Wangze Ni, Tao Wang*, Yu Wu, Lechen Chen, Min Zeng, Jianhua Yang, Nantao Hu, Bowei Zhang, Fuzhen Xuan and Zhi Yang*,
{"title":"基于迁移学习驱动Scentformer模型的电子鼻鲁棒气味检测","authors":"Wangze Ni, Tao Wang*, Yu Wu, Lechen Chen, Min Zeng, Jianhua Yang, Nantao Hu, Bowei Zhang, Fuzhen Xuan and Zhi Yang*, ","doi":"10.1021/acssensors.5c0063010.1021/acssensors.5c00630","DOIUrl":null,"url":null,"abstract":"<p >Mimicking the olfactory system of humans, the use of electronic noses (E-noses) for the detection of odors in nature has become a hot research topic. This study presents a novel E-nose based on deep learning architecture called Scentformer, which addresses the limitations of the current E-nose like a narrow detection range and limited generalizability across different scenarios. Armed with a self-adaptive data down-sampling method, the E-nose is capable of detecting 55 different natural odors with the classification accuracy of 99.94%, and the model embedded in the E-nose is analyzed using Shapley Additive exPlanations analysis, providing a quantitative interpretation of the E-nose performance. Furthermore, leveraging Scentformer’s transfer learning ability, the E-nose efficiently adapts to new odors and gases. Rather than retraining all layers of the model on the new odor data set, only the fully connected layers need to be trained for the pretrained model. Using only 1‰ data of the retrained model, the pretrained model-based E-nose can also achieve classification accuracies of 99.14% across various odor and gas concentrations. This provides a robust approach to the detection of diverse direct current signals in real-world applications.</p>","PeriodicalId":24,"journal":{"name":"ACS Sensors","volume":"10 5","pages":"3704–3712 3704–3712"},"PeriodicalIF":8.2000,"publicationDate":"2025-05-15","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":"0","resultStr":"{\"title\":\"Robust Odor Detection in Electronic Nose Using Transfer-Learning Powered Scentformer Model\",\"authors\":\"Wangze Ni, Tao Wang*, Yu Wu, Lechen Chen, Min Zeng, Jianhua Yang, Nantao Hu, Bowei Zhang, Fuzhen Xuan and Zhi Yang*, \",\"doi\":\"10.1021/acssensors.5c0063010.1021/acssensors.5c00630\",\"DOIUrl\":null,\"url\":null,\"abstract\":\"<p >Mimicking the olfactory system of humans, the use of electronic noses (E-noses) for the detection of odors in nature has become a hot research topic. This study presents a novel E-nose based on deep learning architecture called Scentformer, which addresses the limitations of the current E-nose like a narrow detection range and limited generalizability across different scenarios. Armed with a self-adaptive data down-sampling method, the E-nose is capable of detecting 55 different natural odors with the classification accuracy of 99.94%, and the model embedded in the E-nose is analyzed using Shapley Additive exPlanations analysis, providing a quantitative interpretation of the E-nose performance. Furthermore, leveraging Scentformer’s transfer learning ability, the E-nose efficiently adapts to new odors and gases. Rather than retraining all layers of the model on the new odor data set, only the fully connected layers need to be trained for the pretrained model. Using only 1‰ data of the retrained model, the pretrained model-based E-nose can also achieve classification accuracies of 99.14% across various odor and gas concentrations. This provides a robust approach to the detection of diverse direct current signals in real-world applications.</p>\",\"PeriodicalId\":24,\"journal\":{\"name\":\"ACS Sensors\",\"volume\":\"10 5\",\"pages\":\"3704–3712 3704–3712\"},\"PeriodicalIF\":8.2000,\"publicationDate\":\"2025-05-15\",\"publicationTypes\":\"Journal Article\",\"fieldsOfStudy\":null,\"isOpenAccess\":false,\"openAccessPdf\":\"\",\"citationCount\":\"0\",\"resultStr\":null,\"platform\":\"Semanticscholar\",\"paperid\":null,\"PeriodicalName\":\"ACS Sensors\",\"FirstCategoryId\":\"92\",\"ListUrlMain\":\"https://pubs.acs.org/doi/10.1021/acssensors.5c00630\",\"RegionNum\":1,\"RegionCategory\":\"化学\",\"ArticlePicture\":[],\"TitleCN\":null,\"AbstractTextCN\":null,\"PMCID\":null,\"EPubDate\":\"\",\"PubModel\":\"\",\"JCR\":\"Q1\",\"JCRName\":\"CHEMISTRY, ANALYTICAL\",\"Score\":null,\"Total\":0}","platform":"Semanticscholar","paperid":null,"PeriodicalName":"ACS Sensors","FirstCategoryId":"92","ListUrlMain":"https://pubs.acs.org/doi/10.1021/acssensors.5c00630","RegionNum":1,"RegionCategory":"化学","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":null,"EPubDate":"","PubModel":"","JCR":"Q1","JCRName":"CHEMISTRY, ANALYTICAL","Score":null,"Total":0}
Robust Odor Detection in Electronic Nose Using Transfer-Learning Powered Scentformer Model
Mimicking the olfactory system of humans, the use of electronic noses (E-noses) for the detection of odors in nature has become a hot research topic. This study presents a novel E-nose based on deep learning architecture called Scentformer, which addresses the limitations of the current E-nose like a narrow detection range and limited generalizability across different scenarios. Armed with a self-adaptive data down-sampling method, the E-nose is capable of detecting 55 different natural odors with the classification accuracy of 99.94%, and the model embedded in the E-nose is analyzed using Shapley Additive exPlanations analysis, providing a quantitative interpretation of the E-nose performance. Furthermore, leveraging Scentformer’s transfer learning ability, the E-nose efficiently adapts to new odors and gases. Rather than retraining all layers of the model on the new odor data set, only the fully connected layers need to be trained for the pretrained model. Using only 1‰ data of the retrained model, the pretrained model-based E-nose can also achieve classification accuracies of 99.14% across various odor and gas concentrations. This provides a robust approach to the detection of diverse direct current signals in real-world applications.
期刊介绍:
ACS Sensors is a peer-reviewed research journal that focuses on the dissemination of new and original knowledge in the field of sensor science, particularly those that selectively sense chemical or biological species or processes. The journal covers a broad range of topics, including but not limited to biosensors, chemical sensors, gas sensors, intracellular sensors, single molecule sensors, cell chips, and microfluidic devices. It aims to publish articles that address conceptual advances in sensing technology applicable to various types of analytes or application papers that report on the use of existing sensing concepts in new ways or for new analytes.