Unverified60% confidenceFactExact time
The Straight-Through Estimator (STE) was proposed by Bengio et al. in 2013 and is the theoretical foundation of extreme quantization training.
2
Sources
60%
Confidence
Long-term
Relevance
7/20/2026
First Seen
Sources
Related Claims
UnverifiedSTE (Straight-Through Estimator) performs truncated rounding in the forward pass but passes gradients straight through in the backward pass (approximating the gradient as 1) to address the non-differentiability of quantization operations, and was formally systematized in a 2013 paper by Bengio et al.70% similarUnverified对抗性样本（Adversarial Examples）的理论基础由Szegedy等人于2013年首次系统阐述60% similarUnverifiedFragmented Entangled Representations (FER), from research by Kenneth Stanley and others, describes the internal representational state of a neural network after training in which conceptual features are highly entangled and fragmentarily distributed across parameters.57% similarUnverifiedThe researchers derived testable eigenmodes through a non-local probability wave equation called 'Generalized Behavioral Intelligence' (GBI).56% similarUnverifiedNadeau and Bengio (2003) proposed a method for corrected variance estimation to address statistical inference issues in cross-validation.56% similar
Cite This Claim
Stable URI
https://kongchang.com/claim/571228API
curl https://kongchang.com/api/v1/knowledge/claims/571228MCP
get_claim(id=571228)