Abstract:Graph anomaly detection has long been an important problem in various domains pertaining to information security such as financial fraud, social spam and network intrusion. The majority of existing methods are performed in an unsupervised manner, as labeled anomalies in a large scale are often too expensive to acquire. However, the identified anomalies may turn out to be uninteresting data instances due to the lack of prior knowledge. In real-world scenarios, it is often feasible to obtain limited labeled anomalies, which have great potential to advance graph anomaly detection. However, the work exploring limited labeled anomalies and a large amount of unlabeled nodes in graphs to detect anomalies is relatively limited. Therefore, in this paper, we study an important problem of few-shot graph anomaly detection. Nonetheless, it is challenging to fully leverage the information of few-shot anomalous nodes due to the irregularity of anomalies and the overfitting issue in the few-shot learning. To tackle the above challenges, we propose a novel meta-learning based framework, MetaGAD, that learns to adapt the knowledge from self-supervised learning to few-shot supervised learning for graph anomaly detection. In specific, we formulate the problem as a bi-level optimization, ensuring MetaGAD converging to minimizing the validation loss, thus enhancing the generalization capacity. The comprehensive experiments on six real-world datasets with synthetic anomalies and "organic" anomalies (available in the datasets) demonstrate the effectiveness of MetaGAD in detecting anomalies with few-shot anomalies. The code is available at <a class="link-external link-https" href="https://github.com/XiongxiaoXu/MetaGAD" rel="external noopener nofollow">this https URL</a>.

LogAnMeta: Log Anomaly Detection Using Meta Learning

MetaLog: Generalizable Cross-System Anomaly Detection from Logs with Meta-Learning.

LogAnomaly: Unsupervised Detection of Sequential and Quantitative Anomalies in Unstructured Logs

OMLog: Online Log Anomaly Detection for Evolving System with Meta-learning

Natural Language Processing-based Model for Log Anomaly Detection

Learning to Diagnose: Meta-Learning for Efficient Adaptation in Few-Shot AIOps Scenarios

An Anomaly Detection Approach of Part-of-Speech Log Sequence Via Population Based Training

MLog: Mogrifier LSTM-based Log Anomaly Detection Approach Using Semantic Representation

End-to-End AutoML for Unsupervised Log Anomaly Detection

Machine Learning Based Anomaly Detection of Log Files Using Ensemble Learning and Self-Attention

OneLog: Towards End-to-End Training in Software Log Anomaly Detection

Log-based Anomaly Detection Without Log Parsing

LLMeLog: an Approach for Anomaly Detection Based on LLM-enriched Log Events

Detecting log anomaly using subword attention encoder and probabilistic feature selection

ConAnomaly: Content-Based Anomaly Detection for System Logs

MetaGAD: Meta Representation Adaptation for Few-Shot Graph Anomaly Detection

AutoLog: A Log Sequence Synthesis Framework for Anomaly Detection

Log-based Anomaly Detection of Enterprise Software: An Empirical Study

MoniLog: An Automated Log-Based Anomaly Detection System for Cloud Computing Infrastructures

LogFiT: Log Anomaly Detection Using Fine-Tuned Language Models

LogOnline: A Semi-Supervised Log-Based Anomaly Detector Aided with Online Learning Mechanism.