Predicting the Reliability of an Image Classifier under Image Distortion

2024-12-22

arXiv.org

AI Summary

To address the challenge of predicting classifier reliability under distorted inputs, this work formulates reliability assessment as a distortion-level binary classification task: given a distortion level, predict whether the model’s output meets a user-specified accuracy threshold. To mitigate severe class imbalance during training—particularly the scarcity of unreliable samples—we propose two Gaussian process-based resampling strategies, integrated with supervised learning to construct a reliability warning classifier. Extensive evaluation across six mainstream image datasets demonstrates that our approach significantly outperforms multiple baselines, achieving an average 12.7% improvement in reliability prediction accuracy and exhibiting strong generalization. Our key contributions are: (1) establishing a novel distortion-level reliability binary classification paradigm; and (2) designing Gaussian process-driven resampling techniques that effectively alleviate data scarcity for the unreliable class in low-sample regimes.

Technology Category

• Machine Learning: Calibration & Uncertainty Quantification
• Computer Vision: Adversarial Attacks & Robustness
• Reasoning under Uncertainty: Relational Probabilistic Models

Application Category

• Web Mining and Content Analysis: Web data provenance, reliability, and authenticity
• Economics, Online Markets and Human Computation: Trust and reliance of crowd workers and data experts on GenAI
• Search and Retrieval-Augmented AI: Web learning to rank, online learning, and counterfactual learning for ranking

Abstract

In image classification tasks, deep learning models are vulnerable to image distortions i.e. their accuracy significantly drops if the input images are distorted. An image-classifier is considered"reliable"if its accuracy on distorted images is above a user-specified threshold. For a quality control purpose, it is important to predict if the image-classifier is unreliable/reliable under a distortion level. In other words, we want to predict whether a distortion level makes the image-classifier"non-reliable"or"reliable". Our solution is to construct a training set consisting of distortion levels along with their"non-reliable"or"reliable"labels, and train a machine learning predictive model (called distortion-classifier) to classify unseen distortion levels. However, learning an effective distortion-classifier is a challenging problem as the training set is highly imbalanced. To address this problem, we propose two Gaussian process based methods to rebalance the training set. We conduct extensive experiments to show that our method significantly outperforms several baselines on six popular image datasets.

Problem

• Predict image classifier reliability under distortion
• Address imbalanced training set for distortion-classifier
• Improve accuracy of reliability prediction methods

Innovation

• Predict reliability using distortion-classifier model
• Rebalance training set with Gaussian process
• Outperform baselines on six datasets