Beyond Accuracy: Reliability and Uncertainty Estimation in Convolutional Neural Networks

TL;DR

This paper compares Bayesian Monte Carlo Dropout and Conformal Prediction for uncertainty estimation in CNNs, highlighting their calibration differences and practical implications for trustworthy AI systems.

Contribution

It provides an empirical comparison of two uncertainty quantification methods in CNNs, emphasizing calibration and validity in high-stakes applications.

Findings

GoogLeNet offers better calibration than VGG16.

Conformal Prediction guarantees statistically valid prediction sets.

H-CNN VGG16 shows higher accuracy but overconfidence.

Abstract

Deep neural networks (DNNs) have become integral to a wide range of scientific and practical applications due to their flexibility and strong predictive performance. Despite their accuracy, however, DNNs frequently exhibit poor calibration, often assigning overly confident probabilities to incorrect predictions. This limitation underscores the growing need for integrated mechanisms that provide reliable uncertainty estimation. In this article, we compare two prominent approaches for uncertainty quantification: a Bayesian approximation via Monte Carlo Dropout and the nonparametric Conformal Prediction framework. Both methods are assessed using two convolutional neural network architectures; H-CNN VGG16 and GoogLeNet, trained on the Fashion-MNIST dataset. The empirical results show that although H-CNN VGG16 attains higher predictive accuracy, it tends to exhibit pronounced overconfidence, whereas GoogLeNet yields better-calibrated uncertainty estimates. Conformal Prediction additionally demonstrates consistent validity by producing statistically guaranteed prediction sets, highlighting its practical value in high-stakes decision-making contexts. Overall, the findings emphasize the importance of evaluating model performance beyond accuracy alone and contribute to the development of more reliable and trustworthy deep learning systems.