ABSTRACT
Temporal point processes have many applications, from crime forecasting to modeling earthquake aftershocks sequences. Due to the flexibility and expressiveness of deep learning, neural network-based approaches have recently shown promise for modeling point process intensities. However, there is a lack of research on the robustness of such models in regards to adversarial attacks and natural shocks to systems. Precisely, while neural point processes may outperform simpler parametric models on in-sample tests, how these models perform when encountering adversarial examples or sharp non-stationary trends remains unknown. Current work proposes several white-box and blackbox adversarial attacks against temporal point processes modeled by deep neural networks. Extensive experiments confirm that predictive performance and parametric modeling of neural point processes are vulnerable to adversarial attacks. Additionally, we evaluate the vulnerability and performance of these models in the presence of non-stationary abrupt changes, using the crimes dataset, during the Covid-19 pandemic, as an example. © 2022 IEEE.
ABSTRACT
Two models that capture the spread of infectious diseases, the Hawkes point process model and the SEIR compartmental model, are compared with regard to their use in modeling the COVID-19 pandemic. The physical plausibility of the SEIR model is weighed against the parsimony and flexibility of the Hawkes model. The mathematical connection between Hawkes and SEIR models is described.
ABSTRACT
This chapter presents a novel point process model for COVID-19 transmission—the multivariate recursive Hawkes process, which is an extension of the recursive Hawkes model to the multivariate case. Equivalently the model can be viewed as an extension of the multivariate Hawkes model to allow for varying productivity as in the recursive model. Several theoretical properties of this process are stated and proved, including the existence of the multivariate recursive counting process and formulas for the mean and variance. EM-based algorithms are explored for estimating parameters of parametric and semi-parametric forms of the model. Additionally, an algorithm is presented to reconstruct the process from imprecise event times. The performance of the algorithms on both synthetic and real COVID-19 data sets is illustrated through several experiments. © 2022, The Author(s), under exclusive license to Springer Nature Switzerland AG.
ABSTRACT
Detecting the source of an outbreak cluster during a pandemic like COVID-19 can provide insights into the transmission process, associated risk factors, and help contain the spread. In this work we study the problem of source detection from multiple snapshots of spreading on an arbitrary network structure. We use a spatial temporal graph convolutional network based model (SD-STGCN) to produce a source probability distribution, by fusing information from temporal and topological spaces. We perform extensive experiments using popular compartmental simulation models over synthetic networks and empirical contact networks. We also demonstrate the applicability of our approach with real COVID-19 case data.