Wenshui LUO

icon indicating a website link https://randydkx.github.io/

icon company Nanjing University of Science and Technology icon location Nanjing University of Science and Technology icon location I am Wenshui LUO, a student at NJUST.

Results 4 comments of Wenshui LUO

Bi-LSTM attention calc may be wrong

I think so too.

Question about eq(9) in your paper.

@DongXiang-CV It's ok to have unnormalized probabilities, but what confused me is that I think eq9 is p(ci|xi), not p(xi|ci), with prior probabilities fixed for every ci i.e. p(ci=k)=1/C for...

question about implementation of eq(5)

@KaiWU5 Thanks for your reply, I know that but I find the re-weighting factor is qi ^2 for i = j in implementation, it should be 1 in your paper.

When operating a custom dataset, will it work if I just set config.toml?

@limhasic hard to do that, I'm recently on the synthesis of fraud detection transactions, I found that tab-DDPM proposed here can only handle samll feature sets. My trans data includes...