Machine Metacognition and Uncertainty Quantification

Summarise with VIRAT
Authors: Murray Scott Bennett
Cited by: 0

Summary

No summary yet. VIRAT can read this paper and write a plain-language summary of its aim, methods, key findings and limitations.

Summarise with VIRAT Free to try with your Science Catalogs account.

Abstract

This project introduces a latent‑space density approach for estimating neural network uncertainty. By analyzing neighbourhood structure in penultimate‑layer representations, we derive a familiarity metric that identifies high‑risk classifications, detects high‑confidence errors, and supports safer human–AI collaboration. Code, analyses, and reproducible materials are provided.

Fields of Study & Subjects

Adversarial Robustness in Machine Learning Advanced Neural Network Applications Explainable Artificial Intelligence (XAI)

Publication details

Journal / source
OSF Preprints (OSF Preprints)
Publication date
16 March 2026
Language
EN
Document type
Preprint