知识卡片
CNN示例须补齐通道和填充
示例若保持卷积尺寸,5×5卷积在步长1时需相应填充;20通道再卷到50通道,每个普通输出核要覆盖20输入通道。两次2倍池化后50×7×7展平为2450,再连500隐藏单元与10类输出,尺寸不能只靠图面猜测。
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
- · 人工神经网络
原书出处 · 1 处
- · 查看原书依据
Figure 10.12 shows a typical CNN that combines multiple convolutional and pooling layers. The network takes a 28 × 28 image as an input and trains 20 different 5 × 5 convolution kernels to create 20 feature maps of 28 × 28 each. This layer is followed by a MaxPooling layer that downsamples each feature map by a factor of two. These feature maps are
参考来源
- 位置:《Introduction to Autonomous Robots Mechanisms, Sensors, Actuators, and Algorithms (Nikolaus Correll, Bradley Hayes etc.)》第10章《人工神经网络》(来源项目:a19c7933-4eda-4176-aebd-30fe7e1ad292;源文件:extracted/Introduction to Autonomous Robots Mechanisms, Sensors, Actuators, and Algorithms (Nikolaus Correll, Bradley Hayes etc.)/_epub-src/OEBPS/xhtml/chapter_10.xhtml)
- 结论依据:摘录中关于CNN示例须补齐通道和填充的定义、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:Figure 10.12 shows a typical CNN that combines multiple convolutional and pooling layers. The network takes a 28 × 28 image as an input and trains 20 different 5 × 5 convolution kernels to create 20 feature maps of 28 × 28 each. This layer is followed by a MaxPooling layer that downsamples each feature map by a factor of two. These feature maps are