7 February 2025 in Expert Systems with Applications
1 October 2025 in Expert Systems with Applications
6 May 2025 in IEEE Sensors Journal
Taste identification research is a pivotal issue in agriculture and guides agricultural production to guarantee food quality. The bio-inspired electronic tongue (e-tongue) has emerged as a powerful tool for taste classification, demonstrating superiority over human sensory evaluation and physiochemical analysis technologies. However, current taste analysis with e-tongue system focuses on data processing while neglecting the nerve conduction mechanisms (NCMs) underlying human taste perception, which constrains the performance in taste classification. Motivated by this, a gustatory-emotion coupling model (GECM) based on NCMs for human taste and emotion systems (HTESs) is proposed to improve the performance of taste identification. First, the topological structure of nerve transmission pathways for taste and emotion information is built, and the dynamic characteristics of the HTESs are characterized via the GECM. Second, the response characteristics of the brain (including 1/f and chaotic characteristics) are reproduced in the GECM output under stimulation, and the bionic characteristics of the GECM are validated. Finally, a convolutional neural network (CNN) is designed to process the GECM output for taste identification. Compared with multiple classifiers, the best identification performance, which includes the accuracies of 98.88% and 98.45%, the recall scores of 98.91% and 98.23%, the precision scores of 99.17% and 98.90%, the kappa coefficients of 98.61% and 98.07%, and the F1-scores of 99.04% and 98.57%, is acquired using the GECM-CNN. In conclusion, the performance of taste recognition for e-tongue system is improved via the GECM. It provides a foundation for guiding agricultural production and evaluating food quality from NCMs.
27 April 2026 in IEEE Transactions on Instrumentation and Measurement
Apples of different production origins vary in quality due to differences in natural growing conditions. The practice of passing off low-quality apples as high-quality varieties is a widespread issue in the market. To address this issue, a bionic gustatory-emotion coupling model integrated with Hebbianhabituation learning rule (GECM-HHLR), which mimics the nerve conduction and synaptic plasticity mechanisms of the human taste and emotional system (HTESs), is proposed and combined with an electronic tongue (e-tongue) for apple origin recognition. First, GECM-HHLR describes the transmission pathways and learning patterns of HTESs information and improves the 1/f characteristics of brain responses under the stimulation of e-tongue data, demonstrating the bionic performance of GECM-HHLR in processing e-tongue data. Second, GECMHHLR achieves superior results in the classification of apple origins in comparison with mainstream baseline models of the e-tongue system, confirming the adaptability of GECM-HHLR for apple origin analysis. Finally, the fusion of GECM-HHLR and convolutional neural network (CNN) yields the optimal performance with an accuracy of 98.75%, a kappa coefficient of 98.57%, an F1-score of 98.81%, and a precision of 98.64% in apple origin identification, outperforming multiple state-ofthe- art algorithms. In conclusion, effective classification of apple origins is achieved via GECM-HHLR and CNN, providing a biological mechanism-based approach for sensor data processing in food detection.