Unverified50% confidenceFactExact time
ResNet, proposed by Kaiming He et al. in 2015, introduced residual connections (Skip Connections), making it possible to train networks of hundreds or even thousands of layers, and it has become a standard component in the Transformer architecture.
1
Sources
50%
Confidence
Long-term
Relevance
8/31/2026
First Seen
Sources
Related Entities
Related Claims
Unverified跳跃连接最早由Kaiming He等人在2015年提出的ResNet（残差网络）中引入68% similarUnverifiedA developer shared on Reddit a project in which a neural network learned the optimal strategy for tic-tac-toe from scratch, which users can run in a terminal and play against the trained network.58% similarUnverified越来越多的开发工具开始回归原生技术栈，Zed 编辑器是这一潮流的代表54% similarUnverifiedConnectionism began with Rosenblatt's perceptron in 1958, fell into obscurity following Minsky's criticism in 1969, was revived through the backpropagation algorithm in 1986, and fully exploded after AlexNet won the ImageNet competition in 2012.53% similar
Cite This Claim
Stable URI
https://kongchang.com/claim/831000API
curl https://kongchang.com/api/v1/knowledge/claims/831000MCP
get_claim(id=831000)