tf.nn函数总结

1、tf.nn.conv2d

conv2d(
    input,
    filter,
    strides,
    padding,
    use_cudnn_on_gpu=None,
    data_format=None,
    name=None
)

根据给定的四维张量和滤波张量计算2-D滤波。其中，给定输入的张量维度为[batch, in_height, in_weidht, in_channels]和滤波核的张量的维度为[filter_height, filter_width, in_channesl, out_channels],计算会转换为一下步骤

(1)将filter平铺成二维的张量，维度为[filter_height*fliter_width*in_channels, output_channels]

(2)将image转换成一个张量[batch, out_height, out_width, filter_height*filter_width*in_channels]

(3)对于每个patch，右乘滤波矩阵和image块

默认格式为：NHWC

output[b, i, j, k] =
    sum_{di, dj, q} input[b, strides[1] * i + di, strides[2] * j + dj, q] *
                    filter[di, dj, q, k]