Unsupervised Raw IR Spectra Images Recognition: Toward Chemist-Like Chemical Structure Classification

Recent advances in artificial intelligence have significantly improved the spectral data analysis. In this study, we used unsupervised machine learning to classify chemical compounds based on infrared (IR) spectral images, without relying on prior chemical knowledge. This study demonstrated the potential of machine learning for chemical classification by extracting IR spectral images from the Spectral Database for Organic Compounds (SDBS) database and converting them into 218196-dimensional vector data. Hierarchical clustering of 227 compounds revealed distinct main clusters (A-G), each with specific sub-clusters exhibiting higher intra-cluster similarity. Despite challenges, including sensitivity to spectral deviations and difficulty in distinguishing delicate chemical structures in spectra with low transparency in the fingerprint area, this image recognition approach exhibits potential. The Tanimoto coefficient has been used as a metric for molecular similarity, providing valuable insights, although sometimes diverging from chemists’ intuitions. The study also highlighted that the scaling composition formulas and molecular weights did not affect the classification results because high-dimensional features dominated the process. Overall, this research demonstrates the feasibility of using IR spectral image data in machine learning for chemical classification, offering a novel perspective that complements traditional methods, although some classifications may not always align with chemists’ intuition.