Collective Anomaly

Collective anomaly detection focuses on identifying unusual patterns within groups or subsequences of data, where individual elements might appear normal. Current research emphasizes developing advanced machine learning models, including tree-based multi-instance learning frameworks and transformer architectures like BERT, to effectively capture these complex patterns in various data types such as time series and trajectories. This research is crucial for improving anomaly detection in diverse fields, ranging from cybersecurity and network monitoring to healthcare and industrial process control, by enabling more accurate and timely identification of significant events that traditional methods might miss. Furthermore, efforts are underway to improve the explainability and root cause identification capabilities of these models.

Papers

January 20, 2024

TreeMIL: A Multi-instance Learning Framework for Time Series Anomaly Detection with Inexact Supervision
Chen Liu, Shibo He, Haoyu Liu, Shizhong Li
Time Series Multiple Instance Learning Unlabeled Anomaly Anomaly Score Pseudo Supervision Collective Anomaly

December 19, 2023

Collective Anomaly Perception During Multi-Robot Patrol: Constrained Interactions Can Promote Accurate Consensus
Zachary R. Madin, Jonathan Lawry, Edmund R. Hunt
Multi Robot Multi Robot System Collective Perception Group Consensus Consensus Formation Patrolling Strategy Collective Anomaly

March 17, 2023

GADformer: A Transparent Transformer Model for Group Anomaly Detection on Trajectories
Andreas Lohrer, Darpan Malik, Claudius Zelenka, Peer Kröger
Transformer Model Handwritten Trajectory Collective Anomaly

March 7, 2023

Root Cause Identification for Collective Anomalies in Time Series given an Acyclic Summary Causal Graph with Loops
Charles K. Assaad, Imad Ez-zejjari, Lei Zan
Time Series Causal Graph Causal Relation Root Cause Level Loop Collective Anomaly Summary Causal Graph

September 28, 2022

Big data analysis and distributed deep learning for next-generation intrusion detection system optimization
Khloud Al Jallad, Mohamad Aljnidi, Mohammad Said Desouki
Deep Learning Anomaly Detection Big Data Deep Recurrent Neural Network Contextual Attack Collective Anomaly

Collective Anomaly

Papers

TreeMIL: A Multi-instance Learning Framework for Time Series Anomaly Detection with Inexact Supervision

Collective Anomaly Perception During Multi-Robot Patrol: Constrained Interactions Can Promote Accurate Consensus

GADformer: A Transparent Transformer Model for Group Anomaly Detection on Trajectories

Root Cause Identification for Collective Anomalies in Time Series given an Acyclic Summary Causal Graph with Loops

Big data analysis and distributed deep learning for next-generation intrusion detection system optimization