Unverified50% confidenceFactTime unknown
监督学习使用带标签的数据训练模型，例如图像分类中每张图片都标注了类别
1
Sources
50%
Confidence
Medium-term (~90 days)
Relevance
7/2/2026
First Seen
Valid until: 9/30/2026
Sources
Related Claims
Unverified数据标注质量直接决定了模型训练的上限，专业领域的标注需要领域专家参与72% similarPartially VerifiedSupervised learning training uses labeled data consisting of sample pairs of known inputs and their corresponding outputs.71% similarUnverified各家图像生成模型多依赖LAION-5B等大规模开放数据集训练70% similarUnverified轨迹数据集标准化的意义可类比于ImageNet对计算机视觉、Common Crawl对大语言模型预训练的作用70% similarUnverified训练风格统一的参考图即可训练出专属风格的LoRA模型69% similar
Cite This Claim
Stable URI
https://kongchang.com/claim/55358API
curl https://kongchang.com/api/v1/knowledge/claims/55358MCP
get_claim(id=55358)