Clustering method for the construction of machine learning model with high predictive ability

Chemometrics and Intelligent Laboratory Systems, Год журнала: 2024, Номер 246, С. 105084 - 105084

Опубликована: Фев. 9, 2024

Язык: Английский

---

Unsupervised Machine Learning-Based Image Recognition of Raw Infrared Spectra: Toward Chemist-like Chemical Structural Classification and Beyond Numerical Data

Journal of Chemical Information and Modeling, Год журнала: 2025, Номер unknown

Опубликована: Март 19, 2025

Recent advances in artificial intelligence have significantly improved spectral data analysis. In this study, we used unsupervised machine learning to classify chemical compounds based on infrared (IR) images, without relying prior knowledge. The potential of for classification was demonstrated by extracting IR images from the Spectral Database Organic Compounds and converting them into 208,620-dimensional vector data. Hierarchical clustering 230 revealed distinct main clusters (A-G), each with specific subclusters exhibiting higher intracluster similarities. Despite challenges, including sensitivity deviations difficulty distinguishing delicate structures spectra low transparency fingerprint area, proposed image recognition approach exhibits good potential. Both principal component analysis k-means produced similar results. Furthermore, method high robustness noise. Tanimoto coefficient evaluate molecular similarity, providing valuable insights. However, some results deviated chemists' intuitions. study also highlighted that scaling composition formulas weights did not affect because high-dimensional features dominated process. A comparison obtained fingerprints, using adjusted Rand index as a metric, indicated provided better performance than numerical same resolution. Overall, demonstrates feasibility offers novel perspective complements traditional methods, although classifications may always align This has broader implications fields such drug discovery, materials science, automated analysis, where handling large, raw sets is essential.

Язык: Английский

---

Discovering Net-Zero Chemical Processes and Pathways by Developing Circular Reaction Networks and Their Hierarchical Screening

Industrial & Engineering Chemistry Research, Год журнала: 2025, Номер unknown

Опубликована: Май 20, 2025

Язык: Английский

---

Optimal Feature Engineering in Machine Learning of Oxidative Coupling of Methane

Опубликована: Янв. 1, 2024

Язык: Английский

---