Time-Series Event Prediction with Evolutionary State Graph

Hu, Wenjie; Yang, Yang; Cheng, Ziqiang; Yang, Carl; Ren, Xiang

doi:10.1145/3437963.3441827

Cited by 26 publications

(20 citation statements)

References 14 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…However, such a distinction is less relevant when events are external to the data [7]. Recently in [16], the authors apply a dynamic graph to identify events on time series. We focus on the detection of events with graphs as inputs.…”

Section: Related Workmentioning

confidence: 99%

Event Detection on Dynamic Graphs

Kosan¹,

Silva²,

Medya³

et al. 2021

Preprint

View full text Add to dashboard Cite

Event detection is a critical task for timely decisionmaking in graph analytics applications. Despite the recent progress towards deep learning on graphs, event detection on dynamic graphs presents particular challenges to existing architectures. Real-life events are often associated with sudden deviations of the normal behavior of the graph. However, existing approaches for dynamic node embedding are unable to capture the graph-level dynamics related to events.In this paper, we propose DyGED, a simple yet novel deep learning model for event detection on dynamic graphs. DyGED learns correlations between the graph macro dynamics-i.e. a sequence of graph-level representations-and labeled events. Moreover, our approach combines structural and temporal self-attention mechanisms to account for application-specific node and time importances effectively. Our experimental evaluation, using a representative set of datasets, demonstrates that DyGED outperforms competing solutions in terms of event detection accuracy by up to 8.5% while being more scalable than the top alternatives. We also present case studies illustrating key features of our model.

show abstract

Section: Related Workmentioning

confidence: 99%

Event Detection on Dynamic Graphs

Kosan¹,

Silva²,

Medya³

et al. 2021

Preprint

View full text Add to dashboard Cite

show abstract

“…Recently, GNNs have achieved stateof-the-art performance on many tasks [16,31,32,40]. Based on the Message-Passing framework, many works have expanded the capabilities to tackle different types of graph, such as heterogeneous graphs [14] and dynamic graphs [13,45]. These improvements make GNNs better compatible for purchase prediction in real-world applications [6,27].…”

Section: Related Workmentioning

confidence: 99%

How Powerful are Interest Diffusion on Purchasing Prediction

Huang

Yang

Cheng

et al. 2021

Proceedings of the 44th International ACM SIGIR Conference on Research and Development in Information Retrieval

Self Cite

View full text Add to dashboard Cite

A Taocode is a kind of specially coded text-link on Taobao.com (the world's biggest online shopping website), through which users can share messages about products with each other. Analyzing Taocodes can potentially facilitate understanding of the social relationships between users and, more excitingly, their online purchasing behaviors under the influence of Taocode diffusion. This paper innovatively investigates the problem of online purchasing predictions from an information diffusion perspective, with Taocode as a case study. Specifically, we conduct profound observational studies on a large-scale real-world dataset from Taobao, containing over 100M Taocode sharing records. Inspired by our observations, we propose InfNet, a dynamic GNN-based framework that models the information diffusion across Taocode. We then apply InfNet to item purchasing predictions. Extensive experiments on real-world datasets validate the effectiveness of InfNet compared with 8 stateof-the-art baselines.

show abstract

“…Previous works such as graph structural learning provide a feasible solution to constructing a static latent graph from time-series, but to the best of our knowledge, no previous work covers how to learn a dynamic and explainable graph for multivariate anomaly detection. Our work is inspired by the recent progress of evolutionary event graph [2,6] where the nodes in the graph represent the time-sequences segments (events) and directed links represent the transition of the segments (events). Compare to previous works, this line of research naturally models the timevarying relations among time-series states via dynamic connections, and each state carries a physical meaning that is understandable by human.…”

Section: Introductionmentioning

confidence: 99%

“…Compare to previous works, this line of research naturally models the timevarying relations among time-series states via dynamic connections, and each state carries a physical meaning that is understandable by human. However, one major limitation of [6] is that the event nodes employed in this work capture the information across all the time series. Assume there are 𝐾 segment patterns in each time series, and the number of time series is 𝐷.…”

Section: Introductionmentioning

confidence: 99%

“…The model would need 𝐾 𝐷 number of event nodes to represent the multivariate signals. This exponential number of event nodes strongly limits the information processing capability of the evolutionary state graph and therefore allows only a very small number of time series (𝐷) to be analyzed in practice (four in the dataset used in the paper [6]).…”

Section: Introductionmentioning

confidence: 99%

See 1 more Smart Citation

Event2Graph: Event-driven Bipartite Graph for Multivariate Time-series Anomaly Detection

Wu,

Gu,

Wang

et al. 2021

Preprint

View full text Add to dashboard Cite

Modeling inter-dependencies between time-series is the key to achieve high performance in anomaly detection for multivariate time-series data. The de-facto solution to model the dependencies is to feed the data into a recurrent neural network (RNN). However, the fully connected network structure underneath the RNN (either GRU or LSTM) assumes a static and complete dependency graph between time-series, which may not hold in many real-world applications. To alleviate this assumption, we propose a dynamic bipartite graph structure to encode the inter-dependencies between time-series. More concretely, we model time series as one type of nodes, and the time series segments (regarded as event) as another type of nodes, where the edge between two types of nodes describe a temporal pattern occurred on a specific time series at a certain time. Based on this design, relations between time series can be explicitly modelled via dynamic connections to event nodes, and the multivariate time-series anomaly detection problem can be formulated as a self-supervised, edge stream prediction problem in dynamic graphs. We conducted extensive experiments to demonstrate the effectiveness of the design.

show abstract

Time-Series Event Prediction with Evolutionary State Graph

Cited by 26 publications

References 14 publications

Event Detection on Dynamic Graphs

Event Detection on Dynamic Graphs

How Powerful are Interest Diffusion on Purchasing Prediction

Event2Graph: Event-driven Bipartite Graph for Multivariate Time-series Anomaly Detection

Contact Info

Product

Resources

About