Learning Granger Causality for Hawkes Processes

Authors: Hongteng Xu, Mehrdad Farajtabar, Hongyuan Zha

ICML 2016 | Conference PDF | Archive PDF | Plain Text | LLM Run Details

Reproducibility Variable Result LLM Response
Research Type Experimental Experiments on both synthetic and real-world data show that our method can learn the Granger causality graph and the triggering patterns of the Hawkes processes simultaneously.
Researcher Affiliation Academia Hongteng Xu HXU42@GATECH.EDU School of ECE, Georgia Institute of Technology; Mehrdad Farajtabar MEHRDAD@GATECH.EDU College of Computing, Georgia Institute of Technology; Hongyuan Zha ZHA@CC.GATECH.EDU College of Computing, Georgia Institute of Technology
Pseudocode Yes Algorithm 1 Learning Hawkes Processes (MLE-SGLP); Algorithm 2 Selecting basis functions
Open Source Code No The paper does not contain any statements or links indicating that open-source code for the methodology is provided.
Open Datasets Yes We test our algorithm on the IPTV viewing record data set (Luo et al., 2015).
Dataset Splits No The paper mentions training and testing sets ('C = {50, ..., 250} sequences are chosen randomly as training set while the rest 250 sequences are chosen as testing set.') but does not explicitly describe a validation set or its split.
Hardware Specification No The paper mentions a 'PC with 16GB memory' in the context of a competitor's algorithm running out of memory, not as a specification for the hardware used in their own experiments.
Software Dependencies No The paper does not provide specific software names with version numbers that would be needed to replicate the experiment.
Experiment Setup Yes We set αS = 10, αG = 100, αP = 1000. In all trials, Gaussian basis functions are used, whose number and bandlimit are decided by Algorithm 2. We set the time length of impact function to be 8 days... and the number of samples M = 576...