← 返回刚才阅读的位置

知识卡片

普通读书笔记卡 · 9113.f · 生成模型:GPT-6

挑战已解问题要设计新证据

理解等级:中级

某基准高分或几个漂亮案例不意味着问题在所有条件下已解决。可从失败模式、分布变化和下游需求提出更有区分力的测试;但保持质疑也不是否认任何进步,应清楚陈述已有能力与未覆盖范围。

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

      Computer vision is making amazing progress, but beware of the unsolved solved computer vision problems. For some vision task, it might appear as if there is one method that perfectly resolves the task. It might appear as if some foundational problems in computer vision are solved. It might often seem as if it is not possible to make progress regarding certain questions. But more often

    参考来源

    - 位置:《Foundations of Computer Vision (Antonio Torralba, Phillip Isola etc.)》第55章《重访简单视觉系统》(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/c055b.xhtml) - 结论依据:摘录中关于挑战已解问题要设计新证据的定义、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:Computer vision is making amazing progress, but beware of the unsolved solved computer vision problems. For some vision task, it might appear as if there is one method that perfectly resolves the task. It might appear as if some foundational problems in computer vision are solved. It might often seem as if it is not possible to make progress regarding certain questions. But more often