PAC-Wrap: Semi-Supervised PAC Anomaly Detection
Shuo Li, Xiayan Ji, Edgar Dobriban, Oleg Sokolsky, Insup Lee
Abstract
Anomaly detection is essential for preventing hazardous outcomes for safety-critical applications like autonomous driving. Given their safety-criticality, these applications benefit from provable bounds on various errors in anomaly detection. To achieve this goal in the semi-supervised setting, we propose to provide Probably Approximately Correct (PAC) guarantees on the false negative and false positive detection rates for anomaly detection algorithms. Our method (PAC-Wrap) can wrap around virtually any existing semisupervised and unsupervised anomaly detection method, endowing it with rigorous guarantees. Our experiments with various anomaly detectors and datasets indicate that PAC-Wrap is broadly effective.
Ask Lune
Know this paper in minutes, not an afternoon.
Lune can read every page of it and the top-tier papers that cite it, and gives you its claims, its evidence and its limits, each one cited to the paper it comes from. No skimming, no AI slop.
5 top-tier papers have cited it since. Know what they found before your reviewers do.
Lune can draw on
the full paper · the top-tier papers that cite it · related work
Free to start. No credit card required.
Lineage
- Conformal Structured Prediction cites PAC Confidence Sets for Deep Neural Networks via Calibrated Prediction
- Conformal Information Pursuit for Interactively Guiding Large Language Models cites iDECODe: In-Distribution Equivariance for Conformal Out-of-Distribution Detection
- Conformal Information Pursuit for Interactively Guiding Large Language Models cites PAC Confidence Sets for Deep Neural Networks via Calibrated Prediction
- Conformal Inference under High-Dimensional Covariate Shifts via Likelihood-Ratio Regularization cites iDECODe: In-Distribution Equivariance for Conformal Out-of-Distribution Detection
- Unsupervised Anomaly Detection with Rejection cites Deep Semi-Supervised Anomaly Detection
Cited by top-tier papers5
- Unsupervised Anomaly Detection with RejectionLorenzo Perini, Jesse DavisNeurIPS 2023 · 17 citations
- Probabilistic Stability Guarantees for Feature AttributionsHelen Jin, Anton Xue, Weiqiu You, Surbhi Goel et al.NeurIPS 2025 · 12 citations
- Conformal Inference under High-Dimensional Covariate Shifts via Likelihood-Ratio RegularizationSunay Joshi, Shayan Kiyani, George J. Pappas, Edgar Dobriban et al.NeurIPS 2025 · 12 citations
- Conformal Information Pursuit for Interactively Guiding Large Language ModelsKwan Ho Ryan Chan, Yuyan Ge, Edgar Dobriban, Hamed Hassani et al.NeurIPS 2025 · 9 citations
- Conformal Structured PredictionBotong Zhang, Shuo Li, Osbert BastaniICLR 2025
Builds on4
- Deep Semi-Supervised Anomaly DetectionLukas Ruff, Robert A. Vandermeulen, Nico Görnitz, Alexander Binder et al.ICLR 2020 · 678 citations
- Conformal prediction interval for dynamic time-seriesChen Xu, Yao XieICML 2021 · 174 citations
- PAC Confidence Sets for Deep Neural Networks via Calibrated PredictionSangdon Park, Osbert Bastani, Nikolai Matni, Insup LeeICLR 2020 · 77 citations
- iDECODe: In-Distribution Equivariance for Conformal Out-of-Distribution DetectionRamneet Kaur, Susmit Jha, Anirban Roy, Sangdon Park et al.AAAI 2022 · 53 citations
Related papers
- Is Out-of-Distribution Detection Learnable?Zhen Fang, Yixuan Li, Jie Lu, Jiahua Dong et al.NeurIPS 2022 · 188 citations
- Towards neural networks that provably know when they don't knowAlexander Meinke, Matthias HeinICLR 2020 · 151 citations
- AutoUAD: Hyper-parameter Optimization for Unsupervised Anomaly DetectionWei Dai, Jicong FanICLR 2025
- Scaling Out-of-Distribution Detection for Real-World SettingsDan Hendrycks, Steven Basart, Mantas Mazeika, Andy Zou et al.ICML 2022 · 653 citations
- Lifelong Anomaly Detection Through UnlearningMin Du, Zhi Chen, Chang Liu, Rajvardhan Oak et al.CCS 2019 · 145 citations
