Calculation of NHP's log-likelihood

[HaochenWang1243]

Hi. I have doubts on how the code calculates NHP's log-likelihood, which is specified by equation (8) in the NHP paper:

The former part of equation (8) is the sum of individual-type log intensities at all event occurrences. In the compute_loglikelihood method in torch_basemodel.py, however, I saw the type dimension (last dimension) of lambda_at_event is summed:

EasyTemporalPointProcess/easy_tpp/model/torch_model/torch_basemodel.py

Line 104 in 01551fb

event_lambdas = torch.sum(lambda_at_event * lambda_type_mask, dim=-1) + self.eps

which doesn't seem to comply with equation (8) shown above. FYI, the caller of this method in NHP is

EasyTemporalPointProcess/easy_tpp/model/torch_model/torch_nhp.py

Line 265 in 01551fb

event_ll, non_event_ll, num_events = self.compute_loglikelihood(lambda_at_event=lambda_at_event,

where lambda_at_event has shape [batch_size, num_times=max_len-1, num_event_types], as the comments mentioned.

The latter part of equation (8), according to Algorithm 1 of the NHP paper:

is calculated by uniformly sampling the intensity function over the whole time frame of the event sequence. In the code, however, it seems like the sampling is done by drawing equidistant samples in each inter-event interval:

EasyTemporalPointProcess/easy_tpp/model/torch_model/torch_nhp.py

Line 255 in 01551fb

# for every batch and every event point => do a sampling (num_mc_sampling)

which, I imagine, can give quite different results from uniform sampling for event sequences of large inter-event-time variance.

Maybe I'm missing something. Thank you for your help.

[iLampard]

Hi,

Regarding the sampling method used here, we are working on to publish a newer version to fix relevant issues. this will be done shortly.

[HaochenWang1243]

Thank you. It'll be really helpful to my research. For now it's a bit hard to reproduce the results in the original papers.

[iLampard]

We will have quite a few PRs in next month, fixing the potential problems in loglike computation, some inconsistency in reproducing models and other issues.

Besides, we will issue a tech report to explain the details of implementation and re-publish the experiment results.