AI & ML
Detection and Classification in Latent Spaces: High-Dimensional Analysis and Validation
arXiv stat.ML0 views
arXiv:2611.18164v1 Announce Type: new Abstract: Foundation models produce high-dimensional latent representations in which suitably designed statistical decision rules can be effective yet analytically tractable. We study such detectors in the high-dimensional regime, where latent dimension and training sample size grow proportionally. For linear and quadratic statistics, we derive explicit large-deviation characterizations of false-alarm and miss-detection performance under the Neyman-Pearson paradigm, and show how the binary analysis extends to multi-class classification. The resulting error probabilities and corresponding decay rates exhibit non-monotonic dependence on the shape parameter, including double-descent and, in some regimes, multiple-descent. Experiments on vision, text, and audio datasets using modern foundation models show close agreement between theoretical predictions and empirical performance. In particular, the proposed framework allows us to set a prescribed false-alarm level and accurately predict the corresponding miss-detection probability. Overall, our results provide a theoretically grounded and interpretable framework with competitive empirical performance across different modalities.
Source:https://arxiv.org/abs/2611.18164PreviousTraining-point consistency of kernel ridge regression with varying correlation scalesNextLaMe: a Joint Latent Mediation Framework for High-Dimensional Systems
Related posts
Building a public test set of bad AI citations. What should be in it?
0dev.to
Syncing dual-system audio across multiple cameras without timecode, then letting an AI rough-cut it via MCP
0dev.to
Why LLMs Hallucinate on PDF Extraction and How Grounded Extraction Makes It Verifiable
0dev.to
Doing Show HN generates a ton of automated email pitches
0Hacker News
Apple TV App to Get Video Podcasts Section Today
0MacRumors

New Apple TV 4K's A19 Chip Has One Core Switched Off
0MacRumors