Epistemic Deep Learning
This work addresses uncertainty modeling in machine learning, which is crucial for improving reliability in applications like classification, though it appears incremental as it builds on existing belief function theory.
The paper tackles the problem of uncertainty quantification in deep neural networks by introducing epistemic deep learning based on random-set interpretations of belief functions, and demonstrates through experiments that this approach yields better performance results compared to traditional uncertainty estimation methods.
The belief function approach to uncertainty quantification as proposed in the Demspter-Shafer theory of evidence is established upon the general mathematical models for set-valued observations, called random sets. Set-valued predictions are the most natural representations of uncertainty in machine learning. In this paper, we introduce a concept called epistemic deep learning based on the random-set interpretation of belief functions to model epistemic learning in deep neural networks. We propose a novel random-set convolutional neural network for classification that produces scores for sets of classes by learning set-valued ground truth representations. We evaluate different formulations of entropy and distance measures for belief functions as viable loss functions for these random-set networks. We also discuss methods for evaluating the quality of epistemic predictions and the performance of epistemic random-set neural networks. We demonstrate through experiments that the epistemic approach produces better performance results when compared to traditional approaches of estimating uncertainty.