知识卡片
可解释关联不自动构成因果模型
一个特征和现实概念容易对应,可帮助提出机制假设和检查错误,却不能凭预测性或物理解释就认定因果。需要另外说明干预、混杂和适用条件;神经网络学到特征也不使这些识别问题自行消失。
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
- · 神经网络基础
原书出处 · 1 处
- · 查看原书依据
The Holy Grail of feature engineering is finding representations that say something about the physics of the real world. If your features are explainable according to real-world phenomena, you can begin to build confidence that they are more than just predictive. You may have a truly causal model that says something true, in general, about the world—not just for your dataset.
参考来源
- 位置:《Natural Language Processing in Action, Second Edition》第05章《神经网络基础》(来源项目:a19c7933-4eda-4176-aebd-30fe7e1ad292;源文件:extracted/Natural Language Processing in Action, Second Edition/_epub-src/OEBPS/Text/chapter-5.html)
- 结论依据:摘录中关于可解释关联不自动构成因果模型的定义、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:The Holy Grail of feature engineering is finding representations that say something about the physics of the real world. If your features are explainable according to real-world phenomena, you can begin to build confidence that they are more than just predictive. You may have a truly causal model that says something true, in general, about the world—not just for your dataset.