Detecting Adversarial Examples and Other Misclassifications in Neural Networks by Introspection
Not on the map. Why?
Abstract (arXiv)
Despite having excellent performances for a wide variety of tasks, modern neural networks are unable to provide a reliable confidence value allowing to detect misclassifications. This limitation is at the heart of what is known as an adversarial example, where the network provides a wrong prediction associated with a strong confidence to a slightly modified image. Moreover, this overconfidence issue has also been observed for regular errors and out-of-distribution data. We tackle this problem by what we call introspection, i.e. using the information provided by the logits of an already pretrained neural network. We show that by training a simple 3-layers neural network on top of the logit activations, we are able to detect misclassifications at a competitive level.
Key facts. Cited by 16 papers in the map (data through 7 Oct 2026). First posted on arXiv 22 May 2019.
Builds on
No reference list yet.
Built on by
Citing papers per year: 2021: 5; 2022: 4; 2023: 1; 2024: 3; 2025: 2; 2026: 1.
16 papers.
Sort:
By area:
- 2021Adversarial Example Detection for DNN Models: A Review and Experimental Comparisoncited by 51
- 2021Detecting Adversarial Examples from Sensitivity Inconsistency of Spatial-Transform Domaincited by 29
- 2022"That Is a Suspicious Reaction!": Interpreting Logits Variation to Detect NLP Adversarial Attackscited by 24
- 2021Revisiting Model's Uncertainty and Confidences for Adversarial Example Detectioncited by 9
- 2022Threat Assessment in Machine Learning based Systemscited by 9
- 2024ReToMe-VA: Recursive Token Merging for Video Diffusion-based Unrestricted Adversarial Attackcited by 3
- 2025MetaSel: A Test Selection Approach for Fine-tuned DNN Modelscited by 3
- 2024PASA: Attack Agnostic Unsupervised Adversarial Detection using Prediction & Attribution Sensitivity Analysiscited by 2
- 2025Multi-Agent Framework for Threat Mitigation and Resilience in AI-Based Systemscited by 1
- 2021MixDefense: A Defense-in-Depth Framework for Adversarial Example Detection Based on Statistical and Semantic Analysiscited by 0
- 2021Using Anomaly Feature Vectors for Detecting, Classifying and Warning of Outlier Adversarial Examplescited by 0
- 2022DAD: Data-free Adversarial Defense at Test Timecited by 0
4 more
- 2022Detecting and Recovering Adversarial Examples from Extracting Non-robust and Highly Predictive Adversarial Perturbationscited by 0
- 2023DAD++: Improved Data-free Test Time Adversarial Defensecited by 0
- 2024AED-PADA:Improving Generalizability of Adversarial Example Detection via Principal Adversarial Domain Adaptationcited by 0
- 2026LogitDynamics: Reliable ViT Error Detection from Layerwise Logit Trajectoriescited by 0
- 2026LogitDynamics: Reliable ViT Error Detection from Layerwise Logit Trajectories12 Apr 2026
- 2025Multi-Agent Framework for Threat Mitigation and Resilience in AI-Based Systems29 Dec 2025
- 2025MetaSel: A Test Selection Approach for Fine-tuned DNN Models21 Mar 2025
- 2024ReToMe-VA: Recursive Token Merging for Video Diffusion-based Unrestricted Adversarial Attack10 Aug 2024
- 2024AED-PADA:Improving Generalizability of Adversarial Example Detection via Principal Adversarial Domain Adaptation19 Apr 2024
- 2024PASA: Attack Agnostic Unsupervised Adversarial Detection using Prediction & Attribution Sensitivity Analysis12 Apr 2024
- 2023DAD++: Improved Data-free Test Time Adversarial Defense10 Sep 2023
- 2022Threat Assessment in Machine Learning based Systems30 Jun 2022
- 2022Detecting and Recovering Adversarial Examples from Extracting Non-robust and Highly Predictive Adversarial Perturbations30 Jun 2022
- 2022"That Is a Suspicious Reaction!": Interpreting Logits Variation to Detect NLP Adversarial Attacks10 Apr 2022
- 2022DAD: Data-free Adversarial Defense at Test Time4 Apr 2022
- 2021Using Anomaly Feature Vectors for Detecting, Classifying and Warning of Outlier Adversarial Examples1 Jul 2021
4 more
- 2021Adversarial Example Detection for DNN Models: A Review and Experimental Comparison1 May 2021
- 2021MixDefense: A Defense-in-Depth Framework for Adversarial Example Detection Based on Statistical and Semantic Analysis20 Apr 2021
- 2021Revisiting Model's Uncertainty and Confidences for Adversarial Example Detection9 Mar 2021
- 2021Detecting Adversarial Examples from Sensitivity Inconsistency of Spatial-Transform Domain7 Mar 2021
- 2021Adversarial Example Detection for DNN Models: A Review and Experimental Comparisoncited by 51
- 2021Detecting Adversarial Examples from Sensitivity Inconsistency of Spatial-Transform Domaincited by 29
- 2022"That Is a Suspicious Reaction!": Interpreting Logits Variation to Detect NLP Adversarial Attackscited by 24
- 2021Revisiting Model's Uncertainty and Confidences for Adversarial Example Detectioncited by 9
- 2022Threat Assessment in Machine Learning based Systemscited by 9
- 2025MetaSel: A Test Selection Approach for Fine-tuned DNN Modelscited by 3
- 2024PASA: Attack Agnostic Unsupervised Adversarial Detection using Prediction & Attribution Sensitivity Analysiscited by 2
- 2021MixDefense: A Defense-in-Depth Framework for Adversarial Example Detection Based on Statistical and Semantic Analysiscited by 0
- 2021Using Anomaly Feature Vectors for Detecting, Classifying and Warning of Outlier Adversarial Examplescited by 0
- 2022DAD: Data-free Adversarial Defense at Test Timecited by 0
- 2022Detecting and Recovering Adversarial Examples from Extracting Non-robust and Highly Predictive Adversarial Perturbationscited by 0
- 2023DAD++: Improved Data-free Test Time Adversarial Defensecited by 0
12 most cited of 14 All 14 (JSON)
- 2026LogitDynamics: Reliable ViT Error Detection from Layerwise Logit Trajectories12 Apr 2026
- 2025MetaSel: A Test Selection Approach for Fine-tuned DNN Models21 Mar 2025
- 2024AED-PADA:Improving Generalizability of Adversarial Example Detection via Principal Adversarial Domain Adaptation19 Apr 2024
- 2024PASA: Attack Agnostic Unsupervised Adversarial Detection using Prediction & Attribution Sensitivity Analysis12 Apr 2024
- 2023DAD++: Improved Data-free Test Time Adversarial Defense10 Sep 2023
- 2022Threat Assessment in Machine Learning based Systems30 Jun 2022
- 2022Detecting and Recovering Adversarial Examples from Extracting Non-robust and Highly Predictive Adversarial Perturbations30 Jun 2022
- 2022"That Is a Suspicious Reaction!": Interpreting Logits Variation to Detect NLP Adversarial Attacks10 Apr 2022
- 2022DAD: Data-free Adversarial Defense at Test Time4 Apr 2022
- 2021Using Anomaly Feature Vectors for Detecting, Classifying and Warning of Outlier Adversarial Examples1 Jul 2021
- 2021Adversarial Example Detection for DNN Models: A Review and Experimental Comparison1 May 2021
- 2021MixDefense: A Defense-in-Depth Framework for Adversarial Example Detection Based on Statistical and Semantic Analysis20 Apr 2021
12 newest of 14 All 14 (JSON)