知识卡片
生产数据回路需要主动设计
请求、预测与后续结果可帮助发现真实问题并改进训练,但必须设计收集、处理、标注与评价的完整回路。使用量增加只有在数据有质量、可用且反馈机制可靠时才可能促进改善;未经处理的历史预测不会自然成为正确训练标签。
这张卡片看不懂?
这张卡片已通过人工审查,不再接受失败反馈。
已有反馈 · 0 条
还没有人提交反馈。
在整书笔记中的位置
原书出处 · 1 处
- · 查看原书依据
Teams often throw away the data that a model generates in production (e.g., requests and predictions)—data that could be used to improve their ML models. This can occur due to lack of proper data collection, processing, and labeling mechanisms. As a result, teams throw away a valuable feedback mechanism that can enhance the performance of the ML system.
参考来源
- 位置:《Effective Machine Learning Teams (David Tan, Ada Leung, David Colls)》第09章《MLOps与持续交付》(来源项目:a19c7933-4eda-4176-aebd-30fe7e1ad292;源文件:extracted/Effective Machine Learning Teams (David Tan, Ada Leung, David Colls)/_epub-src/OEBPS/ch09.html)
- 结论依据:摘录中关于生产数据回路需要主动设计的定义、比较或条件支撑卡片;原书表述不严谨的范围已在覆盖记录逐项限定,不把案例结果外推为普遍结论。
- 原始内容:Teams often throw away the data that a model generates in production (e.g., requests and predictions)—data that could be used to improve their ML models. This can occur due to lack of proper data collection, processing, and labeling mechanisms. As a result, teams throw away a valuable feedback mechanism that can enhance the performance of the ML system.