使用keras2.2.4运行时遇到一个问题

b = K.batch_dot(o, u_hat_vecs, [2, 3])
ValueError: Can not do batch_dot on inputs with shapes (None, 10, 10, 16) and (None, 10, None, 16) with axes=[2, 3]. x.shape[2] != y.shape[3] (10 != 16)

而且b = K.batch_dot(o, u_hat_vecs, [2, 3])
中b不需要和上一轮的b进行叠加吗

[zhen8838]

从tf1.14之后，batch_dot的预期行为有所改变，本来得到的o应该是(None, 10, 16)的。如果你使用tensorflow，将迭代流程修改为如下即可：

for i in range(self.routings):
    c = softmax(b, 1)  # shape = [None, num_capsule, input_num_capsule]
    # o = K.batch_dot(c, u_hat_vecs, [2, 2])
    o = tf.einsum('bin,binj->bij', c, u_hat_vecs)
    if i < self.routings - 1:
        o = K.l2_normalize(o, -1)
        # b = K.batch_dot(o, u_hat_vecs, [2, 3])
        b = tf.einsum('bij,binj->bin', o, u_hat_vecs)

另外，我尝试了将b与上一轮结果叠加，效果反而变差了。

[bojone]

b = tf.einsum('bij,binj->bin', o, u_hat_vecs)

正解。感谢帮助他人解惑。