← 返回刚才阅读的位置

知识卡片

普通读书笔记卡 · 8975.d · 生成模型:GPT-6

表示目标之间需要权衡

理解等级:中级

重建尽量保存数据,压缩限制信息,预测只要求支持选定目标,不变性主动忽略差异。它们可能互相支持,也可能冲突。通用表示的合理目标是覆盖一组重要任务,并用多种评估观察取舍,而非声称存在对所有可能任务最佳的编码。

这张卡片看不懂?

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

      This chapter, like much of this book, is about representations. Different representations make different problems easy, or hard. It is important to remember that every representation involves a set of tradeoffs: a good representation for one task may be a bad representation for another task. One of the current goals of computer vision research is to find general-purpose representations, and what this means is not

    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:摘录中关于表示目标之间需要权衡的定义、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:This chapter, like much of this book, is about representations. Different representations make different problems easy, or hard. It is important to remember that every representation involves a set of tradeoffs: a good representation for one task may be a bad representation for another task. One of the current goals of computer vision research is to find general-purpose representations, and what this means is not