知识卡片
隐层涌现不等于显式分组
神经网络可能在内部形成与物体、轮廓或层相关的特征,使某些传统手工模块不再单独出现。但隐含表示不自动提供可直接使用的对象身份、关系或一致分割。需要通过适合目标的读出和评估确认,而不是从架构或可视化推定已经完成组织。
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
- · 知觉分组
原书出处 · 1 处
- · 查看原书依据
It may seem like some of the topics in this chapter are out of fashion in modern computer vision. Edge detection and bottom-up segmentation are not the workhorses they once were. But this doesn't mean that perceptual organization is absent from current systems. These systems still see the world in terms of objects, contours, layers, and so forth, but those structures are emergent in their
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:摘录中关于隐层涌现不等于显式分组的定义、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:It may seem like some of the topics in this chapter are out of fashion in modern computer vision. Edge detection and bottom-up segmentation are not the workhorses they once were. But this doesn't mean that perceptual organization is absent from current systems. These systems still see the world in terms of objects, contours, layers, and so forth, but those structures are emergent in their