Trusted detection for Parkinson’s disease based on uncertainty estimation

doi:10.19682/j.cnki.1005-8885.2024.0019

中国邮电高校学报(英文) ›› 2024, Vol. 31 ›› Issue (5): 85-94.doi: 10.19682/j.cnki.1005-8885.2024.0019

• Artificial Intelligence • 上一篇

Trusted detection for Parkinson’s disease based on uncertainty estimation

周林¹,李云¹,Ji Wei¹,刘宇轩¹,郑慧芬²

1. 南京邮电大学
2. 南京医科大学附属老年医院

收稿日期:2023-09-11 修回日期:2024-03-31 出版日期:2024-10-31 发布日期:2024-10-31
通讯作者: 李云 E-mail:liyun@njupt.edu.cn
基金资助:
江苏省高校基础科学(自然科学)重大项目

Trusted detection for Parkinson’s disease based on uncertainty estimation

Received:2023-09-11 Revised:2024-03-31 Online:2024-10-31 Published:2024-10-31
Contact: Yun LI E-mail:liyun@njupt.edu.cn

摘要/Abstract

摘要：

Currently, most deep learning methods used for Parkinson’s disease ( PD) detection lack reliability assessment. This characteristic makes it is difficult to identify erroneous results in practice, leading to potentially serious consequences. To address this issue, a prior network with the distance measure ( PNDM) layer was proposed in this paper. PNDM layer consists of two modules: prior network ( PN) and the distance measure ( DM) layer. The prior network is employed to estimate data uncertainty, and the DM layer is utilized to estimate model uncertainty. The goal of this work is to provide accurate and reliable PD detection through uncertainty estimation. Experiments show that PNDM layer can effectively estimate both model uncertainty and data uncertainty, rendering it more suitable for uncertainty estimation in PD detection compared to existing methods.

关键词:

Parkinson’s disease ( PD), deep learning, uncertainty estimation

Abstract: Currently, most deep learning methods used for Parkinson’s disease ( PD) detection lack reliability assessment. This characteristic makes it is difficult to identify erroneous results in practice, leading to potentially serious consequences. To address this issue, a prior network with the distance measure ( PNDM) layer was proposed in this paper. PNDM layer consists of two modules: prior network ( PN) and the distance measure ( DM) layer. The prior network is employed to estimate data uncertainty, and the DM layer is utilized to estimate model uncertainty. The goal of this work is to provide accurate and reliable PD detection through uncertainty estimation. Experiments show that PNDM layer can effectively estimate both model uncertainty and data uncertainty, rendering it more suitable for uncertainty estimation in PD detection compared to existing methods.

Key words:

Parkinson’s disease ( PD), deep learning, uncertainty estimation

参考文献

[1] SARKAR S, RAYMICK J, IMAM S. Neuroprotective and therapeutic strategies against Parkinson’s disease: recent perspectives. International Journal of Molecular Sciences, 2016, 17(6): Article 904.

[2] HUGHES A J, DANIEL S E, BEN-SHLOMO Y, et al. The accuracy of diagnosis of Parkinsonian syndromes in a specialist movement disorder service. Brain, 2002, 125(4): 861 - 870.

[3] RIZZO G, COPETTI M, ARCUTI S, et al. Accuracy of clinical diagnosis of Parkinson disease: a systematic review and meta- analysis. Neurology, 2016, 86(6): 566 - 576.

[4] MEKYSKA J, GALAZ Z, MZOUREK Z. Assessing progress of Parkinson’s disease using acoustic analysis of phonation. Proceedings of the 4th International Work Conference on Bioinspired Intelligence ( IWOBI’15 ), 2015, Jun 10 - 12, San Sebastian, Spain. Piscataway, NJ, USA: IEEE, 2015: 111 - 118.

[5] MORO-VELAZQUEZ L, VILLALBA J, DEHAK N. Using X-vectors to automatically detect Parkinson’s disease from speech. Proceedings of the 2020 IEEE International Conference on Acoustics, Speech and Signal Processing ( ICASSP’20 ), 2020, May 4 - 8, Barcelona, Spain. Piscataway, NJ, USA: IEEE, 2020: 1155 - 1159.

[6] GAWLIKOWSKI J, TASSI C R N, ALI M, et al. A survey of uncertainty in deep neural networks. Artificial Intelligence Review, 2023, 56: 1513 - 1589.

[7] JOSPIN L V, LAGA H, BOUSSAID F, et al. Hands-on Bayesian neural networks: a tutorial for deep learning users. IEEE Computational Intelligence Magazine, 2022, 17(2): 29 - 48.

[8] MACKAY D J C. A practical Bayesian framework for backpropagation networks. Neural Computation, 1992, 4 ( 3 ): 448 - 472.

[9] MALININ A, GALES M. Predictive uncertainty estimation via prior networks. Proceedings of the 32nd Annual Conference on Neural Information Processing Systems ( NIPS’18 ), 2018, Dec 3 - 8, Montréal, Canada. Red Hook, NY, USA: Curran Associates Inc, 2018: 7047 - 7058.

[10] CHARPENTIER B, ZÜGNER D, GÜNNEMANN S. Posterior network: uncertainty estimation without OOD samples via density- based pseudo-counts. Proceedings of the 34th Annual Conference on Neural Information Processing Systems ( NIPS’20), 2020, Dec 6 - 12, Vancouver, Canada. Red Hook, NY, USA: Curran Associates Inc, 2020: 1356 - 1367.

[11] VÁSQUEZ-CORREA J C, OROZCO-ARROYAVE J R, NÖTH E. Convolutional neural network to model articulation impairments in patients with Parkinson’s disease. Proceedings of the 18th Annual Conference of the International Speech Communication Association ( INTERSPEECH’17), 2017, Aug 20 - 24, Stockholm, Sweden. Red Hook, NY, USA: Curran Associates Inc, 2017: 314 - 318.

[12] VÁSQUEZ-CORREA J C, ARIAS-VERGARA T, RIOS-URREGO C D, et al. Convolutional neural networks and a transfer learning strategy to classify Parkinson’s disease from speech in three different languages. Proceedings of the 24th Iberoamerican Congress ( CIARP’19 ), 2019, Oct 28 - 31, Havana, Cuba. LNCS 11896. Berlin, Germany: Springer, 2019: 697 - 706.

[13] LIU Y Y, REDDY M K, PENTTIL魧 N, et al. Automatic assessment of Parkinson’s disease using speech representations of phonation and articulation. IEEE / ACM Transactions on Audio, Speech, and Language Processing, 2023, 31: 242 - 255.

[14] KOHL S A A, ROMERA-PAREDES B, MEYER C, et al. A probabilistic U-Net for segmentation of ambiguous images. Proceedings of the 32nd Annual Conference on Neural Information Processing Systems ( NIPS’18 ), 2018, Dec 3 - 8, Montr佴al, Canada. Red Hook, NY, USA: Curran Associates Inc, 2018: 6965 - 6975.

[15] SOHN K, LEE H, YAN X C. Learning structured output representation using deep conditional generative models. Proceedings of the 28th Annual Conference on Neural Information Processing Systems ( NIPS’15 ): Vol 2, 2015, Dec 7 - 12, Montrea, Canada. Cambridge, MA, USA: MIT Press, 2015: 3483 - 3491.

[16] GAL Y R, GHAHRAMANI Z. Dropout as a Bayesian approximation: representing model uncertainty in deep learning. Proceedings of the 33rd International Conference on Machine Learning ( ICML’16), 2016, Jun 19 - 24, New York, NY, USA. Journal of Machine Learning Research ( JMLR), 2016: 1050 - 1059.

[17] LAKSHMINARAYANAN B, PRITZEL A, BLUNDELL C. Simple and scalable predictive uncertainty estimation using deep ensembles. Proceedings of the 31st Annual Conference on Neural Information Processing Systems ( NIPS’17 ), 2017, Dec 4 - 9, Long Beach, CA, USA. Red Hook, NY, USA: Curran Associates Inc, 2017: 6405 - 6416.

[18] DIAO Z S, JIANG H Y, SHI T Y. A unified uncertainty network for tumor segmentation using uncertainty cross entropy loss and prototype similarity. Knowledge-Based Systems, 2022, 246: Article 108739.

[19] HEIN M, ANDRIUSHCHENKO M, BITTERWOLF J. Why ReLU networks yield high-confidence predictions far away from the training data and how to mitigate the problem. Proceedings of the 2019 IEEE / CVF Conference on Computer Vision and Pattern Recognition ( CVPR’19), 2019, Jun 15 - 20, Long Beach, CA, USA. Piscataway, NJ, USA: IEEE, 2019: 41 - 50.

[20] TAGASOVSKA N, LOPEZ-PAZ D. Single-model uncertainties for deep learning. Proceedings of the 33rd Annual Conference on Neural Information Processing Systems ( NIPS’19 ), 2019, Dec 8 - 14, Vancouver, Canada. Red Hook, NY, USA: Curran Associates Inc, 2019: 6417 - 6428.

[21] CALANDRA R, PETERS J, RASMUSSEN C E, et al. Manifold Gaussian processes for regression. Proceedings of the 2016 International Joint Conference on Neural Networks ( IJCNN’16 ), 2016, Jul 24 - 29, Vancouver, Canada. Piscataway, NJ, USA: IEEE, 2016: 3338 - 3345.

[22] LIU J Z, LIN Z, PADHY S, et al. Simple and principled uncertainty estimation with deterministic deep learning via distance awareness. Proceedings of the 34th Annual Conference on Neural Information Processing Systems ( NIPS’20 ), 2020, Dec 6 - 12, Vancouver, Canada. Red Hook, NY, USA: Curran Associates Inc, 2020: 7498 - 7512.

[23] VAN AMERSFOORT J, SMITH L, TEH Y W, et al. Uncertainty estimation using a single deep deterministic neural network. Proceedings of the 37th International Conference on Machine Learning ( ICML’20 ), 2020, Jul 13 - 18, Online. Red Hook, NY, USA: Curran Associates Inc, 2020: 9690 - 9700.

[24] VAN ERVEN T, HARREMOS P. R佴nyi divergence and Kullback- Leibler divergence. IEEE Transactions on Information Theory, 2014, 60(7): 3797 - 3820.

[25] DIAS S B, GRAMMATIKOPOULOU A, DINIZ J A, et al. Innovative Parkinson’s disease patients’ motor skills assessment: the i-prognosis paradigm. Frontiers in Computer Science, 2020, 2: Article 20.

[26] HSU W N, BOLTE B, TSAI Y H H, et al. HuBERT: self-supervised speech representation learning by masked prediction of hidden units. IEEE / ACM Transactions on Audio, Speech, and Language Processing, 2021, 29: 3451 - 3460.

[27] NAEINI M P, COOPER G F, HAUSKRECHT M. Obtaining well calibrated probabilities using Bayesian binning. Proceedings of the 29th AAAI Conference on Artificial Intelligence ( AAAI’15 ), 2015, Jan 25 - 30, Austin, TX, USA. Palo Alto, CA, USA: Association for the Advancement of Artificial Intelligence ( AAAI), 2015: 2901 - 2907.

[28] HENDRYCKS D, GIMPEL K. A baseline for detecting misclassified and out-of-distribution examples in neural networks. Proceedings of the 5th International Conference on learning Representations ( ICLR’17 ), 2017, Apr 24 - 26, Toulon, France. 2017: 1 - 12.

Trusted detection for Parkinson’s disease based on uncertainty estimation

Trusted detection for Parkinson’s disease based on uncertainty estimation

PDF

PDF (Mobile)

可视化

摘要/Abstract

引用本文

使用本文

参考文献

相关文章 0

编辑推荐

Metrics

本文评价