From Zero to Hero: Advancing Zero-Shot Foundation Models for Tabular Outlier Detection

2026-02-03

AI Summary

This work addresses the challenge of deploying anomaly detection methods in scenarios where labeled anomalous samples are unavailable. We propose OUTFORMER, the first plug-and-play zero-shot framework for tabular anomaly detection that requires neither fine-tuning nor hyperparameter tuning. Its core innovations lie in a foundation model architecture enhanced by synthetic data pretraining, in-context learning, a synthetic prior mixing mechanism, and a self-evolving curriculum training strategy. Evaluated on AdBench and two newly constructed large-scale benchmarks encompassing over 1,500 datasets, OUTFORMER substantially outperforms existing methods while maintaining high inference efficiency.

Technology Category

• Data Mining & Knowledge Management: Anomaly/Outlier Detection
• Machine Learning: Calibration & Uncertainty Quantification
• Search and Optimization: Learning to Search

Application Category

• Graph Algorithms and Modeling for the Web: Foundation models and LLMs for Web-related graphs
• Economics, Online Markets and Human Computation: Humans versus LLMs for data annotation and labeling
• Search and Retrieval-Augmented AI: Web learning to rank, online learning, and counterfactual learning for ranking

Abstract

Outlier detection (OD) is widely used in practice; but its effective deployment on new tasks is hindered by lack of labeled outliers, which makes algorithm and hyperparameter selection notoriously hard. Foundation models (FMs) have transformed ML, and OD is no exception: Shen et. al. (2025) introduced FoMo-0D, the first FM for OD, achieving remarkable performance against numerous baselines. This work introduces OUTFORMER, which advances FoMo-0D with (1) a mixture of synthetic priors and (2) self-evolving curriculum training. OUTFORMER is pretrained solely on synthetic labeled datasets and infers test labels of a new task by using its training data as in-context input. Inference is fast and zero-shot, requiring merely forward pass and no labeled outliers. Thanks to in-context learning, it requires zero additional work-no OD model training or bespoke model selection-enabling truly plug-and-play deployment. OUTFORMER achieves state-of-the-art performance on the prominent AdBench, as well as two new large-scale OD benchmarks that we introduce, comprising over 1,500 datasets, while maintaining speedy inference.

Problem

• outlier detection
• zero-shot learning
• tabular data
• foundation models
• labeled outliers

Innovation

• zero-shot learning
• foundation models
• tabular outlier detection
• in-context learning
• synthetic data