气体传感器阵列分类的可复制基准:从FE-ELM到ROCKET和TS2I-CNNs
Chang-Hyun Kim1,2, Seung-Hwan Choi1, Sanghun Choi2
1Department of Advanced Mobility Components Group, Korea Institute of Industrial Technology, Daegu 42994, Republic of Korea.
Sensors (Basel, Switzerland)
|October 29, 2025
概括
分类低度气体传感器阵列 (GSA) 数据是具有挑战性的. ROCKET时间序列分类器在GSA数据集上取得了最高的准确性,超过了其他方法,包括时间序列到图像 (TS2I) CNNs.
科学领域:
- 机器学习 机器学习
- 传感器技术 传感器技术
- 数据科学数据科学数据科学
背景情况:
- 低度气体传感器阵列 (GSA) 数据的分类,由于信号噪声比 (SNR) 低,传感器异质性,漂移和有限的样本大小,因此存在重大挑战.
- 现有的方法在这些不利条件下往往难以实现稳健的性能,需要新的方法和全面的基准测试.
研究的目的:
- 对各种时间序列分类方法进行比较,包括时间序列对图像 (TS2I) 卷积神经网络 (CNN),与低度GSA数据的既定基线进行比较.
- 提供不同分类策略的可复制和公平比较,以指导电子鼻子 (e-nose) 应用中的模型选择.
主要方法:
- 重现了一个强大的FE-ELM基线,并将其与矢量基线,传统时间序列分类器 (TCN,MiniROCKET) 和TS2I-CNNs进行比较.
- 使用GSA-LC和GSA-FM数据集进行严格的20x5重复分层交叉验证 (n=100) 以进行可靠的评估.
- 使用准确度和宏F1分数评估性能,使用对联t测试与霍尔姆-邦费罗尼统计显著性校正.
主要成果:
- 在GSA-FM (0.9721 ± 0.0480) 和GSA-LC (0.9578 ± 0.0433) 两种数据集上,ROCKET时间序列分类器实现了最高的准确性,明显超过了TCN和MiniROCKET (p < 0.05).
- 在基于图像的模型中,CNN-RP显示出最强大的稳定性,而CNN-GASF显示出局限性,特别是在GSA-LC数据集上.
- 与ResNet-18的RGB融合策略和转移学习没有提供一致的优势,表明数据集依赖性和有限的概括性.
结论:
- ROCKET是对低度GSA数据进行分类的最高性能模型,提供卓越的准确性和可靠性.
- 对于这些具有挑战性的条件,CNN-RP是最可行的TS2I替代方案,提供了强大的基于图像的方法.
- 该研究建立了一个可重复的基准,为电子鼻子系统的模型选择提供了实际指导,并澄清了TS2I方法的功能.
更多相关视频
09:34Implementation of Portable Emissions Measurement Systems PEMS for the Real-driving Emissions RDE Regulation in Europe
Published on: December 4, 2016
28.7K
09:49Identification of Olfactory Volatiles using Gas Chromatography-Multi-unit Recordings GCMR in the Insect Antennal Lobe
Published on: February 24, 2013
14.7K
相关概念视频
Gas Chromatography: Types of Detectors-II
1.1K
In gas chromatography, different detectors are employed to meet specific analytical needs. These detectors are often categorized based on their detection mechanisms and the types of compounds they are best suited to analyze. Thermal Conductivity Detectors (TCD), Flame Ionization Detectors (FID), and Electron Capture Detectors (ECD) represent common categories, each with unique operating principles and applications. However, beyond these, several other detectors are designed for more specialized...
1.1K
Gas Chromatography: Types of Detectors-I
1.4K
There are different types of detectors used in gas chromatography, each with its own specific properties that make it suitable for detecting certain types of analytes. The most commonly used detectors in GC are thermal conductivity detector (TCD), flame ionization detector (FID), and electron capture detector (ECD).
TCD is the earliest and most widely used detector that operates by measuring the changes in the thermal conductivity of the carrier gas. When a sample compound enters the detector,...
TCD is the earliest and most widely used detector that operates by measuring the changes in the thermal conductivity of the carrier gas. When a sample compound enters the detector,...
1.4K
