リケラボ論文検索は、全国の大学リポジトリにある学位論文・教授論文を一括検索できる論文検索サービスです。

リケラボ 全国の大学リポジトリにある学位論文・教授論文を一括検索するならリケラボ論文検索大学・研究所にある論文を検索できる

リケラボ 全国の大学リポジトリにある学位論文・教授論文を一括検索するならリケラボ論文検索大学・研究所にある論文を検索できる

大学・研究所にある論文を検索できる 「余事象モデルに基づく未学習クラス推定法と生体信号分類への応用」の論文概要。リケラボ論文検索は、全国の大学リポジトリにある学位論文・教授論文を一括検索できる論文検索サービスです。
リケラボ

コピーが完了しました

URLをコピーしました

論文の公開元へ論文の公開元へ
書き出し

余事象モデルに基づく未学習クラス推定法と生体信号分類への応用

迎田 隆幸 横浜国立大学 DOI:info:doi/10.18880/00013934

2021.06.17

概要

Computational automation systems based on machine learning have seen significant development in recent years, with examples including image recognition technology, man-machine interfaces, and temporal data models for stock price prediction. In the fields of welfare and medicine, effectiveness in pattern recognition technology has also been reported for anomaly detection in medical imagery and the development of myoelectric prosthetic hands.

Classifiers in general simply identify classes predefined in training, but cannot be applied to consideration of specific undefined classes with abnormal patterns. If unknown patterns belonging to undefined classes are input into a trained classifier, misclassification relating to predefined classes will inevitably result. This can cause misclassification of unknown diseases in diagnosis assistance and critical errors in interface control. To address this, unlearned patterns not included in training data need to be recognized.

A variety of classifiers are used to detect unexpected outliers during training. The authors also performed research to enable highly accurate anomaly detection with the novel stochastic OVRGMN (One-vs.-Rest Gaussian Mixture Network) approach, with definition and application of complementary Gaussian distribution. The main issues of previous methods involve the difficulty of stable optimization with small pools of learning data and appropriate thresholds, along with the challenges involved in setting appropriate empirical thresholds. The OVRGMN is also premised on application of the static characteristics of input data, and cannot be used to handle time-series data. This research was performed to develop novel unlearned class detection superior to previous methods and support classification and evaluation of biological signals.

The Normal and Complementary Gaussian Mixture Network (NACGMN) was applied as a novel probabilistic neural approach with unlearned class detection for high classification performance and stable training even with small training samples. The NACGMN incorporates Gaussian mixture models (GMMs) and complementary Gaussian mixture models (CGMMs) representing distribution of training and unlearned classes, respectively. Since the parameters of both distributions can be determined as weighting coefficients of the network with relaxed statistical constraints, the NACGMN supports more stable training. The outcomes of classification experiments employing artificial data and EMG signals demonstrated the validity of the NACGMN.

To extend the proposed network to a classifier capable of handling time-series information, novel One-vs.-Rest hidden Markov Model (OVRHMM) classification was proposed in which time-series data for trained classes are modeled using hidden Markov models (HMMs) with GMMs and unlearned patterns are expressed using HMMs with CGMMs. Based on a posterior probabilities estimated using Bayes' theorem, the OVRHMM can be used to stochastically evaluate the degree of abnormality for input signals. Experimental results indicated that the ORVHMM enabled classification of time-series data that cannot be discriminated using static classifiers.

A time-series data classification model based on the Hidden Semi-Markov Model (HSMM), in which an anomaly state with the CGMM is introduced, was also proposed for more detailed anomaly detection. The related novel classification method combining Bayesian discrimination in consideration of temporal information and estimation of state transition sequencing supports highly accurate classification and detailed anomaly detection. To evaluate the effectiveness of the proposed HSMM for real data, the approach was applied to the classification of care-worker motion with outcomes demonstrating accurate detection and recognition of important work among innumerable movements.

The above achievements show that unlearned-class detection based on complementary event models enabled stable training independent of learning conditions and high-precision classification in consideration of time-series data.

論文の公開元へ

関連論文

関連論文をもっと見る

参考文献

[1] K. Shima et al., “Measurement and Evaluation of Finger Tapping Movements Using Log-linearized Gaussian Mixture Networks,” Sensors 2009, vol. 9(3), 2187- 2201 (2009)

[2] H. Jiang et al., “An Automatic Detection System of Lung Nodule Based on Multigroup Patch-Based Deep Learning Network,” IEEE J. Biomed. Heal. Informatics, vol. 22, no. 4, pp. 1227-1237, 2018.

[3] K. Englehart et al., “Classification of the myoelectric signal using time-frequency based representations”, Medical Engineering & Phiysics, vol. 21, pp. 431-138 (1999)

[4] O. Fukuda et al., “A human-assisting manipulator teleoperated by EMG signals and arm motions,” IEEE Trans. on Roboticsand Automation, vol. 19-2, pp.210- 222, 2003.

[5] T. Tsuji et al., “A Recurrent Log-Linearized Gaussian Mixture Network,” IEEE Trans. on Neural Networks, vol. 14-2, pp.304-316, 2003.

[6] A. Furui et al., “A myoelectric prosthetic hand with muscle synergy-based motion determination and impedance model-based biomimetic control,” Science Robotics, vol. 4, no. 31, eaaw6339, 2019.

[7] M. Akhbari et al., “ECG segmentation and fiducial point extraction using multi hidden Markov model,” Computers in Biology and Medicine, vol. 79, pp.21-29, 2016.

[8] P. Perera et al., “Learning Deep Features for One-Class Classification,” IEEE Trans. on Image Processing, vol. 28, Issue 11, pp. 5450-5463, 2019.

[9] K. Simonyan et al., “Very Deep Convolutional Networks for Large-Scale Image Recognition,” arXiv, 1409.1556,, 2015.

[10] H. He et al., “Learning from Imbalanced Data,” IEEE Transactions on Knowledge and Data Engineering, vol. 21, no. 9, pp. 1263-1284, 2009,

[11] M. Luo et al., “Using imbalanced triangle synthetic data for machine learning anomaly detection,” Computers, Materials and Continua, vol. 58, no. 1, pp. 15- 26, 2019.

[12] Zhi-Hua Zhou et al., “Training cost-sensitive neural networks with methods addressing the class imbalance problem,” IEEE Transactions on Knowledge and Data Engineering, vol. 18, no. 1, pp. 63-77, 2006.

[13] H. Masnadi-Shirazi et al., “Cost-Sensitive Boosting,” IEEE Transactions on Pattern Analysis and Machine Intelligence, vol. 33, no. 2, pp. 294-309, 2011.

[14] W. Lu et al., “Feature fusion for imbalanced ECG data analysis,” Biomedical Signal Processing and Control, vol. 41, pp. 152-160, 2018.

[15] M. Ribeiro et al., “A study of deep convolutional auto-encoders for anomaly detection in videos,” Pattern Recognition Letters, vol. 105, pp. 13-22, 2018.

[16] E. M. Knorr, et al., “Distance-based outliers: algorithms and applications,” The VLDB Journal, vol. 8, pp. 237-253, 2000.

[17] C. M. Rocco S et al., “A support vector machine integrated system for the classification of operation anomalies in nuclear components and systems,” Reliability Engineering and System Safety, vol. 92, Issue 5, pp. 593-600, 2007.

[18] J. Mourao-Miranda et al., “Patient classification as an outlier detection problem: An application of the One-Class Support Vector Machine,” Neuroimage, vol. 58, no. 3, pp. 793-804, 2011.

[19] F. Tony Liu et al., “Isolation Forest,” IEEE Int. Conf. on Data Mining, pp. 413- 422, 2008.

[20] R. Chalapathy et al., “Anomaly Detection using One-Class Neural Networks,” arXiv preprint arXiv:1802.06360, 2018.

[21] K. Shima et al., “A Novel Classification Method with Unlearned-class Detection Based on a Gaussian Mixture Model,” IEEE Int. Conf. on Systems, Man, and Cybernetics, pp. 3726-3731, 2014.

[22] T. Tsuji et al., “A Log-Linearized Gaussian Mixture Network and Its Application to EEG Pattern Classification,” vol. 29, no. 1, pp. 60-72, 1999.

[23] S. Z. Yu, “Hidden semi-Markov models”, Artificial Intelligence, vol. 174, Issue 2, pp. 215-243, 2010.

[24] S. Z. Yu and H. Kobayashi, “Practical Implementation of an Efficient Forwardbackward Algorithm for an Explicit-duration Hidden Markov Model”, IEEE Trans. on Signal Processing, Vol. 54, Issue 5, pp. 1947-1951, 2006.

[25] R. Domingues et al., “A comparative evaluation of outlier detection algorithms: Experiments and analyses,” Pattern Recognition, vol. 74, pp. 406-421, 2018.

[26] J. S. Kim et al., “Robust kernel density estimation,” Journal of Machine Learning Research, vol. 13, pp. 2529-2565, 2012.

[27] M. E. Tipping et al., “Probabilistic Principal Component Analysis,” Journal of the Royal Statistical Society: Series B (Statistical Methodology), vol. 61, Issue 3, pp. 611-622, 1999.

[28] C. M. Bishop, “Pattern Recognition and Machine Learning ,” Springer, 2006

[29] N. Nasios et al., “Variational learning for Gaussian mixture models,” IEEE Transactions on Systems, Man, and Cybernetics, Part B (Cybernetics), vol. 36, no. 4, pp. 849-862, 2006.

[30] Z. Zhang, et al., “Learning a multivariate Gaussian mixture model with the reversible jump MCMC algorithm,” Statistics and Computing, no. 14, pp. 343-355, 2004.

[31] C. E. Rasmussen, “The infinite Gaussian mixture model,” Advances in Neural Information Processing Systems, pp. 554-559, 2000.

[32] B. Bose et al., “Detecting Insider Threats Using RADISH: A System for Real-Time Anomaly Detection in Heterogeneous Data Streams,” IEEE Systems Journal, vol. 11, no. 2, pp. 471-482, 2017.

[33] M. Munir et al., “DeepAnT: A Deep Learning Approach for Unsupervised Anomaly Detection in Time Series,” IEEE Access, Vol. 7, pp. 1991-2005, 2019.

[34] A. A. Cook et al., “Anomaly Detection for IoT Time-Series Data: A Survey,” IEEE Internet of Things Journal, vol. 7, no. 7, pp. 6481-6494, 2020.

[35] T. Fuse et al., “Statistical Anomaly Detection in Human Dynamics Monitoring Using a Hierarchical Dirichlet Process Hidden Markov Model,” IEEE Trans. on Intelligent Transportation Systems, vol. 18, no. 11, pp. 3083-3092, 2017.

[36] J. Bae et al., “Gait phase analysis based on a Hidden Markov Model,” Mechatronics, Vol. 21, Issue 6, pp. 961-970, 2011.

[37] T. Mukaeda et al., “A novel hidden Markov model-based pattern discrimination method with the anomaly detection for EMG signals,” Proceedings of the Annual Int. Conf. of the IEEE Engineering in Medicine and Biology Society, pp. 921-924, 2017.

[38] J. Chorowski et al., “Learning Understandable Neural Networks With Nonnegative Weight Constraints,” IEEE Transactions on Neural Networks and Learning Systems, vol. 26, no. 1, pp. 62-69, 2015.

[39] R. R. Selvaraju et al., “Grad-CAM: Visual Explanations from Deep Networks via Gradient-Based Localization,” 2017 IEEE Int. Conf. on Computer Vision (ICCV), pp. 618-626, 2017.

[40] L. R. Rabiner, “A tutorial on hidden Markov models and selected applications in speech recognition”, Proceedings of the IEEE, vol.77, no. 2, pp. 257-286, 1989.

[41] W. Zeng et al., “Classification of neurodegenerative diseases using gait dynamics via deterministic learning”, Information Sciences, vol.317, pp. 246-258, 2015.

[42] D. Y. Yeung, et al., “Host-based intrusion detection using dynamic and static behavioral models”, Pattern Recognition, vol. 36, no. 1, pp.229-243, 2003.

[43] J. M. Hausdorff et al., “Altered fractal dynamics of gait: reduced stride-interval correlations with aging and Huntington’s disease”, J. Appl. Physiol., vol. 82, no. 1, pp. 262-269, 1997

[44] J.M. Hausdorff et al., “Dynamic markers of altered gait rhythm in amyotrophic lateral sclerosis,” J. Appl. Physiol., vol. 88, no. 6, pp. 2045-2053, 2000.

[45] PhysioNet, “Gait Dynamics in Neuro-Degenerative Disease Data Base,” from http://www.physionet.org/physiobank/database/gaitndd

[46] E. Baratin et al., “Wavelet-based characterization of gait signal for neurological abnormalities,” Gait Posture, vol. 41, no. 2, pp. 634-639, 2015.

[47] Q. Ye et al., “Classification of Gait Patterns in Patients with Neurodegenerative Disease Using Adaptive Neuro-Fuzzy Inference System,” Computational and Mathematical Methods in Medicine, vol. 2018, 2018.

[48] S. A. A. Shah et al., “Classification of Control and Neurodegenerative Disease Subjects Using Tree Based Classifiers,” J. Pharm. Res. Int., vol. 32, no. 11, pp. 63-73, 2020.

[49] S. Miyajima et al., “Minimal Inertial Sensor Placement for Work Recognition and Working Posture Assessment,” 2019 IEEE/SICE International Symposium on System Integration (SII), pp. 391-395, 2019.

[50] Z. Cao et al., “OpenPose: Realtime Multi-Person 2D Pose Estimation Using Part Affinity Fields,” in IEEE Transactions on Pattern Analysis and Machine Intelligence, vol. 43, no. 1, pp. 172-186, 2021.

[51] H. P. H. Shum et al., “Real-Time Posture Reconstruction for Microsoft Kinect,” in IEEE Transactions on Cybernetics, vol. 43, no. 5, pp. 1357-1369, 2013.

[52] J. Duchi et al., “Adaptive Subgradient Methods for Online Learning and Stochastic Optimization,” Journal of Machine Learning Research, vol. 12, pp. 2121-2159, 2011.

[53] M. Zak, “Terminal attractors in neural networks,” Neural Networks, vol. 2, Issue 4, pp. 259-274, 1989.

[54] C. He et al., “A Fully-Pipelined Hardware Design for Gaussian Mixture Models,” IEEE Transactions on Computers, vol. 66, no. 11, pp. 1837-1850, 2017.

[55] N. Bu et al., “FPGA implementation of a probabilistic neural network for a bioelectric human interface,” Midwest Symposium on Circuits and Systems, vol. 3, no. 2, pp. 3-6, 2004.

[56] D. Gorur et al., “Dirichlet process gaussian mixture models: Choice of the base distribution,” Journal of Computer Science and Technology, vol. 25, no. 4, pp. 653-664, 2010.

[57] M. J. Beal et al., “The Infinite Hidden Markov Model,” Advances in Neural Information Processing Systems, pp. 577-584, 2002.

参考文献をもっと見る