Hierarchical Contrastive Learning for Temporal Point Processes
Authors: Qingmei Wang, Minjie Cheng, Shen Yuan, Hongteng Xu
AAAI 2023 | Conference PDF | Archive PDF | Plain Text | LLM Run Details
| Reproducibility Variable | Result | LLM Response |
|---|---|---|
| Research Type | Experimental | Experiments To demonstrate the usefulness of our HCL method, we evaluate it on several synthetic and real-world datasets, with a comparison to representative learning methods of TPPs. The following experimental results show the superiority of our HCL method, and we further do ablation studies and analyze the robustness of the method to its hyperparameters. All the experiments are run on a server with two Nvidia 3090 GPUs. |
| Researcher Affiliation | Academia | 1Gaoling School of Artificial Intelligence, Renmin University of China 2Beijing Key Laboratory of Big Data Management and Analysis Methods |
| Pseudocode | Yes | Algorithm 1: The MLE+HCL method of TPPs |
| Open Source Code | Yes | The code is available at https://github. com/qingmeiwangdaily/HCL TPP. |
| Open Datasets | Yes | Hawkes and Missing are synthetic data provided by (Mei, Qin, and Eisner 2019). Retweet (Zhao et al. 2015): The Retweet dataset... Stack Overflow: The Stack Overflow’s open source dataset... Book Order is collated from (Mei, Qin, and Eisner 2019). |
| Dataset Splits | Yes | Given a dataset and a backbone model, we train the model via a learning method with 5-fold cross validation. |
| Hardware Specification | Yes | All the experiments are run on a server with two Nvidia 3090 GPUs. |
| Software Dependencies | No | The paper mentions 'Adam (Kingma and Ba 2014)' as an optimizer but does not specify version numbers for any programming languages, libraries, or frameworks used in the implementation. |
| Experiment Setup | Yes | For all the methods, we set the learning rate as 0.001, the batch size as 1 for the Bookorder dataset... and 4 for the remaining datasets, and the dimension of sequence embedding as D = 64. ... For MLE+HCL, we set the number of negative sequences as K = 20... |