Unverified50% confidenceFactExact time
The neural tangent kernel (NTK) framework shows that as network width tends to infinity, training dynamics can be approximated as kernel regression under a fixed kernel function.
1
Sources
50%
Confidence
Long-term
Relevance
10/25/2026
First Seen
Sources
Related Entities
Related Claims
UnverifiedThe input gradient kernel reuses the weight blocks resident on the NPU and the dequantizer of the forward pass, avoiding transposed weight copies and ensuring that the forward and backward passes use the same effective weights.63% similarUnverifiedNeural networks exhibit permutation symmetry—reordering the neurons in a given layer leaves the network's functionality completely unchanged, as long as the weights of adjacent layers are adjusted accordingly.63% similarUnverifiedCamerata proposed the concept of "recursive generalization": neurons branch and grow along related directions during training, and each branch has a provenance rather than being forcibly stuffed in as superposition.61% similarUnverifiedA random reservoir network (often referring to an echo state network) is a type of recurrent neural network: the internal connection weights are randomly initialized and then kept fixed, and only the linear readout of the output layer is trained.58% similarUnverified液态神经网络的节点连接强度能随输入动态调整，本质上是一种连续时间递归神经网络（CT-RNN）58% similar
Cite This Claim
Stable URI
https://kongchang.com/claim/1097293API
curl https://kongchang.com/api/v1/knowledge/claims/1097293MCP
get_claim(id=1097293)