Semi-supervised Anomaly Detection on Attributed Graphs

Kumagai, Atsutoshi; Iwata, Tomoharu; Fujiwara, Yasuhiro

doi:10.48550/arxiv.2002.12011

Cited by 4 publications

(11 citation statements)

References 28 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…Graph-based anomaly detection has drawn great attention in recent years [2], especially the recently emerged GNN-based approaches [8,13,16,41,51], but most of the studies focus on anomaly (e.g., anomalous nodes or edges) detection in a single large graph. Below we review related work on GAD.…”

Section: Related Work 21 Graph-level Anomaly Detectionmentioning

confidence: 99%

“…We aim to train a detection model that can detect these two types of abnormal graphs. Note that the detection of locally-anomalous graphs is different from anomalous node detection in [8,13,16,41] because the former is to detect graphs by evaluating the nodes/edges across a set of independent and separate graphs while the latter is to detect nodes/edges given a set of dependent nodes and edges from a single graph.…”

Section: Framework 31 Problem Statementmentioning

confidence: 99%

“…anomaly detection in graph data [8,13,16,41,51], but they focus on anomalous node/edge detection in a single large graph.…”

Section: Introductionmentioning

confidence: 99%

See 2 more Smart Citations

Deep Graph-level Anomaly Detection by Glocal Knowledge Distillation

Pang

Chen

et al. 2022

Proceedings of the Fifteenth ACM International Conference on Web Search and Data Mining

View full text Add to dashboard Cite

Graph-level anomaly detection (GAD) describes the problem of detecting graphs that are abnormal in their structure and/or the features of their nodes, as compared to other graphs. One of the challenges in GAD is to devise graph representations that enable the detection of both locallyand globally-anomalous graphs, i.e., graphs that are abnormal in their fine-grained (node-level) or holistic (graph-level) properties, respectively. To tackle this challenge we introduce a novel deep anomaly detection approach for GAD that learns rich global and local normal pattern information by joint random distillation of graph and node representations. The random distillation is achieved by training one GNN to predict another GNN with randomly initialized network weights. Extensive experiments on 16 real-world graph datasets from diverse domains show that our model significantly outperforms seven state-of-the-art models. Code and datasets are available at https://git.io/GLocalKD.

show abstract

Section: Related Work 21 Graph-level Anomaly Detectionmentioning

confidence: 99%

Section: Framework 31 Problem Statementmentioning

confidence: 99%

See 1 more Smart Citation

Deep Graph-level Anomaly Detection by Glocal Knowledge Distillation

Pang

Chen

et al. 2022

Proceedings of the Fifteenth ACM International Conference on Web Search and Data Mining

View full text Add to dashboard Cite

show abstract

“…Besides the widely used MLP based encoder and decoders [3,28,41], convolutional neural networks (CNN) are applied on grid data [6], and recurrent neural networks (RNN) are applied on time series data [25,31]. Recently, GCN-based autoencoders have been used for anomaly detection on graph data [8,10,[20][21][22]50], with an important emphasis node-level anomalies. In comparison, our work aims at graph-level anomaly detection, since each graph corresponds to the network wide traffic condition at a specific time instant that may be disturbed by a large scale event.…”

Section: Autoencoder-based Anomaly Detectionmentioning

confidence: 99%

Graph Convolutional Networks for traffic anomaly

Hu,

Qu,

Work

2020

Preprint

View full text Add to dashboard Cite

Event detection has been an important task in transportation, whose task is to detect points in time when large events disrupts a large portion of the urban traffic network. Travel information Origin-Destination (OD) matrix data by map service vendors has large potential to give us insights to discover historic patterns and distinguish anomalies. However, to fully capture the spatial and temporal traffic patterns remains a challenge, yet serves a crucial role for effective anomaly detection. Meanwhile, existing anomaly detection methods have not well-addressed the extreme data sparsity and high-dimension challenges, which are common in OD matrix datasets. To tackle these challenges, we formulate the problem in a novel way, as detecting anomalies in a set of directed weighted graphs representing the traffic conditions at each time interval. We further propose Context augmented Graph Autoencoder (Con-GAE ), that leverages graph embedding and context embedding techniques to capture the spatial traffic network patterns while working around the data sparsity and high-dimensionality issue. Con-GAE adopts an autoencoder framework and detect anomalies via semi-supervised learning. Extensive experiments show that our method can achieve up can achieve a 0.1-0.4 improvements of the area under the curve (AUC) score over state-of-art anomaly detection baselines, when applied on several real-world large scale OD matrix datasets.

show abstract

“…Recent advancements in the area of machine learning on graphs led to impressive results in solving the specified problems by applying the graph convolutional network (GCN) framework (cf. [31], [32], [33]). However, to extract necessary information and provide substantial results, the neural network needs the graph data to be constructed as a set of features (called "embeddings") without neglecting the relational structure and corresponding attributes.…”

Section: How Can We Specify Anomaly Detection In Terms Of Network Mon...mentioning

confidence: 99%

Statistical learning for change point and anomaly detection in graphs

Malinovskaya¹,

Otto²,

Peters³

2020

Preprint

View full text Add to dashboard Cite

Complex systems which can be represented in the form of static and dynamic graphs arise in different fields, e.g. communication, engineering and industry. One of the interesting problems in analysing dynamic network structures is to monitor changes in their development. Statistical learning, which encompasses both methods based on artificial intelligence and traditional statistics, can be used to progress in this research area. However, the majority of approaches apply only one or the other framework. In this paper, we discuss the possibility of bringing together both disciplines in order to create enhanced network monitoring procedures focussing on the example of combining statistical process control and deep learning algorithms. Together with the presentation of change point and anomaly detection in network data, we propose to monitor the response times of ambulance services, applying jointly the control chart for quantile function values and a graph convolutional network.

show abstract

Semi-supervised Anomaly Detection on Attributed Graphs

Cited by 4 publications

References 28 publications

Deep Graph-level Anomaly Detection by Glocal Knowledge Distillation

Deep Graph-level Anomaly Detection by Glocal Knowledge Distillation

Graph Convolutional Networks for traffic anomaly

Statistical learning for change point and anomaly detection in graphs

Contact Info

Product

Resources

About