机器学习实践:图像分类
使用集合让一切井井有条
根据您的偏好保存内容并对其进行分类。
检查您的理解情况:卷积
将一个二维 3x3 卷积过滤器应用到一个二维 4x4 输入特征图上(未添加内边距):

输出特征图是形状什么的?
2x2
当 3x3 过滤器在 4x4 特征图上滑动时,过滤器可以放置在 4 个不同位置,生成一个 2x2 输出特征图:

3x3
虽然过滤器的尺寸为 3x3,但输出特征图较小,因为在 4x4 输入特征图上,放置过滤器的可能位置少于 9 个(3x3)。
4x4
为了生成与输入特征图尺寸相同的输出特征图(不含内边距),卷积过滤器的形状必须为 1x1。大于 1x1 的过滤器的输出特征图会小于输入特征图。由于我们的过滤器的尺寸为 3x3,因此输出特征图必须小于 4x4。
如未另行说明,那么本页面中的内容已根据知识共享署名 4.0 许可获得了许可,并且代码示例已根据 Apache 2.0 许可获得了许可。有关详情,请参阅 Google 开发者网站政策。Java 是 Oracle 和/或其关联公司的注册商标。
最后更新时间 (UTC):2025-07-27。
[null,null,["最后更新时间 (UTC):2025-07-27。"],[[["\u003cp\u003eApplying a 3x3 convolutional filter to a 4x4 input feature map without padding results in a 2x2 output feature map.\u003c/p\u003e\n"],["\u003cp\u003eThe output feature map is smaller than the input because the filter's size limits the number of positions it can occupy on the input.\u003c/p\u003e\n"],["\u003cp\u003eA 1x1 filter would be required to maintain the input's dimensions in the output without padding.\u003c/p\u003e\n"]]],[],null,["# ML Practicum: Image Classification\n\n\u003cbr /\u003e\n\nCheck Your Understanding: Convolution\n-------------------------------------\n\nA two-dimensional, 3x3 convolutional filter is applied to a two-dimensional 4x4 input feature map\n(no padding added):\n\nWhat is the shape of the output feature map? \n2x2 \nAs the 3x3 filter slides over the 4x4 feature map, there are 4 unique locations in which it can be placed, which results in a 2x2 output feature map: \n3x3 \nWhile the filter itself is 3x3, the output feature map is smaller because there are fewer than 9 (3 times 3) possible locations where the filter can be placed on the 4x4 input feature map. \n4x4 \nTo generate an output feature map with the same dimensions as the input feature map with no padding, the convolutional filter would have to be 1x1 in shape. A filter larger than 1x1 will produce an output feature map that is smaller than the input feature map. Because our filter is 3x3, the output feature map must be smaller than 4x4."]]