Conjugate Bayesian Two-step Change Point Detection for Hawkes Process

Authors: Zeyue Zhang, Xiaoling LU, Feng Zhou

NeurIPS 2024 | Conference PDF | Archive PDF | Plain Text | LLM Run Details

Reproducibility Variable Result LLM Response
Research Type Experimental Extensive experiments on both synthetic and real data demonstrate the superior effectiveness and efficiency of our method compared to baseline methods.
Researcher Affiliation Academia Zeyue Zhang1,2, Xiaoling Lu1,2, Feng Zhou1,3 1Center for Applied Statistics and School of Statistics, Renmin University of China 2Innovation Platform, Renmin University of China 3Beijing Advanced Innovation Center for Future Blockchain and Privacy Computing
Pseudocode Yes Algorithm 1 Gibbs Sampler
Open Source Code Yes Our code is publicly available at https://github.com/Aurora2050/Co Bay-CPD.
Open Datasets Yes Wanna Cry Cyber Attack3 [4]: The Wanna Cry virus infected more than 200,000 computers around the world in 2017 and received much attention. The Wanna Cry Cyber Attack data contains 208 traffic logs information observations. Each observation contains the relevant timestamp. 3https://www.malware-traffic-analysis.net/2017/05/18/index2.html. NYC Vehicle Collisions4 [27]: The New York City vehicle collision dataset comprises approximately 1.05 million vehicle collision records, each containing information about the time and location of the collision. For our experiments, we select the records from Oct.14th, 2017. 4https://data.cityofnewyork.us/Public-Safety/NYPD-Motor-Vehicle-Collisions/h9gi-nx95
Dataset Splits Yes In experiments, we select all hyperparameters through cross-validation.
Hardware Specification No The paper does not provide specific hardware details (e.g., CPU/GPU models, memory, or cloud resources) used for running its experiments. It only reports running time in minutes.
Software Dependencies No The paper does not provide specific software dependencies or versions (e.g., library names with version numbers) needed to replicate the experiment.
Experiment Setup Yes For Co Bay-CPD, we adopt a prior distribution p(w) = N(w|0, K), where K = 0.5I. The detection outcomes are presented in Fig. 1a in Appendix D.2. Furthermore, the estimated parameter λ from Co Bay-CPD for the synthetic data is depicted in Fig. 1b in Appendix D.2.