Confusing about the ComputeNormlizedSGD

[yyangzh]

Hi,
sorry for the bothering,

I'm confused about the function ComputeNormlizedSGD( F1, F2, size1, size2).
In this function, symmetric computation is applied.
However, when you reverse the direction from I1->I2 to I2->I1, should the fundamental F1, F2 be changed to their transposed one F1^T, F2^T?
In your function, d2 = one_iteration(F2, F1, h2, w2, h1, w1) and I can't understand why there is no such transformation.

I'm looking forward to your kindly reply :).

[JiawangBian]

Thanks for your attention on this work. So sorry that as I haven't read this code for a long time, it is hard for me to answer your questions quickly. However, I am confident that it is correct, as I paid much time on checking them at that moment. I hope that you try to read it more several days. If you are still confused after that, I will try to back again to this project and reply to you.

[yyangzh]

Thanks for your reply, I will try my best to figure it out.

[jingedawang]

According to this paper, we should "reverse the roles of F1 and F2" instead of reversing the roles of I1 and I2. In this circumstance, this line should be

    d2 = one_iteration(F2, F1, h1, w1, h2, w2);

Please have a look at this problem.

[yyangzh]

According to this paper, we should "reverse the roles of F1 and F2" instead of reversing the roles of I1 and I2. In this circumstance, this line should be

    d2 = one_iteration(F2, F1, h1, w1, h2, w2);

Please have a look at this problem.

Yes. This way makes sense.