Related Experiment Video
Updated: May 21, 2025
08:46
Electronic Tongue Generating Continuous Recognition Patterns for Protein Analysis
Published on: September 16, 2014
7.8K
Enhancing E-Nose Performance via Metal-Oxide Based MEMS Sensor Arrays Optimization and Feature Alignment for Drug
Ruiwen Kong1,2, Wenfeng Shen2,3,4,5, Yang Gao5,6
1Faculty of Electrical Engineering and Computer Science, Ningbo University, Ningbo 315211, China.
Sensors (Basel, Switzerland)
|March 18, 2025
Summary
This study enhances electronic nose (e-nose) accuracy by optimizing sensor arrays and using feature alignment for reliable drug classification. This approach overcomes device inconsistencies, enabling better odor recognition and model sharing across e-nose systems.
Area of Science:
- Materials Science
- Sensor Technology
- Machine Learning
Background:
- Electronic noses (e-noses) face challenges in classification accuracy due to sensor array variability and feature discrepancies.
- Transferring classification models between e-nose devices is hindered by hardware inconsistencies, limiting practical applications, especially in drug classification.
Purpose of the Study:
- To develop a novel approach for improving e-nose classification accuracy.
- To optimize sensor array selection and implement feature alignment for enhanced odor recognition and model sharing.
- To address challenges in material selection and model transfer for drug classification using e-noses.
Main Methods:
- Fabrication of six tin dioxide (SnO2)-based micro-electro-mechanical systems (MEMS) gas sensors via physical vapor deposition.
- Utilized the ReliefF algorithm for sensor ranking and optimal sensor array selection for drug classification.
- Applied feature alignment techniques from transfer learning to enable model sharing across three inconsistent e-nose devices.
Main Results:
- Identified an optimal sensor array composition for drug classification through ReliefF algorithm analysis.
- Successfully enhanced model sharing capabilities among disparate e-nose devices using feature alignment.
- Demonstrated a method to overcome hardware inconsistencies inherent in batch-produced e-noses.
Conclusions:
- The developed approach significantly improves electronic nose classification accuracy and reliability.
- Feature alignment effectively resolves discrepancies, enabling robust model transfer and addressing batch production hardware inconsistencies.
- This research paves the way for the mass production and widespread adoption of consistent and reliable electronic nose systems.
