Author name cluster

Lijun Yang

Possible papers associated with this exact author name in Arrow. This page groups case-insensitive exact name matches and is not a full identity disambiguation profile.

9 papers

2 author rows

AAAI Conference 2026 Conference Paper

Constrained Particle Seeking: Solving Diffusion Inverse Problems with Just Forward Passes

Hongkun Dou
Zike Chen
Zeyu Li
Hongjue Li
Lijun Yang
Yue Deng

Diffusion models have gained prominence as powerful generative tools for solving inverse problems due to their ability to model complex data distributions. However, existing methods typically rely on complete knowledge of the forward observation process to compute gradients for guided sampling, limiting their applicability in scenarios where such information is unavailable. In this work, we introduce *Constrained Particle Seeking (CPS)*, a novel gradient-free approach that leverages all candidate particle information to actively search for the optimal particle while incorporating constraints aligned with high-density regions of the unconditional prior. Unlike previous methods that passively select promising candidates, CPS reformulates the inverse problem as a constrained optimization task, enabling more flexible and efficient particle seeking. We demonstrate that CPS can effectively solve both image and scientific inverse problems, achieving results comparable to gradient-based methods while significantly outperforming gradient-free alternatives.

PDF Details DOI

JBHI Journal 2026 Journal Article

Multi-Source Discriminant Dynamic Domain Adaptation for Cross-Subject Motor Imagery EEG Recognition

Yifan Gong
Kaiting Shi
Xiaolong Niu
Lijun Yang
Xiaohui Yang
Chen Zheng

Electroencephalography (EEG) has emerged as a widely utilized signal in motor imagery (MI) brain-computer interfaces(BCI) due to its convenience and safety. Recently, deep learning methods have rapidly developed in the field of brain computer interfaces. However, traditional EEG classification methods often face challenges related to limited generalization capability across subjects. To address this issue, this paper proposes a multi-source discriminant dynamic domain adaptation model(MSD-DDA) aimed at fully leveraging domain adaptation to enhance the accuracy of motor imagery classification. The model adeptly handles global and local disparities in motor imagery classification by dynamically minimizing differences between global domain and local subdomain. Furthermore, to ensure discriminability and diversity in the target domain, we introduce batch kernel norm maximization of the difference, thereby enhancing the model’s discriminability in the target domain while preserving prediction diversity. To tackle variations in similarity between different source domains and the target domain, we devise a weighted joint prediction mechanism. This mechanism automatically adjusts the contribution weight of each source domain based on its similarity to the target domain, facilitating more precise discriminant prediction and improved adaptability to scenarios with multiple source domains. To evaluate our approach, we conducted a large number of experiments on datasets 1 and 2a of the Fourth BCI Competition and on the openBMI dataset, with average classification accuracy of 92. 43%, 79. 24% and 71. 96%, respectively. Finally, we compare the proposed method with several classical and recent algorithms, and prove that its performance is better than the existing methods.

Details DOI

JBHI Journal 2025 Journal Article

A Fusion Network With Stacked Denoise Autoencoder and Meta Learning for Lateral Walking Gait Phase Recognition and Multi-Step-Ahead Prediction

Wujing Cao
Changyu Li
Lijun Yang
Meng Yin
Chunjie Chen
Worawarit Kobsiriphat
Thanak Utakapan
Yizhuang Yang

Lateral walking gait phase recognition and prediction are the premise of hip exoskeleton application in lateral resistance walk exercise. We presented a fusion network with stacked denoise autoencoder and meta learning (SDA-NN-ML) to recognize gait phase and predict gait percentage from IMU signals. Experiments were conducted to detect the four lateral walking gait phases and predict their percentage across different speeds. The performance of SDA-NN-ML and Support Vector Machine (SVM), Adaptive Boosting (AdaBoost) and Long Short Term Memory (LSTM) were evaluated. The cross-subject recognition accuracy of SDA-NN-ML (89. 94%) decreased by 4. 62% compared to the training accuracy, which outperformed SVM (8. 60%), AdaBoost (5. 61%), and LSTM (7. 12%). For real-time and cross-subject prediction of gait phase percentage, the RMSE of SDA-NN-ML (0. 2043) outperformed that of a single regression network (0. 2426). With a signal noise ratio of 100: 30, the cross-subject recognition accuracy decreased by a mere 5. 70%, while the prediction result (RMSE) of SDA-NN-ML increased by 0. 0167 when compared to the noise-free results. SDA-NN-ML demonstrates a stable multi-step-ahead prediction ability with an accuracy higher than 82. 50% and an RMSE of less than 0. 23 when the ahead time is less than 200 ms. The results demonstrated that the proposed method has high accuracy and robust performance in lateral walking gait recognition and prediction.

Details DOI

ICLR Conference 2025 Conference Paper

Hybrid Regularization Improves Diffusion-based Inverse Problem Solving

Hongkun Dou
Zeyu Li
Jinyang Du
Lijun Yang
Wen Yao 0001
Yue Deng 0001

Diffusion models, recognized for their effectiveness as generative priors, have become essential tools for addressing a wide range of visual challenges. Recently, there has been a surge of interest in leveraging Denoising processes for Regularization (DR) to solve inverse problems. However, existing methods often face issues such as mode collapse, which results in excessive smoothing and diminished diversity. In this study, we perform a comprehensive analysis to pinpoint the root causes of gradient inaccuracies inherent in DR. Drawing on insights from diffusion model distillation, we propose a novel approach called Consistency Regularization (CR), which provides stabilized gradients without the need for ODE simulations. Building on this, we introduce Hybrid Regularization (HR), a unified framework that combines the strengths of both DR and CR, harnessing their synergistic potential. Our approach proves to be effective across a broad spectrum of inverse problems, encompassing both linear and nonlinear scenarios, as well as various measurement noise statistics. Experimental evaluations on benchmark datasets, including FFHQ and ImageNet, demonstrate that our proposed framework not only achieves highly competitive results compared to state-of-the-art methods but also offers significant reductions in wall-clock time and memory consumption.

Details

EAAI Journal 2025 Journal Article

Non-contact weight intelligent estimation based on yak skeleton localization

Fei Wang
Xinghua Zou
Zhijiang Chen
Qi Tang
Tianshuo Li
Shuiying Wang
Lijun Yang
Dongming Tang

Weight estimation is a vital method for monitoring the growth and health of yaks, however, traditional techniques-such as relying on herders’ experience or using weighbridge-are labor-intensive, time-consuming and pose safety risks. Currently, many studies have shown that yak body size can be an effective indicator of weight. With the advancement of computer vision, non-contact weight estimation has become increasingly feasible for livestock. Yet, studies focusing on yak weight estimation, particularly in high-altitude plateau regions, remain limited. Herein, we propose a novel weight estimation approach based on deep learning and binocular vision technology to address this gap. The method involves four main steps: (1) yak image acquisition, (2) skeletal key point localization, (3) body size calculation (4) weight estimation using Gaussian process regression. To enhance practicality and mobility, we also developed two edge-intelligent devices: an intelligent inspection vehicle and a handheld detection unit, enabling convenient and non-invasive weight estimation. Our models are trained and tested on a yak dataset collected by our team on the Tibetan Plateau. Experimental results demonstrate the effectiveness of our approach, achieving an Mean Absolute Percentage Error (MAPE) of 0. 12 percent, a Mean Absolute Error (MAE) of 25. 4 kilograms (kg) and a Coefficient of determination ( R 2 ) value of 0. 72. It not only provides a new technical solution for the yak industry but also provides innovative insights for advancing intelligent animal husbandry. The code and data can be accessed at https: //github. com/FeiWang-swun/YakWeight.

Details DOI

ICLR Conference 2025 Conference Paper

Physics-aligned field reconstruction with diffusion bridge

Zeyu Li
Hongkun Dou
Shen Fang
Wang Han
Yue Deng 0001
Lijun Yang

The reconstruction of physical fields from sparse measurements is pivotal in both scientific research and engineering applications. Traditional methods are increasingly supplemented by deep learning models due to their efficacy in extracting features from data. However, except for the low accuracy on complex physical systems, these models often fail to comply with essential physical constraints, such as governing equations and boundary conditions. To overcome this limitation, we introduce a novel data-driven field reconstruction framework, termed the Physics-aligned Schr\"{o}dinger Bridge (PalSB). This framework leverages a diffusion bridge mechanism that is specifically tailored to align with physical constraints. The PalSB approach incorporates a dual-stage training process designed to address both local reconstruction mapping and global physical principles. Additionally, a boundary-aware sampling technique is implemented to ensure adherence to physical boundary conditions. We demonstrate the effectiveness of PalSB through its application to three complex nonlinear systems: cylinder flow from Particle Image Velocimetry experiments, two-dimensional turbulence, and a reaction-diffusion system. The results reveal that PalSB not only achieves higher accuracy but also exhibits enhanced compliance with physical constraints compared to existing methods. This highlights PalSB's capability to generate high-quality representations of intricate physical interactions, showcasing its potential for advancing field reconstruction techniques. The source code can be found at https://github.com/lzy12301/PalSB.

Details

EAAI Journal 2024 Journal Article

Electroencephalogram-based emotion recognition using factorization temporal separable convolution network

Lijun Yang
Yixin Wang
Rujie Ouyang
Xiaolong Niu
Xiaohui Yang
Chen Zheng

Temporal Convolutional Networks (TCNs) expand their receptive field through dilated convolutions, which is essential for capturing dependencies in longer sequences. This characteristic is especially critical for detecting long-term patterns in Electroencephalogram (EEG) data, making TCNs a suitable choice for EEG-based emotion recognition. In this study, to effectively capture the interaction between features, we incorporate Factorization Machines (FM) into the TCN model, proposing the Factorization Temporal Convolution Network (FTCN) for EEG-based emotion recognition. On one hand, the FTCN model enhances understanding of temporal dynamics by capturing long-term dependencies in time series data through TCN. On the other hand, it combines FM to increase the model’s expression ability in the feature dimension, allowing for a more comprehensive understanding of EEG data. Building on this, separable convolutions are incorporated into the FTCN to develop the Factorization Temporal Separable Convolution Network (FTSCN). This approach reduces the model’s parameter count by splitting standard convolutions into two simpler operations, thus accelerating training and inference. Experiments on the DEAP and SEED datasets demonstrate the effectiveness of these two models, showcasing competitive recognition accuracy compared to published methods. Specifically, on the DEAP dataset, the recognition accuracies for arousal and valence achieve 97. 39% ± 1. 93 and 97. 55% ± 1. 65. For the four-class task, the accuracy reaches 95. 43% ± 2. 43; on the SEED dataset, the recognition accuracy can reach 89. 13% ± 4. 49.

Details DOI

EAAI Journal 2024 Journal Article

QFS-CNN for sub-synchronous oscillation source location based on dissipating energy flow

Wenxu Xiang
Pan Li
Jianliang Song
Lijun Yang

Sub-synchronous oscillation (SSO) caused by large-scale renewable energy generations can threaten the safe and reliable operation of power systems. In this paper, a dissipating energy-based quaternion feature set convolutional neural network (QFS-CNN) method is proposed to locate the SSO sources quickly and accurately, providing the foundation for isolation of the SSO sources from power systems in time. Firstly, aiming at the difficulty in feature extraction of SSO sources due to the poor observability of power system, a temporal and spatial feature extraction method based on oscillation energy is proposed to characterize the SSO sources information in the form of images with limited measurement data. The temporal and spatial feature images are extracted based on the variation of oscillation energy in time and the direction of energy flow power in space, respectively. Then, the correlation between the oscillation energy and the SSO sources location can be revealed by the feature images with less calculation and storage space. Secondly, a QFS-CNN method based on temporal and spatial feature images for SSO sources location is proposed to address the insufficient labeled data of SSO. The QFS data augmentation technique increases the feature set via meaningful recombination of the existing labeled feature images. Then, the augmentative feature set is used to ensure the model training accuracy of QFS-CNN to improve the location accuracy of SSO sources. Finally, the proposed method is demonstrated and evaluated by a modified IEEE 39-bus wind power generation system. Simulation results show that the method has high accuracy and stronger anti-noise ability in the case of poor system observability and few sample data.

Details DOI

EAAI Journal 2023 Journal Article

Label consistency-based deep semisupervised NMF for tumor recognition

Lijun Yang
Lulu Yan
Xiaoge Wei
Xiaohui Yang

Tumor has become a hot topic in the field of image processing and pattern recognition. Gene expression data is an important way to study tumor. Since gene expression data is characterized by high-dimensional small samples, it is one of the key steps to extract discriminative gene features to distinguish different tumor types. Nonnegative matrix factorization (NMF) is an unsupervised feature representation method that does not depend on the label information of data. NMF can achieve nonlinear dimension reduction and is widely used in tumor recognition. Considering that some data may carry label information, models’ feature representation capability will be improved if the label information can be used effectively. Therefore, this paper proposes a semisupervised NMF model based on label consistency (LC-NMF), which uses both labeled and unlabeled data to obtain feature representations. Furthermore, to alleviate the sensitivity of the NMF model to initial values and excavate the deep features of data, a label consistency-based deep semisupervised NMF model (LC-DNMF) is constructed, which combines the LC-NMF model with the layer-by-layer pretraining and multilayer representation strategy in deep learning. The performance of the proposed models (i. e. , LC-NMF and LC-DNMF) is verified by applying them to the tumor recognition tasks. The experimental results on seven datasets show that the two models achieve good results and can obtain competitive recognition accuracies compared with the state-of-the-art methods. Furthermore, the performance of the LC-DNMF model outperforms that of the LC-NMF model, which verifies the effectiveness of introducing the layer-by-layer pretraining and multilayer representation strategy.

Details DOI