Fast and Accurate Anomaly Detection in Dynamic Graphs with a Two-Pronged Approach

Yoon, Minji; Hooi, Bryan; Shin, Kijung; Faloutsos, Christos

doi:10.1145/3292500.3330946

Cited by 70 publications

(53 citation statements)

References 29 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…NetWalk proposed clique embedding and reservoir sampling to quickly capture dynamically changing graph dimensions [55]. AnomRank classifies the types of anomalies into structural and node-relational anomalies and proposes a method to detect anomalies in two ways [56]. In addition, GAD approaches that incorporate graph clustering or community detection methods have been proposed for dynamic graphs [57,58].…”

Section: Graph-based Anomaly Detectionmentioning

confidence: 99%

Graph Representation Learning-Based Early Depression Detection Framework in Smart Home Environments

Kim

Sohn

2022

Sensors

View full text Add to dashboard Cite

Although the diagnosis and treatment of depression is a medical field, ICTs and AI technologies are used widely to detect depression earlier in the elderly. These technologies are used to identify behavioral changes in the physical world or sentiment changes in cyberspace, known as symptoms of depression. However, although sentiment and physical changes, which are signs of depression in the elderly, are usually revealed simultaneously, there is no research on them at the same time. To solve the problem, this paper proposes knowledge graph-based cyber–physical view (CPV)-based activity pattern recognition for the early detection of depression, also known as KARE. In the KARE framework, the knowledge graph (KG) plays key roles in providing cross-domain knowledge as well as resolving issues of grammatical and semantic heterogeneity required in order to integrate cyberspace and the physical world. In addition, it can flexibly express the patterns of different activities for each elderly. To achieve this, the KARE framework implements a set of new machine learning techniques. The first is 1D-CNN for attribute representation in relation to learning to connect the attributes of physical and cyber worlds and the KG. The second is the entity alignment with embedding vectors extracted by the CNN and GNN. The third is a graph extraction method to construct the CPV from KG with the graph representation learning and wrapper-based feature selection in the unsupervised manner. The last one is a method of activity-pattern graph representation based on a Gaussian Mixture Model and KL divergence for training the GAT model to detect depression early. To demonstrate the superiority of the KARE framework, we performed the experiments using real-world datasets with five state-of-the-art models in knowledge graph entity alignment.

show abstract

Section: Graph-based Anomaly Detectionmentioning

confidence: 99%

Graph Representation Learning-Based Early Depression Detection Framework in Smart Home Environments

Kim

Sohn

2022

Sensors

View full text Add to dashboard Cite

show abstract

“…(Jiang et al 2016) detects groups of nodes who form dense subgraphs in a temporally synchronized manner. • Anomalous event detection: ) detects sudden appearance of many unexpected edges, and (Yoon et al 2019) spots sudden changes in 1st and 2nd derivatives of PageRank. Anomaly detection in edge streams use as input a stream of edges over time.…”

Section: Related Workmentioning

confidence: 99%

Midas: Microcluster-Based Detector of Anomalies in Edge Streams

Bhatia

Hooi

Yoon

et al. 2020

AAAI

Self Cite

View full text Add to dashboard Cite

Given a stream of graph edges from a dynamic graph, how can we assign anomaly scores to edges in an online manner, for the purpose of detecting unusual behavior, using constant time and memory? Existing approaches aim to detect individually surprising edges. In this work, we propose Midas, which focuses on detecting microcluster anomalies, or suddenly arriving groups of suspiciously similar edges, such as lockstep behavior, including denial of service attacks in network traffic data. Midas has the following properties: (a) it detects microcluster anomalies while providing theoretical guarantees about its false positive probability; (b) it is online, thus processing each edge in constant time and constant memory, and also processes the data 108–505 times faster than state-of-the-art approaches; (c) it provides 46%-52% higher accuracy (in terms of AUC) than state-of-the-art approaches.

show abstract

“…The dataset may be characterized by one or more type of data properties. In order to give an insight into different data types used in anomaly detection tasks, there are a number of papers that summarize the main differences between the types and present anomaly detection approaches for them [45]- [54].…”

Section: Anomaly Ratiomentioning

confidence: 99%

Novel Meta-Features for Automated Machine Learning Model Selection in Anomaly Detection

et al. 2021

View full text Add to dashboard Cite

A growing number of research papers shed light on automated machine learning (AutoML) frameworks, which are becoming a promising solution for building complex machine learning models without human expertise and assistance. The key challenge in enabling AutoML frameworks to build an efficient model for anomaly detection tasks is to determine the best underlying model for a given task and optimization metric. The meta-learning approaches based on a set of meta-features that describes data properties can enable efficient model selection in AutoML frameworks. The existing meta-learning approaches based on statistical and information-theoretic meta-features require large amounts of data and computational resources to extract data properties. This paper proposes a novel set of meta-features for model selection in anomaly detection tasks based on domain-specific properties of data which overcomes the shortcomings of existing meta-features by introducing simple but effective meta-features that can be efficiently extracted or estimated by using a low amount of data. Experiments with 63 datasets from different repositories with varying schemas show that the proposed set of meta-features achieves an accuracy of 87% for model selection, while the achieved accuracy for simple meta-features is 74%, for statistical meta-features 68%, for information theory meta-feature 70%, and for a comprehensive set of meta-features by pyMFE 73%. This demonstrates that the proposed set can be adopted by AutoML frameworks across a diverse range of domains.

show abstract

Fast and Accurate Anomaly Detection in Dynamic Graphs with a Two-Pronged Approach

Cited by 70 publications

References 29 publications

Graph Representation Learning-Based Early Depression Detection Framework in Smart Home Environments

Graph Representation Learning-Based Early Depression Detection Framework in Smart Home Environments

Midas: Microcluster-Based Detector of Anomalies in Edge Streams

Novel Meta-Features for Automated Machine Learning Model Selection in Anomaly Detection

Contact Info

Product

Resources

About