Unverified50% confidenceFactExact time
Since the α-stable distribution has infinite variance, using it directly for neural network training would cause gradient explosion, so a truncated version is often adopted in practical applications.
1
Sources
50%
Confidence
Long-term
Relevance
10/16/2026
First Seen
Sources
Related Entities
Related Claims
Unverified深度神经网络天然倾向于过拟合训练分布，对分布外样本的处理能力往往大幅下降77% similarUnverified一段数据参与神经网络训练后，其影响以分布式方式弥散在数十亿参数中，难以被真正删除74% similarUnverifiedNaïve post-training quantization causes individual weight errors to accumulate and amplify across multiple layers—a phenomenon known as compounding error—and at 2-bit or 1-bit precision the model structure collapses72% similarVerifiedVanishing gradient refers to the exponential decay of gradient values during backpropagation in deep neural networks, causing parameters near the input layer to receive almost no updates.72% similarUnverifiedNormalization layers in neural networks (BatchNorm, LayerNorm) continuously perform data purification during training, smoothly estimating noise through batch statistics, and the perturbation of anomalous samples on the overall mean and variance is diluted by the large number of normal samples.72% similar
Cite This Claim
Stable URI
https://kongchang.com/claim/1046097API
curl https://kongchang.com/api/v1/knowledge/claims/1046097MCP
get_claim(id=1046097)