# Can We Predict Anomaly Detection Performance from Embedding-Space Geometry?

arXiv: https://arxiv.org/abs/2609.26460 · HTML: https://mlsanity.com/paper/2609.26460/

## Key facts

- Cited by 1 papers in the map (data through 14 Oct 2026).
- Builds on 24 of its 44 references that are in the map.
- First posted on arXiv 22 Sep 2026.

## Builds on (most cited first, 24 of 24)

- [Deep Anomaly Detection with Outlier Exposure](https://mlsanity.com/paper/1812.04606/) (2018)
- [Towards Total Recall in Industrial Anomaly Detection](https://mlsanity.com/paper/2106.08265/) (2021)
- [Out-of-Distribution Detection with Deep Nearest Neighbors](https://mlsanity.com/paper/2204.06507/) (2022)
- [Fine-tuning CNN Image Retrieval with No Human Annotation](https://mlsanity.com/paper/1711.02512/) (2017)
- [A Unifying Review of Deep and Shallow Anomaly Detection](https://mlsanity.com/paper/2009.11732/) (2020)
- [Detecting Spacecraft Anomalies Using LSTMs and Nonparametric Dynamic Thresholding](https://mlsanity.com/paper/1802.04431/) (2018)
- [BEATs: Audio Pre-Training with Acoustic Tokenizers](https://mlsanity.com/paper/2212.09058/) (2022)
- [Characterizing Adversarial Subspaces Using Local Intrinsic Dimensionality](https://mlsanity.com/paper/1801.02613/) (2018)
- [EAT: Self-Supervised Pre-Training with Efficient Audio Transformer](https://mlsanity.com/paper/2401.03497/) (2024)
- [Scaling up masked audio encoder learning for general audio classification](https://mlsanity.com/paper/2406.06992/) (2024)
- [ToyADMOS2: Another dataset of miniature-machine operating sounds for anomalous sound detection under domain shift conditions](https://mlsanity.com/paper/2106.02369/) (2021)
- [MIMII DG: Sound Dataset for Malfunctioning Industrial Machine Investigation and Inspection for Domain Generalization Task](https://mlsanity.com/paper/2205.13879/) (2022)
- [Description and Discussion on DCASE 2022 Challenge Task 2: Unsupervised Anomalous Sound Detection for Machine Condition Monitoring Applying Domain Generalization Techniques](https://mlsanity.com/paper/2206.05876/) (2022)
- [Description and Discussion on DCASE 2023 Challenge Task 2: First-Shot Unsupervised Anomalous Sound Detection for Machine Condition Monitoring](https://mlsanity.com/paper/2305.07828/) (2023)
- [Description and Discussion on DCASE 2024 Challenge Task 2: First-Shot Unsupervised Anomalous Sound Detection for Machine Condition Monitoring](https://mlsanity.com/paper/2406.07250/) (2024)
- [Description and Discussion on DCASE 2025 Challenge Task 2: First-shot Unsupervised Anomalous Sound Detection for Machine Condition Monitoring](https://mlsanity.com/paper/2506.10097/) (2025)
- [On Out-of-Distribution Detection for Audio with Deep Nearest Neighbors](https://mlsanity.com/paper/2210.15283/) (2022)
- [Finite-Sample Analysis of Fixed-k Nearest Neighbor Density Functional Estimators](https://mlsanity.com/paper/1606.01554/) (2016)
- [Model Selection of Anomaly Detectors in the Absence of Labeled Validation Data](https://mlsanity.com/paper/2310.10461/) (2023)
- [Local Density-Based Anomaly Score Normalization for Domain Generalization](https://mlsanity.com/paper/2509.10951/) (2025)
- [Deep Generic Representations for Domain-Generalized Anomalous Sound Detection](https://mlsanity.com/paper/2409.05035/) (2024)
- [Exploring Self-Supervised Audio Models for Generalized Anomalous Sound Detection](https://mlsanity.com/paper/2508.12230/) (2025)
- [Why do Angular Margin Losses work well for Semi-Supervised Anomalous Sound Detection?](https://mlsanity.com/paper/2309.15643/) (2023)
- [Temporal Pooling Strategies for Training-Free Anomalous Sound Detection with Self-Supervised Audio Embeddings](https://mlsanity.com/paper/2603.04605/) (2026)

## Built on by (most cited first, 1 of 1)

- [Anomaly-Free Self-Optimization via AUC Bounds](https://mlsanity.com/paper/2609.27362/) (2026)

## Abstract (arXiv)

Anomaly detection systems are often trained using normal data alone, while model selection and evaluation typically require labeled anomalies. We study whether anomaly detection performance can be predicted without access to anomalous data. For kNN-based detectors, we derive a lower bound on the area under the ROC curve (AUC) that relates detection performance to the separation between inlier and outlier scores and to their respective variances. Under a local scaling model, we use this bound to characterize how density variation, intrinsic-dimensional heterogeneity, and cross-domain mismatch contribute to score variability. We then investigate anomaly-free model selection and show that inlier score variance alone does not reliably predict performance across different representations. To address this limitation, we introduce simple pseudo-anomaly probes that provide a reference for estimating relative score separation. Experiments on the DCASE 2022-2025 benchmarks, spanning four embedding models and 208 candidate systems, show that pseudo-anomaly-based estimators substantially improve anomaly-free model selection. In particular, diverse pseudo-anomalies enable anomaly-free model selection to outperform conventional development-set selection under domain shift. These results show that embedding-space geometry contains predictive information about anomaly detection performance while also highlighting the representation-dependent nature of inlier-only performance estimates.