Transformer-Hawkes-Process
Transformer-Hawkes-Process copied to clipboard
some question about calculating probalitility
I found that the intensity function is not used when calculating the probability,I don't know if you can answer for me. Thank you.
This paper didn't utilize the probability to predict the future event, it used the equation in the bottom of Page 7, please check.
In practice, using the intensity function doesn't work very well. Utilising a non-linearity of the model output and then a regression layer works well for event type prediction.
I also have the same question, and this paper states the conditional intensity function is used to predict the future event type and time stamp. However, I find the model in this repo only uses a linear layer to get the prediction result. I hope someone can help me, thanks a lot!
I also have the same question, and this paper states the conditional intensity function is used to predict the future event type and time stamp. However, I find the model in this repo only uses a linear layer to get the prediction result. I hope someone can help me, thanks a lot!
it used the equation in the bottom of Page 7~
I also have the same question, and this paper states the conditional intensity function is used to predict the future event type and time stamp. However, I find the model in this repo only uses a linear layer to get the prediction result. I hope someone can help me, thanks a lot!
it used the equation in the bottom of Page 7~
Thanks for your reply! I only find the equation in the bottom of Page 7 is used in Utils.log_likelihood function to get log-likelihood. And I am not sure if this equation is used in the prediction process. Maybe I miss or misunderstand some details. I will read the codes carefully. Thanks!
I also have the same question, and this paper states the conditional intensity function is used to predict the future event type and time stamp. However, I find the model in this repo only uses a linear layer to get the prediction result. I hope someone can help me, thanks a lot!
it used the equation in the bottom of Page 7~
Thanks for your reply! I only find the equation in the bottom of Page 7 is used in Utils.log_likelihood function to get log-likelihood. And I am not sure if this equation is used in the prediction process. Maybe I miss or misunderstand some details. I will read the codes carefully. Thanks!
Referring to the 3rd floor, the 5.3 experimental part of the paper mentions that Eq.7 has a general effect, and a linear layer is used for this
Thanks a lot!