← 返回刚才阅读的位置

知识卡片

普通读书笔记卡 · 8903 · 生成模型:GPT-6

采样选择信息与成本

理解等级:初级

采样把连续域变成离散位置集合,采样越密通常需要越多存储和计算。决定采样间距应依据要保留的频率、任务精度和采集成本,而不是单纯追求更多像素。适当带限后已经满足重建条件时,继续加密并不自动增加原信号中的独立信息。

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

      Understanding how information is transformed when we convert a continuous signal into a discrete signal is important because we often have to optimize two competing factors: Maximize the amount of available information. This requires as many samples as possible. Minimize the computational cost and memory requirements. This implies that we want to minimize the number of samples we collect. Let's consider a one-dimensional (1D) continuous

    参考来源

    - 位置:《Foundations of Computer Vision (Antonio Torralba, Phillip Isola etc.)》第20章《图像采样与混叠》(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/c020b.xhtml) - 结论依据:摘录中关于采样选择信息与成本的定义、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:Understanding how information is transformed when we convert a continuous signal into a discrete signal is important because we often have to optimize two competing factors: Maximize the amount of available information. This requires as many samples as possible. Minimize the computational cost and memory requirements. This implies that we want to minimize the number of samples we collect. Let's consider a one-dimensional (1D) continuous