Deterministic Uncertainty Estimation for Multi-Modal Regression With Deep Neural Networks

Prediction interval (PI) is a common method to represent predictive uncertainty in regression by deep neural networks. This paper proposes an extension of the prediction interval by using a union of disjoint intervals. Since previous PI methods assumed a single-interval PI (one lower and upper bound...

Full description

Saved in:
Bibliographic Details
Main Authors: Jaehak Cho, Jae Myung Kim, Seungyub Han, Jungwoo Lee
Format: Article
Language:English
Published: IEEE 2025-01-01
Series:IEEE Access
Subjects:
Online Access:https://ieeexplore.ieee.org/document/10909531/
Tags: Add Tag
No Tags, Be the first to tag this record!
_version_ 1850023164216606720
author Jaehak Cho
Jae Myung Kim
Seungyub Han
Jungwoo Lee
author_facet Jaehak Cho
Jae Myung Kim
Seungyub Han
Jungwoo Lee
author_sort Jaehak Cho
collection DOAJ
description Prediction interval (PI) is a common method to represent predictive uncertainty in regression by deep neural networks. This paper proposes an extension of the prediction interval by using a union of disjoint intervals. Since previous PI methods assumed a single-interval PI (one lower and upper bound), it suffers from performance degradation in uncertainty estimation when the conditional density function is multi-modal. This paper demonstrates the need to include multi-modality in uncertainty estimation for regression. To address the issue, we propose a novel method that generates a union of disjoint PI&#x2019;s. With UCI benchmark experiments, the proposed method is shown to improve over current state-of-the-art uncertainty quantification methods, reducing an average PI width by over <inline-formula> <tex-math notation="LaTeX">$27~\%$ </tex-math></inline-formula>. With qualitative experiments, it is shown that multi-modality often exists in real-world datasets, and our method produces high-quality PI&#x2019;s compared to existing PI methods.
format Article
id doaj-art-b65e42e9e75e4553a4f832a2a011ccc9
institution DOAJ
issn 2169-3536
language English
publishDate 2025-01-01
publisher IEEE
record_format Article
series IEEE Access
spelling doaj-art-b65e42e9e75e4553a4f832a2a011ccc92025-08-20T03:01:28ZengIEEEIEEE Access2169-35362025-01-0113452814528910.1109/ACCESS.2025.354791110909531Deterministic Uncertainty Estimation for Multi-Modal Regression With Deep Neural NetworksJaehak Cho0Jae Myung Kim1Seungyub Han2https://orcid.org/0009-0001-8704-8968Jungwoo Lee3https://orcid.org/0000-0002-6804-980XCognitive and Machine Learning Laboratory, Interdisciplinary Program in Artificial Intelligence, Seoul National University, Seoul, South KoreaExplainable Machine Learning Laboratory, Cluster of Excellence Machine Learning, University of T&#x00FC;bingen, T&#x00FC;bingen, GermanyCognitive and Machine Learning Laboratory, Interdisciplinary Program in Artificial Intelligence, Seoul National University, Seoul, South KoreaCognitive and Machine Learning Laboratory, Interdisciplinary Program in Artificial Intelligence, Seoul National University, Seoul, South KoreaPrediction interval (PI) is a common method to represent predictive uncertainty in regression by deep neural networks. This paper proposes an extension of the prediction interval by using a union of disjoint intervals. Since previous PI methods assumed a single-interval PI (one lower and upper bound), it suffers from performance degradation in uncertainty estimation when the conditional density function is multi-modal. This paper demonstrates the need to include multi-modality in uncertainty estimation for regression. To address the issue, we propose a novel method that generates a union of disjoint PI&#x2019;s. With UCI benchmark experiments, the proposed method is shown to improve over current state-of-the-art uncertainty quantification methods, reducing an average PI width by over <inline-formula> <tex-math notation="LaTeX">$27~\%$ </tex-math></inline-formula>. With qualitative experiments, it is shown that multi-modality often exists in real-world datasets, and our method produces high-quality PI&#x2019;s compared to existing PI methods.https://ieeexplore.ieee.org/document/10909531/Deep learninguncertainty estimationmulti-modal distributionregression
spellingShingle Jaehak Cho
Jae Myung Kim
Seungyub Han
Jungwoo Lee
Deterministic Uncertainty Estimation for Multi-Modal Regression With Deep Neural Networks
IEEE Access
Deep learning
uncertainty estimation
multi-modal distribution
regression
title Deterministic Uncertainty Estimation for Multi-Modal Regression With Deep Neural Networks
title_full Deterministic Uncertainty Estimation for Multi-Modal Regression With Deep Neural Networks
title_fullStr Deterministic Uncertainty Estimation for Multi-Modal Regression With Deep Neural Networks
title_full_unstemmed Deterministic Uncertainty Estimation for Multi-Modal Regression With Deep Neural Networks
title_short Deterministic Uncertainty Estimation for Multi-Modal Regression With Deep Neural Networks
title_sort deterministic uncertainty estimation for multi modal regression with deep neural networks
topic Deep learning
uncertainty estimation
multi-modal distribution
regression
url https://ieeexplore.ieee.org/document/10909531/
work_keys_str_mv AT jaehakcho deterministicuncertaintyestimationformultimodalregressionwithdeepneuralnetworks
AT jaemyungkim deterministicuncertaintyestimationformultimodalregressionwithdeepneuralnetworks
AT seungyubhan deterministicuncertaintyestimationformultimodalregressionwithdeepneuralnetworks
AT jungwoolee deterministicuncertaintyestimationformultimodalregressionwithdeepneuralnetworks