Machine Learning for Chemical Sciences and Industry
A.Y. 2025/2026
Learning objectives
The course main goal is to introduce students in industrial chemistry and chemistry to the fundamentals of machine learning and some of its applications to chemical sciences and industrial chemistry.
The course will aim at presenting:
- the topic of supervised machine learning through description of data sampling for training and test, with illustration and discussion of different models (linear, kernels, regression, neural networks);
- applications of supervised machine learning to the construction of potential energy surfaces for chemical spectroscopy, kinetics, and thermochemistry;
- unsupervised machine learning with topics of interest in industrial chemistry (flux maps, control charts), and chemical sciences (principal component analysis);
- the possibility to exploit the numerous instrumentation and control data acquired in the chemical industry to improve diagnostics of problems, quality prediction, and control optimization.
The course will aim at presenting:
- the topic of supervised machine learning through description of data sampling for training and test, with illustration and discussion of different models (linear, kernels, regression, neural networks);
- applications of supervised machine learning to the construction of potential energy surfaces for chemical spectroscopy, kinetics, and thermochemistry;
- unsupervised machine learning with topics of interest in industrial chemistry (flux maps, control charts), and chemical sciences (principal component analysis);
- the possibility to exploit the numerous instrumentation and control data acquired in the chemical industry to improve diagnostics of problems, quality prediction, and control optimization.
Expected learning outcomes
At the end of the class, the students will be able to:
1. illustrate the basic concepts of machine learning;
2. understand the theoretical and practical difference between supervised and unsupervised models;
3. proficiently discuss the theoretical foundations and practical aspects of different learning models;
4. appreciate the importance of machine learning for chemical sciences in applications involving spectroscopy, kinetics, and thermochemistry;
5. appreciate the importance of machine learning to improve industrial and control processes, advance quality prediction, enhance environmental sustainability and optimize energy consumption in the chemical industry.
1. illustrate the basic concepts of machine learning;
2. understand the theoretical and practical difference between supervised and unsupervised models;
3. proficiently discuss the theoretical foundations and practical aspects of different learning models;
4. appreciate the importance of machine learning for chemical sciences in applications involving spectroscopy, kinetics, and thermochemistry;
5. appreciate the importance of machine learning to improve industrial and control processes, advance quality prediction, enhance environmental sustainability and optimize energy consumption in the chemical industry.
Lesson period: Second semester
Assessment methods: Esame
Assessment result: voto verbalizzato in trentesimi
Single course
This course cannot be attended as a single course. Please check our list of single courses to find the ones available for enrolment.
Course syllabus and organization
Single session
Responsible
Lesson period
Second semester
Course syllabus
The class will deal with:
- basic mathematical introductory concepts: basics of linear algebra; linear systems resolution; matrix diagonalization; linear regression;
- introductory topics in machine learning: supervised models, unsupervised models, reinforcement learning;
- supervised machine learning for chemistry: description of the main goal of supervised learning; training and test sets; data sampling; generalization and basics of statistical learning theory; regularization; model selection and validation;
- linear models, kernels, and trees for supervised models: multiple linear regression; non linear regression and kernel ridge regression; selection of kernel functions and hyperparameters; Gaussian process regression; trees and random forests; featurization or how to transform chemical structures into numbers: molecular graphs, lasso method, and random forests;
- neural networks: perceptrons; multilayer perceptrons and hidden layers; deep learning; optimization and training; regularization and hyperparameter selection; applications of machine learning and neural networks to potential energy surfaces for kinetics, thermochemistry and spectroscopy calculations.
- unsupervised machine learning: flux maps; control charts; principal component analysis; clusterization of data.
- machine learning in chemical industry: instrumentation and control data as operational historians; industrial applications: diagnostics; condition of predictive monitoring; quality prediction; control optimization; scheduling.
- basic mathematical introductory concepts: basics of linear algebra; linear systems resolution; matrix diagonalization; linear regression;
- introductory topics in machine learning: supervised models, unsupervised models, reinforcement learning;
- supervised machine learning for chemistry: description of the main goal of supervised learning; training and test sets; data sampling; generalization and basics of statistical learning theory; regularization; model selection and validation;
- linear models, kernels, and trees for supervised models: multiple linear regression; non linear regression and kernel ridge regression; selection of kernel functions and hyperparameters; Gaussian process regression; trees and random forests; featurization or how to transform chemical structures into numbers: molecular graphs, lasso method, and random forests;
- neural networks: perceptrons; multilayer perceptrons and hidden layers; deep learning; optimization and training; regularization and hyperparameter selection; applications of machine learning and neural networks to potential energy surfaces for kinetics, thermochemistry and spectroscopy calculations.
- unsupervised machine learning: flux maps; control charts; principal component analysis; clusterization of data.
- machine learning in chemical industry: instrumentation and control data as operational historians; industrial applications: diagnostics; condition of predictive monitoring; quality prediction; control optimization; scheduling.
Prerequisites for admission
No particular prerequisite is needed apart from the basic knowledge of mathematics commonly provided to students of bachelor degrees in the chemical field. Some lectures may use computer simulations presented by the lecturer, but there is no need for any particular coding knowledge or ability by the students.
Teaching methods
Classroom lectures supported by slides and in-depth analysis at the whiteboard. Attendance is warmly and highly recommended.
Teaching Resources
Main source of information will come from the slides shown during the lectures and in-depth analysis at the whiteboard. All teaching material will be uploaded on the MyAriel website of the course. Some review articles describing topics treated in classroom will be suggested during the lectures.
A book suggested especially for the supervised learning part of the program is Machine Learning in Molecular Sciences edited by Chen Qu and Hanchao Liu, Volume 36 in Challenges and Advances in Computational Chemistry and Physics, Springer (2024). Other books will be suggested during the lectures.
A book suggested especially for the supervised learning part of the program is Machine Learning in Molecular Sciences edited by Chen Qu and Hanchao Liu, Volume 36 in Challenges and Advances in Computational Chemistry and Physics, Springer (2024). Other books will be suggested during the lectures.
Assessment methods and Criteria
The exam will consist of an about 30-minute oral examination on several topics dealt with during the course. A final grade will be proposed and the exam passed if the grade ranges from a minimum of 18/30 to a maximum of 30/30 with possibility of honors (cum laude). Lower grades will be considered not sufficient to pass the exam.
Professor(s)
Reception:
To be agreed via email. Please send an email to [email protected]
Department of Chemistry, First Floor, Sector A, Room 131O