Unverified50% confidenceFactExact time
深度学习中的深度指网络包含多个隐藏层
1
Sources
50%
Confidence
Long-term
Relevance
8/31/2026
First Seen
Sources
Related Entities
Related Claims
Unverified深度学习模型浅层网络学习的边缘、纹理等低层特征具有较强领域无关性，深层网络编码的高层语义特征更具领域特异性80% similarVerified深度学习使用多层神经网络，其结构受人脑神经元连接方式的启发77% similarUnverified2014 年 Yosinski 等人在论文《How transferable are features in deep neural networks?》中证明网络浅层学习的特征具有极高通用性，而深层特征逐渐变得任务特定75% similarUnverifiedSPLADE和DeepImpact是学习稀疏检索的代表，产生稀疏向量但每个词权重由神经网络学习得出，能做词项扩展75% similarUnverified《Dive into Deep Learning》是覆盖深度学习核心内容的开源教材75% similar
Cite This Claim
Stable URI
https://kongchang.com/claim/828240API
curl https://kongchang.com/api/v1/knowledge/claims/828240MCP
get_claim(id=828240)