Unverified50% confidenceFactExact time
深度知识追踪（DKT）将LSTM引入知识追踪任务，能够捕捉跨知识点的复杂依赖关系
1
Sources
50%
Confidence
Long-term
Relevance
7/20/2026
First Seen
Sources
Related Claims
Unverified2015年Piech等人提出深度知识追踪（DKT），用LSTM循环神经网络替代HMM，预测精度显著提升77% similarUnverifiedDify supports integration with mainstream large models such as DeepSeek, Tongyi Qianwen, and Zhipu, as well as custom tools, and provides drag-and-drop node visual workflow orchestration.66% similarUnverifiedDPR的核心创新是用双塔神经网络分别编码问题和文档，使语义相近内容在向量空间中聚拢64% similarUnverifiedDeepSeek-R1 has distilled versions based on knowledge distillation technology.63% similarUnverified内容感知裁剪结合了计算机视觉中的显著性检测和目标识别技术，现代深度学习方案如 U²-Net、DeepGaze 使用卷积神经网络学习视觉焦点62% similar
Cite This Claim
Stable URI
https://kongchang.com/claim/566711API
curl https://kongchang.com/api/v1/knowledge/claims/566711MCP
get_claim(id=566711)