Prediction of Potato Rot Level by Using Electronic Nose Based on Data Augmentation and Channel Attention Conditional Convolutional Neural Networks DOI Creative Commons

Jiayu Mai,

Haonan Lin, Xuezhen Hong

и другие.

Chemosensors, Год журнала: 2024, Номер 12(12), С. 275 - 275

Опубликована: Дек. 20, 2024

This study introduces a novel approach for predicting the decay levels of potato by integrating an electronic nose system combined with feature-optimized deep learning models. The was utilized to collect volatile gas data from potatoes at different stages, offering non-invasive method classify levels. To mitigate scarcity and improve training model robustness, Gaussian Mixture Embedded Generative Adversarial Network (GMEGAN) used generate synthetic data, resulting in augmented datasets that increased diversity improved performance. Several machine models, including traditional classifiers (SVM, LR, RF, ANN) advanced neural networks (CNN, ECA-CNN, CAM-CNN, Conditional CNN), were trained evaluated. Models incorporating channel attention modules (f-CAM, f-ECA) achieved classification accuracy up 90.28%, significantly outperforming models (72–77%) standard CNN (83.33%). inclusion GMEGAN-generated further enhanced performance, especially observed increase 5.55%. A comprehensive evaluation feature mapping consistency, distribution similarity, quality metrics, demonstrated generated closely resembled real thereby effectively enhancing dataset diversity. proposed shows significant potential improving robustness agricultural assessment, particularly potatoes.

Язык: Английский

An Electronic Nose Analysis of the Headspace from Extra-Virgin Olive Oil–Saliva Interactions and Its Ability to Differentiate Between Individuals Based on Body Mass Index DOI Creative Commons
Alessandro Genovese, Andrea Balivo, Nicola Caporaso

и другие.

Chemosensors, Год журнала: 2025, Номер 13(2), С. 40 - 40

Опубликована: Янв. 29, 2025

The interaction between fatty foods and saliva in individuals of different body weights may lead to differences the release volatile compounds mouth. This study investigates ability an electronic nose (E-nose) discriminate headspace profiles extra-virgin olive oil (EVOO) mixed with 55 subjects mass indices (BMI). resulting data were analysed using linear discriminant analysis (LDA) principal component (PCA) evaluate E-nose’s groups. W5S, W1S, W2S, W2W sensors exhibited greatest variation response intensity; particular, they highlighted obese non-obese subjects. LDA plot demonstrated a clear separation samples corresponding three BMI groups, first second components accounting for 61.25% 23.97% variance, respectively. Overall, percentage correct classification cross-validation results was 87.3%. These highlight potential use as rapid objective tool screening olfactory associated food matrix–saliva providing valuable insight further research on food–saliva interactions.

Язык: Английский

Процитировано

0
Electronic Nose and GC-MS Analysis to Detect Mango Twig Tip Dieback in Mango (<em>Mangifera indica</em>) and Panama Disease (TR4) in Banana (<em>Musa acuminata</em>) DOI Open Access

Wathsala Ratnayake,

Stanley E. Bellgard, Hao Wang

и другие.

Опубликована: Апрель 11, 2024

Volatile organic compounds (VOCs) released from plants have been correlated with disease-status. Analysis of VOCs using GC-MS is time-consuming, laboratory-based, and requires specialist training. Electronic nose devices (E-nose) provide a portable alternative. Three different E-nose were compared to assess how accurately they could detect Mango Twig Tip Dieback Panama disease in banana. The initially trained on known volatiles, then pure cultures Pantoea sp., Staphylococcus Fusarium odoratissimum, finally, infected healthy mango leaves field-collected, banana pseudo-stems. experiments repeated three times six replicates for each host-pathogen pair. variation between host materials was evaluated by principal component linear discriminant analysis, cross-validation chemometric data analysis. analysis conducted contemporaneously identified an 80% similarity plant material. C 320 100% successful discriminating volatiles but had low capability differentiating substrates. advanced (PEN 3 / MSEM 160) successfully detected diseased samples high variance. results suggest that are more sensitive accurate detecting changes headspace GC-MS.

Язык: Английский

Процитировано

3
Prediction of Potato Rot Level by Using Electronic Nose Based on Data Augmentation and Channel Attention Conditional Convolutional Neural Networks DOI Creative Commons

Jiayu Mai,

Haonan Lin, Xuezhen Hong

и другие.

Chemosensors, Год журнала: 2024, Номер 12(12), С. 275 - 275

Опубликована: Дек. 20, 2024

This study introduces a novel approach for predicting the decay levels of potato by integrating an electronic nose system combined with feature-optimized deep learning models. The was utilized to collect volatile gas data from potatoes at different stages, offering non-invasive method classify levels. To mitigate scarcity and improve training model robustness, Gaussian Mixture Embedded Generative Adversarial Network (GMEGAN) used generate synthetic data, resulting in augmented datasets that increased diversity improved performance. Several machine models, including traditional classifiers (SVM, LR, RF, ANN) advanced neural networks (CNN, ECA-CNN, CAM-CNN, Conditional CNN), were trained evaluated. Models incorporating channel attention modules (f-CAM, f-ECA) achieved classification accuracy up 90.28%, significantly outperforming models (72–77%) standard CNN (83.33%). inclusion GMEGAN-generated further enhanced performance, especially observed increase 5.55%. A comprehensive evaluation feature mapping consistency, distribution similarity, quality metrics, demonstrated generated closely resembled real thereby effectively enhancing dataset diversity. proposed shows significant potential improving robustness agricultural assessment, particularly potatoes.

Язык: Английский

Процитировано

0