Sensing, Spectroscopy & Scientific Imaging
7 published worksArtificial Intelligence in Gas Sensing: A Review
Surveys how machine learning, deep learning, and explainable AI are reshaping gas sensing, from spectral classification to real-time sensor deployment.
Deep Learning for Gas Sensing via Infrared Spectroscopy
Deep learning model trained on simulated absorbance spectra to classify up to three-component gas mixtures across a broad 400–4000 cm⁻¹ range, highlighting the frequencies driving each prediction.
TSMC-Net: Deep-Learning Multigas Classification Using THz Absorption Spectra
Deep learning architecture for terahertz absorption spectra (220–330 GHz) that classifies gas mixtures with up to eight components and identifies the frequencies driving each prediction.
VOC-Net: Deep Learning Volatile Organic Compounds Classifier
1D CNN trained on THz spectral signatures for rapid, selective identification of volatile organic compounds, where conventional analytical techniques often struggle.
Rotational Spectra Classification
Links simulated and measured THz rotational spectra to improve robustness to noise and instrument variability in high-selectivity gas classification.
IR Spectra Classification
SVM models trained on simulated absorbance data achieve reliable multigas IR classification, showing lightweight models can complement deep learning in resource-constrained sensing.