Unverified50% confidenceFactExact time
Integrated Gradients is a gradient-based attribution method proposed by Sundararajan et al. in 2017 that satisfies axioms such as sensitivity and implementation invariance.
1
Sources
50%
Confidence
Long-term
Relevance
10/22/2026
First Seen
Sources
Related Entities
Related Claims
UnverifiedThe "probing" methodology in NLP was systematically proposed by Alain and Bengio in 2017, becoming a standard tool in interpretability research.60% similarUnverifiedThe divsteps algorithm was originally proposed by Bernstein and Yang in 2019; it is an iterative method for computing polynomial greatest common divisors and modular inverses, involving only additions, subtractions, and shifts in each step, making it naturally suited to reversible implementation.60% similarUnverifiedNadeau and Bengio (2003) proposed a method for corrected variance estimation to address statistical inference issues in cross-validation.60% similarUnverifiedSTE (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.59% similarUnverifiedThe authors built the DISTRACT benchmark to quantify the view-bias phenomenon.59% similar
Cite This Claim
Stable URI
https://kongchang.com/claim/1084250API
curl https://kongchang.com/api/v1/knowledge/claims/1084250MCP
get_claim(id=1084250)