Unverified50% confidenceFactExact time
In the two-neuron ReLU teacher-student model, any critical point in the population squared-loss landscape at which all student weights are nonzero is either a global minimum or a saddle point.
1
Sources
50%
Confidence
Long-term
Relevance
11/2/2026
First Seen
Sources
Related Entities
Related Claims
UnverifiedIn the task of predicting final exam scores from midterm scores, a basic neural network using only students' own scores has high error.66% similarUnverifiedPReLU (Parametric ReLU) treats the slope of the negative half-axis as a learnable parameter, rather than fixing it at 0 or some constant.65% similarUnverifiedThe study tests on four types of loss surfaces: well-conditioned linear regression, ill-conditioned linear regression, logistic regression, and non-convex small neural networks.64% similarUnverifiedA billing item is counted only when a model and direction exist in both regions at the same time; a model missing in a region contributes zero rather than an infinite ratio.63% similarUnverifiedThe paper proves that, under appropriate assumptions, relative to the target amplitude, the amplitude of the learned solution at high frequencies consistently tends to zero throughout training.63% similar
Cite This Claim
Stable URI
https://kongchang.com/claim/1125632API
curl https://kongchang.com/api/v1/knowledge/claims/1125632MCP
get_claim(id=1125632)