知识卡片
描述长度原则需要完整代价
在数据拟合相当时偏好更简洁的解释,可作为表示学习的取向。但最小描述长度比较的是模型及数据的完整编码成本,不能仅数潜在向量维度,也不能从简单重建网络直接宣称拥有规范化数据似然或必然更真实的解释。
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
- · 表示学习
原书出处 · 1 处
- · 查看原书依据
3. Compression is an embodiment of Occam's razor: among competing hypotheses that all explain the data equally well, the simplest is most likely to be true. As we will see below, and in the next chapter, many representation learning algorithms seek simple representations from which you can regenerate the raw data (e.g., autoencoders). Such representations explain the data in the formal sense that they assign
参考来源
- 位置:《Foundations of Computer Vision (Antonio Torralba, Phillip Isola etc.)》第30章《表示学习》(来源项目:a19c7933-4eda-4176-aebd-30fe7e1ad292;源文件:extracted/Foundations of Computer Vision (Antonio Torralba, Phillip Isola etc.)/_epub-src/OPS/xhtml/c030b.xhtml)
- 结论依据:摘录中关于描述长度原则需要完整代价的定义、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:3. Compression is an embodiment of Occam's razor: among competing hypotheses that all explain the data equally well, the simplest is most likely to be true. As we will see below, and in the next chapter, many representation learning algorithms seek simple representations from which you can regenerate the raw data (e.g., autoencoders). Such representations explain the data in the formal sense that they assign