Unverified50% confidenceFactExact time
He initialization targets ReLU and its variants, scaling the variance to twice that of Xavier.
1
Sources
50%
Confidence
Long-term
Relevance
10/12/2026
First Seen
Sources
Related Entities
Related Claims
UnverifiedHe初始化针对ReLU激活函数设计，通过将权重方差设为2/n来保证各层激活值方差稳定65% similarUnverifiedReLU即max(0,x)，由Hahnloser等人于2000年提出51% similarUnverifiedThe XOR function is completely non-linearly separable for a single-layer linear model; this is the central counterexample put forward by Minsky and Papert in "Perceptrons" in 1969.50% similar
Cite This Claim
Stable URI
https://kongchang.com/claim/1011890API
curl https://kongchang.com/api/v1/knowledge/claims/1011890MCP
get_claim(id=1011890)