy is H(p) = -∑ p(x_i) log p(x_i).75% similarVerifiedCross-entropy = entropy of the true distribution + KL divergence.63% similarUnverifiedIn 1951, von Neumann proved the von Neumann inequality: for any contraction operator T and any polynomial p, ‖p(T)‖ ≤ sup{|p(z)| : |z| ≤ 1}.63% similarVerifiedClaude Shannon published 'A Mathematical Theory of Communication' in 1948, proposing the concept of information entropy, whereby low-probability events carry higher information content, with the formula I(x) = -log₂P(x).60% similarUnverifiedThe mathematical expression for Ledoit-Wolf shrinkage is Σ_shrunk = δ·F + (1-δ)·S, where S is the sample covariance matrix, F is the shrinkage target, and δ∈[0,1] is the shrinkage intensity.60% similar
Cite This Claim
Stable URI
https://kongchang.com/claim/900285API
curl https://kongchang.com/api/v1/knowledge/claims/900285MCP
get_claim(id=900285)