# Interpretable Failure Detection with Human-Level Concepts

Source: [Interpretable Failure Detection with Human-Level Concepts](<https://www.sota2.com/research/paper/68fa78121e209131cea76a26>)

Collection index: [llms.txt](<https://www.sota2.com/research/llms.txt>)

Data updated: 2025-10-26T04:31:26.317Z

Authors: Kien X. Nguyen, Tang Li, Xi Peng

Published: 2025-02-07T19:19:40.000Z

## Abstract

Reliable failure detection holds paramount importance in safety-critical applications. Yet, neural networks are known to produce overconfident predictions for misclassified samples. As a result, it remains a problematic matter as existing confidence score functions rely on category-level signals, the logits, to detect failures. This research introduces an innovative strategy, leveraging human-level concepts for a dual purpose: to reliably detect when a model fails and to transparently interpret why. By integrating a nuanced array of signals for each category, our method enables a finer-grained assessment of the model's confidence. We present a simple yet highly effective approach based on the ordinal ranking of concept activation to the input image. Without bells and whistles, our method significantly reduce the false positive rate across diverse real-world image classification benchmarks, specifically by 3.7% on ImageNet and 9% on EuroSAT.

## Recorded benchmark results

A research result applies to its stated dataset, split, metric and evaluation configuration. Compare like-for-like settings, preserve the metric's unit and direction, and distinguish paper-reported results from independent reproduction. The best result recorded here is not a claim of exhaustive, current global coverage.

## Original sources and implementation

[Original paper](<https://arxiv.org/abs/2502.05275>)

[PDF](<https://arxiv.org/pdf/2502.05275.pdf>)