← 返回起始笔记位置

知识卡片

普通读书笔记卡 · 8979.a · 生成模型:GPT-6

离散分组连接连续表示

理解等级:中级

知觉组织偏向对象、部分、关系等模块化结构,表示学习常以连续向量为起点。二者可通过学习特征后聚类、对区域作编码等方式结合。连续特征和离散标签分别保留不同信息,选用哪种应依据后续推理所需的结构与不确定性。

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

      Perceptual organization aims to represent visual content in terms of modular and disentangled elements that make subsequent processing easier. In this way, perceptual organization shares the same goal as visual representation learning, although the former deals more with discrete structures whereas the latter has focused mostly on continuous representations. These two fields evolved somewhat separately, with perceptual organization having its roots in psychology and representation

    参考来源

    - 位置:《Foundations of Computer Vision (Antonio Torralba, Phillip Isola etc.)》第31章《知觉分组》(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/c031.xhtml) - 结论依据:摘录中关于离散分组连接连续表示的定义、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:Perceptual organization aims to represent visual content in terms of modular and disentangled elements that make subsequent processing easier. In this way, perceptual organization shares the same goal as visual representation learning, although the former deals more with discrete structures whereas the latter has focused mostly on continuous representations. These two fields evolved somewhat separately, with perceptual organization having its roots in psychology and representation