4th Workshop on Automated Spatial and Temporal Anomaly Detection (ASTAD)
Overview
ASTAD brings together researchers and practitioners working on AI methods for detecting, predicting, explaining, and responding to anomalies, abnormalities, outliers, risks, failures, and deviations from expected or learned normal behavior across spatial, temporal, and multimodal data. The workshop welcomes methodological advances as well as real-world applications in industry, IT systems, finance, healthcare, and beyond, spanning anomaly detection and prediction, out-of-distribution and outlier detection, fraud detection, predictive maintenance, risk analysis and assessment, fault and failure detection/prediction, accident and incident detection/prediction, early warning systems, and abnormal behavior recognition.
Call for Papers
We invite researchers and practitioners to submit their original research contributions to the 4th Workshop on Automated Spatial and Temporal Anomaly Detection (ASTAD), held as part of AAAI-27. Topics include, but are not limited to:
- Anomaly, abnormality, novelty, outlier, and out-of-distribution detection
- Anomaly and abnormal-event prediction / forecasting
- Predictive maintenance, fault diagnosis, and failure prediction in industrial and IT systems
- Risk analysis, risk prediction, and early-warning systems in healthcare, finance, and infrastructure
- Fraud detection and financial anomaly detection
- Accident, incident, and hazard detection and prediction
- Detection and prediction of deviations from expected or learned normal behavior
- Root-cause analysis, diagnosis, localization, and attribution across clinical, industrial, and IT settings
- Foundation models, LLMs, VLMs, and generalist models for detection and prediction
- Agentic AI for monitoring, diagnosis, prediction, and autonomous response
- Generative, self-supervised, unsupervised, few-shot, and zero-shot approaches
- Explainable, interpretable, causal, and reasoning-based methods
- Multimodal, spatial, temporal, spatiotemporal, graph, image, video, signal, and tabular data
- Real-time, streaming, continual, and on-device methods
- Datasets, benchmarks, evaluation, uncertainty, robustness, and distribution shift
- Applications in healthcare, manufacturing, IT operations, automotive, transportation, robotics, cybersecurity, remote sensing, infrastructure, energy, finance, and safety-critical systems
Submission Requirements
We welcome original research as full papers of up to 7 pages, plus 2 additional pages for references only. Submissions must use the official AAAI-27 author kit and will undergo double-blind peer review.
All submissions will be handled electronically via OpenReview. Only PDF files are accepted.
Submission Site: OpenReview
Format
ASTAD is a one-day event combining paper presentations, invited talks from leading researchers, and interactive poster sessions, with ample time for Q&A and discussion.
Attendance
Attendance is open to all registered AAAI-27 participants, including researchers, students, and industry professionals.
Important Dates
- Paper submission deadline: November 20, 2026
- Author notification: December 2, 2026
- Camera-ready deadline: To be announced
- Workshop Date: February 22–23, 2027
- Workshop Location: Montréal, Canada
Schedule
The workshop schedule will be announced soon.
Speakers
Prof. Leman Akoglu
Heinz College of Information Systems and Public Policy, Carnegie Mellon University
Dr. Akoglu is the Heinz College Dean’s Associate Professor at Carnegie Mellon University’s Heinz College of Information Systems and Public Policy. She also hold courtesy appointments at the Machine Learning Department (MLD) and the Computer Science Department (CSD) of School of Computer Science (SCS). At Heinz, she directs the Data Analytics Techniques Algorithms (DATA) Lab. Her research interests are broadly in data mining, graph mining, machine learning, and knowledge discovery, with specific focus on anOmaLiEs—identifying and characterizing ‘what stands out’ in large-scale, time-varying, multi-modal data sources through scalable computational methods.
Prof. Yue Zhao
Department of Computer Science, University of Southern California
Yue Zhao is an Assistant Professor of Computer Science at the University of Southern California, where he leads the FORTIS Lab and serves as Associate Co-Director of the Institute on Ethics and Trust in Computing. His research focuses on auditable AI systems, connecting anomaly detection with the evaluation, security, and efficiency of AI agents. He develops open-source tools and benchmarks, including PyOD, ADBench, and TrustLLM. He received his Ph.D. from Carnegie Mellon University in 2023. His honors include multiple Amazon Research Awards, the NVIDIA Academic Grant Program Award, and selection for AAAI New Faculty Highlights. He serves as an Action Editor for Transactions on Machine Learning Research and an Associate Editor for IEEE Transactions on Neural Networks and Learning Systems.
Dr. Samet Akcay
Principal AI Research Engineer, Intel Corporation
Samet Akcay is a Principal Engineer at Intel, within the AI Models and Robotics organization in Edge Computing Group. He provides technical leadership and architecture for machine learning, deep learning, and robot learning, with a focus on advancing Vision AI and Physical AI technologies for intelligent systems at the edge. He is the creator of Anomalib, a widely adopted open-source framework for anomaly detection. His research spans computer vision, anomaly detection, multimodal and physical AI, with over 40 publications and more than 6,000 citations. Samet holds a PhD in Computer Science from Durham University, UK, and an MSc in Electrical Engineering from Pennsylvania State University, USA. His current work focuses on advancing Vision AI, multimodal learning, and Physical AI, bridging cutting-edge research with practical, scalable solutions for real-world applications in computer vision, robotics, and edge computing.
More keynote speakers to be announced soon!
Workshop Poster
FAQ
Accepted papers from the previous edition of ASTAD will appear in archival proceedings. We intend to publish this year's accepted papers in archival proceedings as well; however, this has not yet been confirmed. Any updates will be posted on this page.
No. ASTAD does not allow dual submission. A paper submitted to ASTAD must not be under review at, or accepted by, any other venue. If ASTAD papers are published in archival proceedings, your accepted paper may not be resubmitted elsewhere in the same form, but substantially extended versions are allowed and encouraged. Otherwise, you are free to submit it to another venue afterwards.
ASTAD follows the attendance policy of the main AAAI conference: at least one author of each accepted paper must register for the workshop and present the paper in person.
No. All presentations must be given in person.
Email us and include a link to your OpenReview profile.
Organizers
Narges Armanfard (Chair)
Associate Professor - narges.armanfard@mcgill.ca
Narges Armanfard is an Associate Professor of Electrical and Computer Engineering at McGill University and Mila - Quebec AI Institute. She is also affiliated with McGill Centre for Intelligent Machines (CIM), McGill initiative in Computational Medicine (MiCM) and McGill Institute for Aerospace Engineering (MIAE). Dr. Armanfard is the founder and principal investigator of the iSMART Lab with the mission to pioneer advanced algorithms in artificial intelligence, with expertise in computer vision, time series analysis, tabular data, large language models, and visual language models.
Thi Kieu Khanh Ho
AI Researcher (Postdoctoral; Lab member since 2021) - thi.k.ho@mail.mcgill.ca
Khanh received her PhD degree in AI at McGill University and her Master’s degree at Gwangju Institute of Science and Technology (GIST), South Korea. She was a researcher at Seoul National University Hospital (SNUH), South Korea, before moving to Montreal for her PhD. She has received the prestigious McGill Engineering Doctoral Award (MEDA), the GREAT Award, and the highly prestigious Vanier Scholarship and Chwang Seto Award. Her research focuses on time-series data analysis using generative models and graphs.
Thomas Lai
AI Researcher (PhD; Lab member since 2023) - thomas.lai@mail.mcgill.ca
Thomas is a PhD student at the iSMART Lab, McGill University and Mila - Quebec AI Institute. He received his B.Eng. in Bioengineering from McGill University. His research develops deep learning algorithms and foundation models for time-series analysis, with a focus on open-set anomaly detection, with work published at ECAI. He also works on applications in brain signal analysis and industrial automation, and is co-first author of EpiiSLM, developed with clinical collaborators at the MNI. He has received the NSERC Postgraduate Scholarship–Doctoral (PGS-D) and the Vadasz Fellowship. He was a co-organizer of the ASTAD workshop at WACV 2025.
