知识卡片
卷积模式按核与输入长度定义
一维卷积的full长度为L+K−1,valid为较长长度减较短长度再加1,same通常取两者较长长度。常见L≥K时same等于输入长;它不是一般意义的外推,valid也不是只输出K−1个值。
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
NumPy convolution gives you three possible modes for doing the convolution, in order of increasing output length: valid—Only output len(kernel) - 1 values for the convolution as our pure Python. same—Output a signal that is the same length as the input by extrapolating the signal beyond the beginning and end of the array. full—The output signal will have more sample than the input signal. The
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:摘录中关于卷积模式按核与输入长度定义的定义、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:NumPy convolution gives you three possible modes for doing the convolution, in order of increasing output length: valid—Only output len(kernel) - 1 values for the convolution as our pure Python. same—Output a signal that is the same length as the input by extrapolating the signal beyond the beginning and end of the array. full—The output signal will have more sample than the input signal. The