Closed ocsponge closed 6 years ago
forgive my poor English :)
Hi, tail prediction and head prediction are using two sets of De and Dr. Hence 2k + 2k = 4K.
Best,
On Thu, May 31, 2018 at 5:38 AM ZhangYu notifications@github.com wrote:
hi, @bxshi https://github.com/bxshi, in your paper, you said
Therefore, compared to simple models like TransE, ProjE only increases the number of parameters by 5k + 1, where 1, 4k, and k are introduced as the projection bias, combination weights, and combination bias respectively
but the combination weights De and Dr both are k*k diagonal matrices, each of them should have k parameters so that there would be 2k, why 4k in your paper?
— You are receiving this because you were mentioned. Reply to this email directly, view it on GitHub https://github.com/bxshi/ProjE/issues/7, or mute the thread https://github.com/notifications/unsubscribe-auth/ABQQQ2ng1LWKEXwciL7xhd79_soyRyhQks5t39YmgaJpZM4UU5Ki .
so for the relation prediction task, there would be another 2k parameters which is different from head prediction(2k) and tail prediction task(2k)? is that correct?
Right
On Thu, May 31, 2018 at 8:43 AM ZhangYu notifications@github.com wrote:
so for the relation prediction task, there would be another 2k parameters which is different from head prediction(2k) and tail prediction task(2k)? is that correct?
— You are receiving this because you were mentioned. Reply to this email directly, view it on GitHub https://github.com/bxshi/ProjE/issues/7#issuecomment-393553668, or mute the thread https://github.com/notifications/unsubscribe-auth/ABQQQ5VDUG7_HpzhiqsIhr8KixcYdRsYks5t4AF_gaJpZM4UU5Ki .
thank you for your quick reply!
hi, @bxshi, in your paper, you said
but the combination weights De and Dr both are k*k diagonal matrices, each of them should have k parameters so that there would be 2k, why 4k in your paper?