Alessandro Butté: Katalogdaten im Frühjahrssemester 2023 |
| Name | Herr Dr. Alessandro Butté |
| Adresse | Lehre Chemie u. Ang. Biowiss. ETH Zürich, HCI F 137 Vladimir-Prelog-Weg 1-5/10 8093 Zürich SWITZERLAND |
| buttea@ethz.ch | |
| Departement | Chemie und Angewandte Biowissenschaften |
| Beziehung | Dozent |
| Nummer | Titel | ECTS | Umfang | Dozierende | |
|---|---|---|---|---|---|
| 529-0017-00L | Chemometrics and Machine Learning for Chemical Engineers  | 6 KP | 3G | A. Butté | |
| Kurzbeschreibung | This course will offer a broad overview on several statistical techniques that can be applied in the field of (bio)chemical engineering for process modeling and experimental design. During the course, the student will be initially given basic statistical notions (variance, covariance, p-values, etc.), followed by an overview of the main so-called chemometric techniques, with particular focus on mu | ||||
| Lernziel | The course has the following objectives: 1.Introduce the student to the main statistical techniques that are typically used for research and industrial purposes, while emphasizing on the role that machine learning will play in the future. Several application examples from (bio)chemical engineering will be provided. 2.Provide some guidance to the choice of the statistical tools for different purposes, and to the pros and cons of such choice. 3.Provide major insights into such techniques, so to avoid most common errors and misusage of such techniques. 4.To some extent, demystify machine learning techniques as simple solution to all problems, highlight major limitations of such techniques when applied to (bio)chemical processes, and discuss the importance of integrating such techniques with theoretical knowledge. | ||||
| Inhalt | Lecture contents: 1. Course motivation and Fundamentals of Statistics 2. Linear regressions (incl lasso and ridge) 3. From Process Data to PCA 4. PLS (comparing also with PCR) 5. PLS (and PLS2) variable importance and advanced interpretation 6. Machine learning: general intro, supervised & unsupervised clustering, decision trees 7. Random Forests and Support Vector Machines 8. Artificial Neural Networks (ANN) and their Variants 9. Gaussian Processes (theory, application for regression, missing data) 10. Hybrid Models: Intro 11. Hybrid Models: Advanced application of Hybrid Models 12. Kalman filtering 13. Model-based experimental design versus classical DoE | ||||
| Skript | Before each class, the student will receive a PowerPoint presentation with the lecture. In the third hour of the lecture, an exercise will be presented to the students. The students are asked to solve the exercise in groups. The exercise will require the numerical solution of some problems using Matlab (or equivalent software). All main functions for the solution will be supplied. The solution of the exercise will be discussed during the next class. | ||||
| Literatur | 1. Practical Guide To Chemometrics, by Paul Gemperline (Editor), ISBN-13: 978-1574447835 2. Multivariate Analysemethoden, by Backhaus, K., Erichson, B., Plinke, W., Weiber, R. (Authors), ISBN-13: 978-3-662-46076-4. 3. Machine Learning Engineering, by Andriy Burkov (Authors), ISBN-13: 978-1999579579 | ||||
| Voraussetzungen / Besonderes | Numerical and statistical methods for chemical engineers. | ||||