Jia Yan
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
Research Subjects
Research Interests
Jia Yan has a total of 233 co-authors. Here are the top co-authors by number of shared publications:
- Shukai Duan
Southwest University
26 shared publications Shukai Duan
College of Artficial Intelligence, Key Laboratory of Brain-like Computing and Intelligent Control (Chongqing), Key Laboratory of Luminescence Analysis and Molecular Sensors (Ministry of Education)
18 shared publications- Lidan Wang
Southwest University
15 shared publications Xiaoyan Peng
11 shared publicationsLidan Wang
College of Artficial Intelligence, Key Laboratory of Brain-like Computing and Intelligent Control (Chongqing), Key Laboratory of Luminescence Analysis and Molecular Sensors (Ministry of Education)
9 shared publications- Pengfei Jia
Guangxi University
9 shared publications
Latest Publications
- Open Access
- 7 July 2026
- Published by Elsevier in SSRN Electronic Journal
Efficient Gas Classification for Electronic Noses Using NbOx Memristor-Based Neurons
Electronic noses emulate the human olfactory system to detect and identify volatile compounds in applications such as food quality assessment, environmental monitoring, and medical diagnostics. Although conventional machine learning and deep learning methods have achieved considerable success in gas classification, their ability to fully exploit the nonlinear and temporally evolving characteristics of sensor responses remains limited. Spiking neural networks (SNNs), inspired by the event-driven communication of biological neurons, offer advantages for temporal information processing and low-power neuromorphic computing. However, hardware implementation of SNNs using conventional CMOS circuits remains challenging because of circuit complexity and limited scalability. In this work, we experimentally demonstrate a compact artificial leaky integrate-and-fire (LIF) neuron implemented using a volatile NbOx memristor. To evaluate the relevance of the proposed neuron to olfactory computing, we performed SNN-based gas classification using data acquired from a 15-sensor electronic nose array. Using linear discriminant analysis (LDA)- preprocessed features, the classifier achieved a test accuracy of 96.06% for ten gas species at multiple concentration levels. Compared with previously reported memristive neuron studies, the proposed approach exhibits tunable spiking behavior and shows potential for compact neuromorphic olfactory systems. Our results support the feasibility of employing NbOx memristor-based neurons in future neuromorphic gas-sensing hardware.
- 7 July 2026
- Published by Elsevier in SSRN Electronic Journal
Open AccessEfficient Gas Classification for Electronic Noses Using NbOx Memristor-Based Neurons
Electronic noses emulate the human olfactory system to detect and identify volatile compounds in applications such as food quality assessment, environmental monitoring, and medical diagnostics. Although conventional machine learning and deep learning methods have achieved considerable success in gas classification, their ability to fully exploit the nonlinear and temporally evolving characteristics of sensor responses remains limited. Spiking neural networks (SNNs), inspired by the event-driven communication of biological neurons, offer advantages for temporal information processing and low-power neuromorphic computing. However, hardware implementation of SNNs using conventional CMOS circuits remains challenging because of circuit complexity and limited scalability. In this work, we experimentally demonstrate a compact artificial leaky integrate-and-fire (LIF) neuron implemented using a volatile NbOx memristor. To evaluate the relevance of the proposed neuron to olfactory computing, we performed SNN-based gas classification using data acquired from a 15-sensor electronic nose array. Using linear discriminant analysis (LDA)- preprocessed features, the classifier achieved a test accuracy of 96.06% for ten gas species at multiple concentration levels. Compared with previously reported memristive neuron studies, the proposed approach exhibits tunable spiking behavior and shows potential for compact neuromorphic olfactory systems. Our results support the feasibility of employing NbOx memristor-based neurons in future neuromorphic gas-sensing hardware.
- Open Access
- 24 June 2026
- Published by MDPI AG in Education Sciences
- Vol. 16  (7) , 999
Stage-Dependent Behavioral Patterns in MOOC Dropout: An Explainable Learning Analytics Study
The high dropout rate in massive open online courses (MOOCs) continues to limit their potential in promoting inclusive and sustainable learning. Although many prediction models have been used to identify potential dropouts, most studies view dropout as a static classification problem and fail to clearly reveal the dynamic trajectory of learner participation over time. Therefore, this study introduces a phased analysis perspective, treating MOOC dropout as a process that continuously evolves at different stages. On the basis of the KDDCUP2015 dataset, we constructed behavioral characteristics at three time points: the first week, the third week, and the fifth week. By combining robust feature analysis and interpretable models, we systematically examined the changing patterns of dropout modes. The results revealed significant differences across the different stages. In the early stage of the course, dropout was related mainly to the unstable interaction behaviors of learners, such as restricted access to resources and irregular participation rhythms. In the middle and late stages, task-oriented behaviors, especially those related to video-based learning activities, gradually became key factors. Notably, high-frequency video participation does not always reduce the risk of dropout; when video activity is high but the overall interaction rate is low, it is more likely to indicate an increase in the risk of dropout. These results indicate that the combination of behaviors is more crucial than mere activity levels. By revealing the changing characteristics of behaviors at different stages, this study helps support the design of more practical early warning methods.
- 24 June 2026
- Published by MDPI AG in Education Sciences
- Vol. 16  (7) , 999
Open AccessStage-Dependent Behavioral Patterns in MOOC Dropout: An Explainable Learning Analytics Study
The high dropout rate in massive open online courses (MOOCs) continues to limit their potential in promoting inclusive and sustainable learning. Although many prediction models have been used to identify potential dropouts, most studies view dropout as a static classification problem and fail to clearly reveal the dynamic trajectory of learner participation over time. Therefore, this study introduces a phased analysis perspective, treating MOOC dropout as a process that continuously evolves at different stages. On the basis of the KDDCUP2015 dataset, we constructed behavioral characteristics at three time points: the first week, the third week, and the fifth week. By combining robust feature analysis and interpretable models, we systematically examined the changing patterns of dropout modes. The results revealed significant differences across the different stages. In the early stage of the course, dropout was related mainly to the unstable interaction behaviors of learners, such as restricted access to resources and irregular participation rhythms. In the middle and late stages, task-oriented behaviors, especially those related to video-based learning activities, gradually became key factors. Notably, high-frequency video participation does not always reduce the risk of dropout; when video activity is high but the overall interaction rate is low, it is more likely to indicate an increase in the risk of dropout. These results indicate that the combination of behaviors is more crucial than mere activity levels. By revealing the changing characteristics of behaviors at different stages, this study helps support the design of more practical early warning methods.
- 19 February 2026
- Published by Institute of Electrical and Electronics Engineers (IEEE) in IEEE Sensors Letters
- Vol. 10  (4) , 1-4
Robust Pseudolabel Subspace Learning for E-Nose Drift Compensation
Sensor drift severely degrades the long-term reliability of electronic nose (E-nose) systems, as temporal and device-induced variations distort the feature distributions of gas responses. Existing drift compensation approaches often require labeled target-domain or calibration samples, which are difficult to obtain in practical deployments, and conventional pseudolabeling strategies are vulnerable to noisy labels under severe drift. To address these challenges, this letter proposes a robust pseudolabel-based subspace learning (RPSL) method for unsupervised E-nose drift compensation. RPSL employs a dual-stage pseudolabel filtering mechanism that selects high-confidence target samples via support vector machine margins and further enforces prediction consistency through prototype learning. On the basis of these reliable samples, a unified subspace learning model is constructed to align source and target distributions by minimizing the mean, covariance, and prototype discrepancies while integrating graph embedding linear discriminant analysis to increase discriminability in the projected space. Experiments on two benchmark E-nose drift datasets under multiple transfer settings show that RPSL achieves stable and high classification accuracy without requiring any target-domain labels, demonstrating its effectiveness and robustness for practical drift compensation.
- 19 February 2026
- Published by Institute of Electrical and Electronics Engineers (IEEE) in IEEE Sensors Letters
- Vol. 10  (4) , 1-4
Robust Pseudolabel Subspace Learning for E-Nose Drift Compensation
Sensor drift severely degrades the long-term reliability of electronic nose (E-nose) systems, as temporal and device-induced variations distort the feature distributions of gas responses. Existing drift compensation approaches often require labeled target-domain or calibration samples, which are difficult to obtain in practical deployments, and conventional pseudolabeling strategies are vulnerable to noisy labels under severe drift. To address these challenges, this letter proposes a robust pseudolabel-based subspace learning (RPSL) method for unsupervised E-nose drift compensation. RPSL employs a dual-stage pseudolabel filtering mechanism that selects high-confidence target samples via support vector machine margins and further enforces prediction consistency through prototype learning. On the basis of these reliable samples, a unified subspace learning model is constructed to align source and target distributions by minimizing the mean, covariance, and prototype discrepancies while integrating graph embedding linear discriminant analysis to increase discriminability in the projected space. Experiments on two benchmark E-nose drift datasets under multiple transfer settings show that RPSL achieves stable and high classification accuracy without requiring any target-domain labels, demonstrating its effectiveness and robustness for practical drift compensation.
- Open Access
Localized Surface Plasmon Resonance-Based Gas Sensor with a Metal–Organic-Framework-Modified Gold Nano-Urchin Substrate for Volatile Organic Compounds Visualization
Volatile organic compound (VOC) monitoring is crucial for environmental safety and health, but conventional gas sensors often suffer from poor selectivity or lack spatial information. Here, we report a localized surface plasmon resonance (LSPR) gas sensor based on Au nano-urchins coated with a zeolitic imidazolate framework (ZIF-8) for both the quantitative detection and visualization of VOCs. Substrates were fabricated by immobilizing Au nano-urchins (~90 nm) on 3-aminopropyltriethoxysilane-modified glass and subsequently growing ZIF-8 crystals (~250 nm) for different durations. Scanning electron microscopy and optical analysis revealed that 90 min of ZIF-8 growth provided the optimal coverage and strongest plasmonic response. Using a spectrometer-based LSPR system, the optimized substrate exhibited clear, concentration-dependent responses to three representative VOCs, 2-pentanone, acetic acid, and ethyl acetate, over nine concentrations, with detection limits of 12.7, 14.5, and 36.3 ppm, respectively. Furthermore, a camera-based LSPR visualization platform enabled real-time imaging of gas plumes and evaporation-driven diffusion, with differential pseudo-color mapping providing intuitive spatial distributions and concentration dependence. These results demonstrate that ZIF-8-modified Au nano-urchin substrates enable sensitive and reproducible VOC detection and, importantly, transform plasmonic sensing into a visual modality, offering new opportunities for integrated LSPR–surface-enhanced Raman scattering dual-mode gas sensing in the future.
Open AccessLocalized Surface Plasmon Resonance-Based Gas Sensor with a Metal–Organic-Framework-Modified Gold Nano-Urchin Substrate for Volatile Organic Compounds Visualization
Volatile organic compound (VOC) monitoring is crucial for environmental safety and health, but conventional gas sensors often suffer from poor selectivity or lack spatial information. Here, we report a localized surface plasmon resonance (LSPR) gas sensor based on Au nano-urchins coated with a zeolitic imidazolate framework (ZIF-8) for both the quantitative detection and visualization of VOCs. Substrates were fabricated by immobilizing Au nano-urchins (~90 nm) on 3-aminopropyltriethoxysilane-modified glass and subsequently growing ZIF-8 crystals (~250 nm) for different durations. Scanning electron microscopy and optical analysis revealed that 90 min of ZIF-8 growth provided the optimal coverage and strongest plasmonic response. Using a spectrometer-based LSPR system, the optimized substrate exhibited clear, concentration-dependent responses to three representative VOCs, 2-pentanone, acetic acid, and ethyl acetate, over nine concentrations, with detection limits of 12.7, 14.5, and 36.3 ppm, respectively. Furthermore, a camera-based LSPR visualization platform enabled real-time imaging of gas plumes and evaporation-driven diffusion, with differential pseudo-color mapping providing intuitive spatial distributions and concentration dependence. These results demonstrate that ZIF-8-modified Au nano-urchin substrates enable sensitive and reproducible VOC detection and, importantly, transform plasmonic sensing into a visual modality, offering new opportunities for integrated LSPR–surface-enhanced Raman scattering dual-mode gas sensing in the future.
- 23 September 2025
- Published by Elsevier in Journal of Food Composition and Analysis
- Vol. 148
A novel manifold discriminant extreme learning machine combined with an E-nose for chili pepper identification via aroma analysis
- 23 September 2025
- Published by Elsevier in Journal of Food Composition and Analysis
- Vol. 148
A novel manifold discriminant extreme learning machine combined with an E-nose for chili pepper identification via aroma analysis
- 14 August 2025
- Published by Elsevier in Biomedical Signal Processing and Control
- Vol. 112
Preoperative prediction of rectal cancer stage via CT imaging and an adaptive attention multiscale feature fusion network
- 14 August 2025
- Published by Elsevier in Biomedical Signal Processing and Control
- Vol. 112
Preoperative prediction of rectal cancer stage via CT imaging and an adaptive attention multiscale feature fusion network
- Open Access
An Integrated Lightweight Neural Network Design and FPGA-Accelerated Edge Computing for Chili Pepper Variety and Origin Identification via an E-Nose
A chili pepper variety and origin detection system that integrates a field-programmable gate array (FPGA) with an electronic nose (e-nose) is proposed in this paper to address the issues of variety confusion and origin ambiguity in the chili pepper market. The system uses the AIRSENSE PEN3 e-nose from Germany to collect gas data from thirteen different varieties of chili peppers and two specific varieties of chili peppers originating from seven different regions. Model training is conducted via the proposed lightweight convolutional neural network ChiliPCNN. By combining the strengths of a convolutional neural network (CNN) and a multilayer perceptron (MLP), the ChiliPCNN model achieves an efficient and accurate classification process, requiring only 268 parameters for chili pepper variety identification and 244 parameters for origin tracing, with 364 floating-point operations (FLOPs) and 340 FLOPs, respectively. The experimental results demonstrate that, compared with other advanced deep learning methods, the ChiliPCNN has superior classification performance and good stability. Specifically, ChiliPCNN achieves accuracy rates of 94.62% in chili pepper variety identification and 93.41% in origin tracing tasks involving Jiaoyang No. 6, with accuracy rates reaching as high as 99.07% for Xianjiao No. 301. These results fully validate the effectiveness of the model. To further increase the detection speed of the ChiliPCNN, its acceleration circuit is designed on the Xilinx Zynq7020 FPGA from the United States and optimized via fixed-point arithmetic and loop unrolling strategies. The optimized circuit reduces the latency to 5600 ns and consumes only 1.755 W of power, significantly improving the resource utilization rate and processing speed of the model. This system not only achieves rapid and accurate chili pepper variety and origin detection but also provides an efficient and reliable intelligent agricultural management solution, which is highly important for promoting the development of agricultural automation and intelligence.
Open AccessAn Integrated Lightweight Neural Network Design and FPGA-Accelerated Edge Computing for Chili Pepper Variety and Origin Identification via an E-Nose
A chili pepper variety and origin detection system that integrates a field-programmable gate array (FPGA) with an electronic nose (e-nose) is proposed in this paper to address the issues of variety confusion and origin ambiguity in the chili pepper market. The system uses the AIRSENSE PEN3 e-nose from Germany to collect gas data from thirteen different varieties of chili peppers and two specific varieties of chili peppers originating from seven different regions. Model training is conducted via the proposed lightweight convolutional neural network ChiliPCNN. By combining the strengths of a convolutional neural network (CNN) and a multilayer perceptron (MLP), the ChiliPCNN model achieves an efficient and accurate classification process, requiring only 268 parameters for chili pepper variety identification and 244 parameters for origin tracing, with 364 floating-point operations (FLOPs) and 340 FLOPs, respectively. The experimental results demonstrate that, compared with other advanced deep learning methods, the ChiliPCNN has superior classification performance and good stability. Specifically, ChiliPCNN achieves accuracy rates of 94.62% in chili pepper variety identification and 93.41% in origin tracing tasks involving Jiaoyang No. 6, with accuracy rates reaching as high as 99.07% for Xianjiao No. 301. These results fully validate the effectiveness of the model. To further increase the detection speed of the ChiliPCNN, its acceleration circuit is designed on the Xilinx Zynq7020 FPGA from the United States and optimized via fixed-point arithmetic and loop unrolling strategies. The optimized circuit reduces the latency to 5600 ns and consumes only 1.755 W of power, significantly improving the resource utilization rate and processing speed of the model. This system not only achieves rapid and accurate chili pepper variety and origin detection but also provides an efficient and reliable intelligent agricultural management solution, which is highly important for promoting the development of agricultural automation and intelligence.
- 26 May 2025
- Published by Institute of Electrical and Electronics Engineers (IEEE) in IEEE Sensors Journal
- Vol. 25  (13) , 25088-25097
A Collaborative Strategy for Multiple Intelligent Sensory Technologies Based on a Discriminant Extreme Learning Machine for Detecting Gases
Factory emissions, vehicle emissions, and fossil fuel combustion, along with other activities, release a significant amount of harmful gases that pose a serious threat to human health and the environment. However, when detection is performed via a single sensing mode alone, the accuracy of identification often fails to reach ideal results. In this paper, we propose a guided fusion model based on a discriminant extreme learning machine framework (GF-DELM) to fuse gas sensor data and thermal image data for industrial gas detection; this model achieves effective fusion by purposefully constraining the fusion space of extreme learning machines. Specifically, to enhance the separability of the fused data between different categories, we consider controlling the within-class and between-class dispersion during the fusion process, aiming to minimize within-class while maximizing between-class distances. Additionally, this method further strengthens the relationship between the fused features and labels via the Hilbert–Schmidt independence criterion (HSIC). The experimental results demonstrate that the proposed fusion model outperforms the other fusion methods. Our approach can be further extended to various other gas detection applications by incorporating a range of sensors and additional modal data, thus displaying significant potential for practical implementation.
- 26 May 2025
- Published by Institute of Electrical and Electronics Engineers (IEEE) in IEEE Sensors Journal
- Vol. 25  (13) , 25088-25097
A Collaborative Strategy for Multiple Intelligent Sensory Technologies Based on a Discriminant Extreme Learning Machine for Detecting Gases
Factory emissions, vehicle emissions, and fossil fuel combustion, along with other activities, release a significant amount of harmful gases that pose a serious threat to human health and the environment. However, when detection is performed via a single sensing mode alone, the accuracy of identification often fails to reach ideal results. In this paper, we propose a guided fusion model based on a discriminant extreme learning machine framework (GF-DELM) to fuse gas sensor data and thermal image data for industrial gas detection; this model achieves effective fusion by purposefully constraining the fusion space of extreme learning machines. Specifically, to enhance the separability of the fused data between different categories, we consider controlling the within-class and between-class dispersion during the fusion process, aiming to minimize within-class while maximizing between-class distances. Additionally, this method further strengthens the relationship between the fused features and labels via the Hilbert–Schmidt independence criterion (HSIC). The experimental results demonstrate that the proposed fusion model outperforms the other fusion methods. Our approach can be further extended to various other gas detection applications by incorporating a range of sensors and additional modal data, thus displaying significant potential for practical implementation.