CreINNs: Credal-Set Interval Neural Networks for Uncertainty Estimation in Classification Tasks

2024-01-10

arXiv.org

AI Summary

To address low uncertainty estimation accuracy and high computational overhead in neural network classification, this paper proposes Credal Interval Neural Networks (CreINNs). CreINNs are the first to integrate deterministic interval-valued outputs with credal set theory, enabling single-pass forward propagation to directly produce probability upper and lower bounds for each class—thereby unifying epistemic and aleatoric uncertainty quantification. The architecture natively supports interval-valued inputs, conferring inherent robustness. On multi- and binary-class benchmarks, CreINNs achieve uncertainty calibration quality comparable to or better than variational Bayesian neural networks (BNNs) and deep ensembles, while accelerating inference by several-fold and eliminating the need for sampling or ensemble averaging. The core innovation lies in a lightweight, deterministic framework that decouples and quantifies multiple uncertainty types without stochasticity or redundancy.

Technology Category

• Machine Learning: Calibration & Uncertainty Quantification
• Reasoning under Uncertainty: Uncertainty Representations
• Cognitive Modeling & Cognitive Systems: Computational Creativity

Application Category

• Graph Algorithms and Modeling for the Web: Graph neural networks and deep learning approaches for Web-related graphs
• Social Networks and Social Media: Influence propagation, information diffusion, and the prediction on networks
• Search and Retrieval-Augmented AI: Web learning to rank, online learning, and counterfactual learning for ranking

Abstract

Effective uncertainty estimation is becoming increasingly attractive for enhancing the reliability of neural networks. This work presents a novel approach, termed Credal-Set Interval Neural Networks (CreINNs), for classification. CreINNs retain the fundamental structure of traditional Interval Neural Networks, capturing weight uncertainty through deterministic intervals. CreINNs are designed to predict an upper and a lower probability bound for each class, rather than a single probability value. The probability intervals can define a credal set, facilitating estimating different types of uncertainties associated with predictions. Experiments on standard multiclass and binary classification tasks demonstrate that the proposed CreINNs can achieve superior or comparable quality of uncertainty estimation compared to variational Bayesian Neural Networks (BNNs) and Deep Ensembles. Furthermore, CreINNs significantly reduce the computational complexity of variational BNNs during inference. Moreover, the effective uncertainty quantification of CreINNs is also verified when the input data are intervals.

Problem

• Uncertainty Estimation
• Neural Network Classification
• Reliability Improvement

Innovation

• CreINNs
• Uncertainty Estimation
• Interval Data