Jiaxi Li
Current Organizations / Institutes
The organizations/institutes from the scholar's latest article
Previous Organizations / Institutes
The organizations / institutes from the scholar's other articles in the past ten years
Jiaxi Li has a total of 18 co-authors. Here are the top co-authors by number of shared publications:
- Zhelong Wang
Key Laboratory of Intelligent Control and Optimization for Industrial Equipment of Ministry of Education, Dalian University of Technology, Dalian, China
4 shared publications - Fang Lin
Key Laboratory of Intelligent Control and Optimization for Industrial Equipment of the Ministry of Education and the School of Control Science and Engineering, Dalian University of Technology, Dalian, China
3 shared publications - Hongyu Zhao
Key Laboratory of Intelligent Control and Optimization for Industrial Equipment of Ministry of Education, Dalian University of Technology, Dalian, China
3 shared publications - Sen Qiu
School of Control Science and Engineering, Dalian University of Technology, Dalian, China
3 shared publications - Ke Zhang
Dalian University of Technology
2 shared publications - Xin Shi
Key Laboratory of Intelligent Control and Optimization for Industrial Equipment of Ministry of Education, Dalian University of Technology, Dalian, China
2 shared publications
Latest Publications
- 10 June 2025
- Published by Institute of Electrical and Electronics Engineers (IEEE) in IEEE Transactions on Human-Machine Systems
- Vol. 55 (4) , 508-518
Fine-Grained Assessment of Upper-Limb Bradykinesia Through Multimodal Feature Enhancement and Deep Learning
Bradykinesia is a hallmark symptom of Parkinson’s disease (PD) that significantly affects patients’ functional abilities and quality of life. This study proposed a fine-grained classification method for evaluating the level of bradykinesia in PD patients. Based on inertial signals, surface electromyographic (sEMG) signals, and videos obtained from 40 PD patients and 13 healthy subjects, the proposed data preprocessing method extracts 69-D features from inertial and sEMG (IE) signals, and 7-D skeleton features from videos. A two-stream network, including IE stream, skeleton stream, and decision fusion module, was developed using long short-term memory, full convolutional neural networks, and fully connected neural networks. In addition, the IE stream incorporated a feature shrinking module to process high-dimensional features to reduce redundant features. Furthermore, an LSTM-variational autoencoders method was proposed for data augmentation of categories with fewer samples. The proposed method achieved higher recognition rates (pro/supination movements of hands: 85.51%, finger tapping: 88.06%, hand movements: 90.00%) compared to other methods. With low-cost, compact and lightweight methods, bradykinesia in PD patients can be intelligently assessed, which will enhance patient management and treatment efficiency.
- 10 June 2025
- Published by Institute of Electrical and Electronics Engineers (IEEE) in IEEE Transactions on Human-Machine Systems
- Vol. 55 (4) , 508-518
Fine-Grained Assessment of Upper-Limb Bradykinesia Through Multimodal Feature Enhancement and Deep Learning
Bradykinesia is a hallmark symptom of Parkinson’s disease (PD) that significantly affects patients’ functional abilities and quality of life. This study proposed a fine-grained classification method for evaluating the level of bradykinesia in PD patients. Based on inertial signals, surface electromyographic (sEMG) signals, and videos obtained from 40 PD patients and 13 healthy subjects, the proposed data preprocessing method extracts 69-D features from inertial and sEMG (IE) signals, and 7-D skeleton features from videos. A two-stream network, including IE stream, skeleton stream, and decision fusion module, was developed using long short-term memory, full convolutional neural networks, and fully connected neural networks. In addition, the IE stream incorporated a feature shrinking module to process high-dimensional features to reduce redundant features. Furthermore, an LSTM-variational autoencoders method was proposed for data augmentation of categories with fewer samples. The proposed method achieved higher recognition rates (pro/supination movements of hands: 85.51%, finger tapping: 88.06%, hand movements: 90.00%) compared to other methods. With low-cost, compact and lightweight methods, bradykinesia in PD patients can be intelligently assessed, which will enhance patient management and treatment efficiency.
- 1 June 2024
- Published by Elsevier in Information Fusion
- Vol. 111
A comprehensive evaluation method for frailty based on semi-supervised learning and transfer-learning
- 1 June 2024
- Published by Elsevier in Information Fusion
- Vol. 111
A comprehensive evaluation method for frailty based on semi-supervised learning and transfer-learning
- Open Access
Revisiting semi-supervised training objectives for differentiable particle filters
Differentiable particle filters combine the flexibility of neural networks with the probabilistic nature of sequential Monte Carlo methods. However, traditional approaches rely on the availability of labelled data, i.e., the ground truth latent state information, which is often difficult to obtain in real-world applications. This paper compares the effectiveness of two semi-supervised training objectives for differentiable particle filters. We present results in two simulated environments where labelled data are scarce.
Open AccessRevisiting semi-supervised training objectives for differentiable particle filters
Differentiable particle filters combine the flexibility of neural networks with the probabilistic nature of sequential Monte Carlo methods. However, traditional approaches rely on the availability of labelled data, i.e., the ground truth latent state information, which is often difficult to obtain in real-world applications. This paper compares the effectiveness of two semi-supervised training objectives for differentiable particle filters. We present results in two simulated environments where labelled data are scarce.
- 18 March 2024
- Published by Institute of Electrical and Electronics Engineers (IEEE) in IEEE Journal of Biomedical and Health Informatics
- Vol. 28 (5) , 3102-3113
Method for Incomplete and Imbalanced Data Based on Multivariate Imputation by Chained Equations and Ensemble Learning
The classification analysis of incomplete and imbalanced data is still a challenging task since these issues could negatively impact the training of classifiers, which were also found in our study on the physical fitness assessments of patients. And in fields such as healthcare, there are higher requirements for the accuracy of the generated imputation values. To train a high-performance classifier and pursue high accuracy, we attempted to resolve any potential negative impact by using a novel algorithmic approach based on the combination of multivariate imputation by chained equations and the ensemble learning method (MICEEN), which can solve the two problems simultaneously. We used multivariate imputation by chained equations to generate more accurate imputation values for the training set passed to ensemble learning to build a predictor. On the other hand, missing values were introduced into minority classes and used them to generate new samples belonging to the minority classes in order to balance the distribution of classes. On real-world datasets, we perform extensive experiments to assess our method and compare it to other state-of-the-art approaches. The advantages of the proposed method are demonstrated by experimental results for the benchmark datasets and self-collected datasets of physical fitness assessment of tumor patients with varying missing rates.
- 18 March 2024
- Published by Institute of Electrical and Electronics Engineers (IEEE) in IEEE Journal of Biomedical and Health Informatics
- Vol. 28 (5) , 3102-3113
Method for Incomplete and Imbalanced Data Based on Multivariate Imputation by Chained Equations and Ensemble Learning
The classification analysis of incomplete and imbalanced data is still a challenging task since these issues could negatively impact the training of classifiers, which were also found in our study on the physical fitness assessments of patients. And in fields such as healthcare, there are higher requirements for the accuracy of the generated imputation values. To train a high-performance classifier and pursue high accuracy, we attempted to resolve any potential negative impact by using a novel algorithmic approach based on the combination of multivariate imputation by chained equations and the ensemble learning method (MICEEN), which can solve the two problems simultaneously. We used multivariate imputation by chained equations to generate more accurate imputation values for the training set passed to ensemble learning to build a predictor. On the other hand, missing values were introduced into minority classes and used them to generate new samples belonging to the minority classes in order to balance the distribution of classes. On real-world datasets, we perform extensive experiments to assess our method and compare it to other state-of-the-art approaches. The advantages of the proposed method are demonstrated by experimental results for the benchmark datasets and self-collected datasets of physical fitness assessment of tumor patients with varying missing rates.
- Open Access
Learning Differentiable Particle Filter on the Fly
Differentiable particle filters are an emerging class of sequential Bayesian inference techniques that use neural networks to construct components in state space models. Existing approaches are mostly based on offline supervised training strategies. This leads to the delay of the model deployment and the obtained filters are susceptible to distribution shift of test-time data. In this paper, we propose an online learning framework for differentiable particle filters so that model parameters can be updated as data arrive. The technical constraint is that there is no known ground truth state information in the online inference setting. We address this by adopting an unsupervised loss to construct the online model updating procedure, which involves a sequence of filtering operations for online maximum likelihood-based parameter estimation. We empirically evaluate the effectiveness of the proposed method, and compare it with supervised learning methods in simulation settings including a multivariate linear Gaussian state-space model and a simulated object tracking experiment.
Open AccessLearning Differentiable Particle Filter on the Fly
Differentiable particle filters are an emerging class of sequential Bayesian inference techniques that use neural networks to construct components in state space models. Existing approaches are mostly based on offline supervised training strategies. This leads to the delay of the model deployment and the obtained filters are susceptible to distribution shift of test-time data. In this paper, we propose an online learning framework for differentiable particle filters so that model parameters can be updated as data arrive. The technical constraint is that there is no known ground truth state information in the online inference setting. We address this by adopting an unsupervised loss to construct the online model updating procedure, which involves a sequence of filtering operations for online maximum likelihood-based parameter estimation. We empirically evaluate the effectiveness of the proposed method, and compare it with supervised learning methods in simulation settings including a multivariate linear Gaussian state-space model and a simulated object tracking experiment.
- 28 November 2022
- Published by Institute of Electrical and Electronics Engineers (IEEE) in IEEE Transactions on Emerging Topics in Computational Intelligence
- Vol. 7 (4) , 1290-1300
Physical Fitness Assessment for Cancer Patients Using Multi-Model Decision Fusion Based on Multi-Source Data
Objective and quantitative physical fitness assessment is the basis of pre-rehabilitation treatment for cancer patients, which can speed up the treatment process and reduce the adverse effects after treatment. In this study, in order to make up for the shortcomings of the traditional assessment by oncologists according to the scale, a method using multi-model decision fusion based on multi-source data is proposed. After fully considering the physical weakness of the cancer patients, a novel experimental paradigm is designed. The multi-source physiological data of 201 patients suffer from cancers have been collected and a series of features which can be used to assess the physical fitness are extracted. An improved data oversampling method based on the Mahalanobis distance and boundary constraints (MDBC) is proposed to overcome the problem of class imbalance. Moreover, a multi-model decision fusion (MMDF) method is proposed for classification, which combines multiple individual machine learning classifiers with a meta learner, to improve the accuracy of the physical fitness assessment. The recognition accuracy of our proposed method reaches 95.1%. The overall experimental results demonstrate that our proposed method is effective in physical fitness assessment of cancer patients.
- 28 November 2022
- Published by Institute of Electrical and Electronics Engineers (IEEE) in IEEE Transactions on Emerging Topics in Computational Intelligence
- Vol. 7 (4) , 1290-1300
Physical Fitness Assessment for Cancer Patients Using Multi-Model Decision Fusion Based on Multi-Source Data
Objective and quantitative physical fitness assessment is the basis of pre-rehabilitation treatment for cancer patients, which can speed up the treatment process and reduce the adverse effects after treatment. In this study, in order to make up for the shortcomings of the traditional assessment by oncologists according to the scale, a method using multi-model decision fusion based on multi-source data is proposed. After fully considering the physical weakness of the cancer patients, a novel experimental paradigm is designed. The multi-source physiological data of 201 patients suffer from cancers have been collected and a series of features which can be used to assess the physical fitness are extracted. An improved data oversampling method based on the Mahalanobis distance and boundary constraints (MDBC) is proposed to overcome the problem of class imbalance. Moreover, a multi-model decision fusion (MMDF) method is proposed for classification, which combines multiple individual machine learning classifiers with a meta learner, to improve the accuracy of the physical fitness assessment. The recognition accuracy of our proposed method reaches 95.1%. The overall experimental results demonstrate that our proposed method is effective in physical fitness assessment of cancer patients.