Distribution Shift Detection

Ask about this topic

Overview

Distribution shift detection studies whether incoming data, model inputs, or model behavior differ from a reference, and whether those changes threaten reliable predictions. Methods range from statistical tests such as Kolmogorov–Smirnov, maximum mean discrepancy, and change-point scans to representation-based OOD scores, conformal tests, and online betting martingales. Work spans image and language OOD benchmarks, time-series and tabular streams, and deployed systems, with growing attention to locating affected subgroups and triggering adaptation only when needed. A central challenge is balancing sensitivity and detection speed against false alarms, especially under high-dimensional shifts, changing data sources, and shifts that are detectable but harmless—or harmful yet difficult to detect.

Latest papers

CardsList

63 papers13 papers in the last four weeks, up 333% on the four weeks before. 0.1% of all new papers.

  1. Detecting Fine-Grained Differences between Image Datasets

    Oct 28, 2026Leander Girrbach, Alexander von Recum, Kirill Bykov +1Distribution Shift Detection

    Technical University of Munich, Munich Center for Machine Learning (MCML) · Helmholtz Munich · Ludwig Maximilian University of Munich

  2. COMET: A coordinated multi-layer runtime monitoring framework for human-supervised AI-enabled safety systems

    Oct 21, 2026Niclas Flehmig, Mary Ann Lundteigen, Shen YinAI SafetyAI Agent Monitoring

    Department of Mechanical and Industrial Engineering, Norwegian University of Science and Technology, Trondheim, Norway · Department of Cybernetics Engineering, Norwegian University of Science and Technology, Trondheim, Norway

  3. No Frames, Just Frequencies: Bandwidth-Efficient Deployment-Time Adaptation for Smart Glasses

    Oct 21, 2026Francesco Corti, Moritz Becker, Olga SaukhDomain AdaptationTest-Time Adaptation

    Graz University of Technology, Graz, Austria · Graz University of Technology, Complexity Science Hub, Graz, Vienna, Austria

  4. Information-Geometric Blind Spots of Conformal Test Martingales

    Oct 16, 2026Johan Hallberg SzabadváryDistribution Shift DetectionConformal Prediction

    Department of Mathematics, Stockholm University · Department of Computing, Jönköping University

  5. Diagnosing Low-Confidence Patient Profiles via Localized Harmful Covariate Shift Detection

    Oct 13, 2026Olivier Lefebvre, Hakima Laribi, Dan Poenaru +1Distribution Shift DetectionClinical Prediction

    Department of Computer Science, Université de Sherbrooke, Sherbrooke, Canada · Department of Pediatric Surgery, Montreal Children's Hospital, Montreal, Canada · Centre for Outcomes Research and Evaluation +1

  6. Learning to Detect Shifts in Unstructured Data Streams via Monitoring-by-Betting

    Oct 12, 2026Siyu Zhou, Jialin Ouyang, Fan Xia +8Distribution Shift DetectionChange-Point Detection

    University of California, San Francisco · University of Maryland · Johns Hopkins University

  7. Monitoring for Shifts at Unknown Times: Parameter-Free Conditional Conformal Martingales

    Oct 12, 2026Tuo Liu, Drew Prinster, Francesco OrabonaDistribution Shift DetectionChange-Point Detection

    King Abdullah University of Science and Technology, Saudi Arabia · Johns Hopkins University, USA

  8. Unsupervised Data Drift Monitoring for Deployed Deepfake Speech Detectors

    Oct 12, 2026Lukáš Prokeš, Vojtěch Staněk, Anton Firc +1Distribution Shift DetectionUnsupervised Anomaly Detection

  9. Does the Explanation Drift Before the Guarantee? Label-Free Early Warning on Evolving Graphs

    Oct 12, 2026Tomasz Szandała, Damian RaczkowskiDistribution Shift DetectionGNN Explainability

    Faculty of Information and Communication Technology, Wroclaw University of Science and Technology · independent researcher

  10. Beyond Marginal Monitoring: Distributed Joint-Distribution Testing for Data Concept Drift in Large Scale E-Commerce Operations

    Oct 6, 2026Cagdas Pullu, Mahmut Emir Arslan, Bugra Balkac +4Two-Sample TestingConcept Drift

    Data & Analytics, Trendyol Group, Istanbul, 34485, Istanbul, Türkiye. · Department of Data Science and Analytics, Istanbul Technical University, Maslak, Istanbul, 34469, Istanbul, Türkiye.

  11. The Blind Spot Paradox: When Adaptive Classifiers Defeat Drift Detectors

    Oct 5, 2026Raphaël Minato, Fabrice Popineau, Arpad Rimmel +1Concept DriftDistribution Shift Detection

    Université Paris-Saclay, CNRS, Laboratoire Interdisciplinaire des Sciences du Numérique, 91190, Gif-sur-Yvette, France

  12. Zero Flux: Flow-Based Comparison of High-Dimensional Discrete Distributions

    Oct 1, 2026Leyang Wang, Yakun Wang, Song Liu +1Distribution ShiftDistribution Shift Detection

    Department of Mathematical Informatics, The University of Tokyo, Japan · Center for Advanced Intelligence Project, RIKEN, Japan · School of Mathematics, University of Bristol, United Kingdom

  13. Dataset Identity, Not Novelty: The Source of an Inflated OOD Detection Gain

    Oct 1, 2026Donghoon Lee, Shinjin KangDistribution Shift DetectionOOD Detection

    Hongik University

  14. When to Adapt: Multi-Signal Domain Shift Detection for Efficient Training-Free Adaptation in Open-Vocabulary Segmentation

    Sep 29, 2026Michele Antonazzi, Alejandra C. Hernandez, José Araujo +2Distribution ShiftContinual Test-Time Adaptation

    EECS and Digital Futures, KTH Royal Institute of Technology, Stockholm, Sweden · Ericsson Research, Ericsson AB, Stockholm, Sweden

  15. Behavioral Monitoring of JEPA World Models with Jacobian Centroids

    Sep 27, 2026Thomas Walker, Randall Balestriero, Richard BaraniukDistribution ShiftWorld Model-Based Planning

    Rice University · Brown University

  16. Towards Universal Representation-Based Process Control

    Sep 25, 2026Jinmyeong Choi, Taesup Kim, Artur DubrawskiDistribution Shift Detection

    Auton Lab, Robotics Institute, Carnegie Mellon University, Pittsburgh, USA · Graduate School of Data Science, Seoul National University, Seoul, Korea

  17. A Decision-Support Audit Protocol for Supervision Drift in Proxy-Labeled Credit-Risk Prediction

    Sep 14, 2026Mehrdad Shoeibi, Muhammad Shabanpour, Waldemar Karwowski +1AI Risk ManagementAlgorithmic Auditing

    Department of Industrial Engineering and Management Systems University of Central Florida, Orlando, FL, USA · Department of Economics, Northeastern University, Boston, MA, USA

  18. Deciding When to Decide: Testing Operational Suboptimality Under Distributional Shift

    Aug 29, 2026Minxing Zheng, Holly Wiberg, Shixiang ZhuDistribution Shift DetectionStochastic Optimization

    Carnegie Mellon University

  19. RankShift: In-Database Detection and Explanation of Categorical Shifts

    Aug 28, 2026Omair Shafi AhmedDistribution Shift DetectionOnline Anomaly Detection

    Microsoft

  20. Change Detection in Probability Flow ODE: Online Testing in Diffusion Latent Spaces

    Aug 24, 2026Artem Kraevskiy, Artem ProkhorovDistribution Shift DetectionChange-Point Detection

    National Research University Higher School of Economics Moscow, Russia · The University of Sydney Business School & CIREQ & CEBDA Sydney, Australia

  21. Separating Covariate Shift from Mechanism Change with Two Discriminators: CJSD, a Conditional Discrepancy with an Exact Covariate-Concept Decomposition

    Aug 20, 2026Kentaro OdaDistribution Shift DetectionCovariate Shift

    Center for Management of Information Technologies, Kagoshima University

  22. Performance Drift Detection in Machine Learning as a Service (MLaaS) for IoT Environments

    Aug 19, 2026Deepak Kanneganti, Sajib Mistry, Sheik Mohammad Mostakim Fattah +3Distribution Shift DetectionAnomaly Detection

    School of Electrical Engineering, Computing and Mathematical Sciences (EECMS), Curtin University, Perth, WA, Australia · Department of Computing Science, Umeå University, Umeå, Sweden

  23. Maintaining IoT Device Identification under Concept Drift via Budget-Aware Traffic Labeling

    Aug 16, 2026Shayan Azizi, Norihiro Okui, Masataka Nakahara +3Concept DriftData-Efficient Learning

    School of Electrical Engineering and Telecommunications, University of New South Wales, Sydney, NSW 2052, Australia · KDDI Research, Japan · School of Computer Science and Engineering, University of New South Wales, Sydney, NSW 2052, Australia

  24. Regime-Conditional Verification: Correctness Estimation for Adapting and Monitoring Safety Classifiers

    Aug 14, 2026Thiago Sandoval, Ufuk TopcuDistribution ShiftSelective Prediction

    The University of Texas at Austin

  25. High-dimensional nonparametric changepoint detection via low-rank degree-two density projection

    Aug 14, 2026Guoqing Zhang, Zhaixin ChenDistribution Shift DetectionChange-Point Detection

    Operations Research Program North Carolina State University Raleigh, NC, USA · H. Milton Stewart School of ISyE Georgia Institute of Technology Atlanta, GA, USA

  26. How fine a change can moments see? A scale law for detecting distribution shift, with a kernel calibration rule

    Aug 2, 2026Adel KalecheTwo-Sample TestingDistribution Shift

    Independent researcher, France

  27. Level, Sharpness, and Corpus: Why Zero-Shot OOD Detector Rankings Do Not Transfer

    Jul 29, 2026Ignacio M. De la Jara, Cristian Rodriguez-Opazo, Stephen Gould +1VLM EvaluationDistribution Shift Detection

    University of Adelaide · 3Naval Group Pacific · 2Australian National University

  28. Representation Trajectories Matters: Complementary Evidence for OOD Detection and Image Classification

    Jul 29, 2026Ignacio M. De la Jara, Cristian Rodriguez-Opazo, Hamed Damirchi +2Distribution Shift DetectionOOD Detection

    University of Adelaide · Naval Group Pacific · Australian National University

  29. Black-Mamba: Biologically-Inspired Leaky Accumulation for Conceptual Knowledge under Distribution Drift

    Jul 21, 2026Giuseppe Soriano, Nicola Tonellotto, Alberto GottaTest-Time AdaptationConcept Drift

    University of Pisa National Research Council (CNR) Pisa, Italy · University of Pisa Pisa, Italy · National Research Council (CNR) Pisa, Italy

  30. When Drift Detectors cry Wolf: False Alarm Rates in continuous ML Monitoring

    Jul 19, 2026Raj Shekhar SinghDistribution Shift Detection

    Indian Institute of Technology, Roorkee Roorkee, Uttarakhand, India