Hyper Evidential Deep Learning to Quantify Composite Classification Uncertainty

TL;DR

This paper introduces Hyper-Evidential Neural Networks (HENN), a novel framework that models and quantifies uncertainty in composite classification tasks using belief theory, improving performance on image datasets.

Contribution

The paper proposes a new hyper-evidential neural network framework that explicitly models composite classification uncertainty with belief theory, a novel approach in deep learning.

Findings

HENN outperforms state-of-the-art methods on four image datasets.

Introduces a new uncertainty measure called vagueness for hyper-opinions.

Effectively models composite class label uncertainty in neural networks.

Abstract

Deep neural networks (DNNs) have been shown to perform well on exclusive, multi-class classification tasks. However, when different classes have similar visual features, it becomes challenging for human annotators to differentiate them. This scenario necessitates the use of composite class labels. In this paper, we propose a novel framework called Hyper-Evidential Neural Network (HENN) that explicitly models predictive uncertainty due to composite class labels in training data in the context of the belief theory called Subjective Logic (SL). By placing a grouped Dirichlet distribution on the class probabilities, we treat predictions of a neural network as parameters of hyper-subjective opinions and learn the network that collects both single and composite evidence leading to these hyper-opinions by a deterministic DNN from data. We introduce a new uncertainty type called vagueness originally designed for hyper-opinions in SL to quantify composite classification uncertainty for DNNs. Our results demonstrate that HENN outperforms its state-of-the-art counterparts based on four image datasets. The code and datasets are available at: https://github.com/Hugo101/HyperEvidentialNN.