知识卡片
平均与求和池化保留不同量
平均池化概括窗口平均响应,求和对出现次数和有效长度更敏感,最大池化强调峰值。选择需与目标及填充掩码相配,不能把卷积内部局部求和混成跨全序列的全局池化,也不能认为一种汇聚总最好。
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
原书出处 · 1 处
- · 查看原书依据
Note If the patterns you are looking for are spread out over several different locations within a passage of text, then you may want to try mean pooling for some of your kernels. Mean pooling is the process of taking the average of all the values in a window rather than just the peak or max.
参考来源
- 位置:《Natural Language Processing in Action, Second Edition》第07章《卷积网络识别文本模式》(来源项目:a19c7933-4eda-4176-aebd-30fe7e1ad292;源文件:extracted/Natural Language Processing in Action, Second Edition/_epub-src/OEBPS/Text/chapter-7.html)
- 结论依据:摘录中关于平均与求和池化保留不同量的定义、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:Note If the patterns you are looking for are spread out over several different locations within a passage of text, then you may want to try mean pooling for some of your kernels. Mean pooling is the process of taking the average of all the values in a window rather than just the peak or max.