Artificial Intelligence for Engineering Applications

26 – 30 August 2024

This summer school focuses on harvesting the power of AI to extract actionable information from raw data. The course integrates AI principles with engineering applications, offering students the opportunity to learn advanced computational techniques and apply them to complex engineering challenges. This synergy of AI and engineering is designed to empower students with the skills to innovate and optimize in the field of engineering and beyond.

image generated with AI: Nanci ©- stock.adobe.com

Keywords: artificial intelligence, supervised inference, unsupervised inference, statistical analysis, learning algorithms, model evaluation

APPLICATION OPEN
Location

TU Delft, the Netherlands

Participants
Application is open to Master and PhD Students of the member universities from the IDEA League Alliance.
Expenses
There are no registration and accommodation fees. Students from IDEA League member universities selected to participate in a summer school only have to pay for their own travel costs where applicable.
Requirements

Curriculum vitae & publications list
Letter of motivation
Letter of recommendation (optional)
Supervisor approval (for PhD students from Chalmers)

Course Content

Data-driven decision-making is becoming a crucial skill in dealing with engineering systems that generate vast amounts of data from the automation system. Data science is improving our understanding of complex phenomena even faster than physical models have done in the past. Engineering Systems are composed of many complex elements, and their mutual interaction is not trivial to evaluate and predict by adopting the conventional first principles physics-based models.

This course will exploit advanced statistical techniques to build models directly based on the large amount of historical data collected by the recently advanced automation systems without prior knowledge of the underlying physical system. The course will focus on Artificial Intelligence (AI), Machine Learning (ML), and Data-driven models (DDMs) for engineering applications, including linear and non- linear models, shallow and deep models, and the best practices for model selection and error estimation. Numerical examples and real-life problems will be proposed and analysed, from bearings fault prediction, to fuel consumption optimisation. All course material will be freely available in PDF format for a complete understanding of the related subjects as well as for future consultation. During the afternoon session, hands-on workshops will be delivered with numerical examples focused on various aspects of AI, ML, and DDMs.

The course is designed for students interested in data analysis with an engineering background, numerical skills, and a rudimentary understanding of statistics. This course covers methodologies necessary for inferring useful information and identifying underlying patterns from raw, incomplete, noisy, and corrupted data that is present in real-life engineering applications. This is achieved by introducing concepts and methods used to model a wide range of systems based on available data.

Learning Objectives

On completion of the course the student is expected to be able to achieve the fol- lowing Learning Objectives (LO):

  • LO 1: Describe several models for supervised and unsupervised inference from data. Critically evaluate statistical analysis. Critically assess the fit of statistical models.
  • LO 2: Assess the strengths and weaknesses of each of these models and inter- pret the mathematical equations from linear algebra, statistics, and probability theory used in the learning models.
  • LO 3: Implement efficient learning algorithms in the MATLAB language, applied to engineering problems.
  • LO 4: Design test procedures to evaluate the model hyperparameters (model selection) and its error (error estimation). Develop an appropriate exper- imental research design for an engineering case study considering practical limitations.

Prerequisites

The course is designed for students interested in data analysis and machine learning applications. An engineering background and statistical and numerical skills would be beneficial but not necessary.

Course Exam

To ensure that students grasp theoretical concepts and excel in their practical application, the course’s learning objectives will be assessed through a comprehensive coursework project.

Objective: The coursework is designed to assess students’ understanding of theoretical AI principles and their ability to develop code that addresses tangible engineering challenges. It offers a platform for students to integrate and apply their knowledge in a practical, real-world context.

Introduction and Discussion: The coursework will be introduced and thoroughly discussed on the last day of the course, which is scheduled for 30 August 2024. This session will provide students with a clear understanding of the coursework requirements, the scope of the problems to be addressed, and the expectations regarding coding and analytical skills.

Duration and Submission: Upon introduction, students will have three weeks to complete and submit their coursework. This timeframe allows students to engage with the problem thoroughly, apply their skills effectively, and develop a well-considered solution.

Components of Submission: The submission will consist of two primary components: 1) a detailed report where the students articulate the problem-solving approach, the application of AI methods, and the analysis of the results; 2) a Developed code well-documented, highlighting the application of AI techniques learned during the course.

Assessment and Grading: Both the report and the code will be critically assessed to gauge the student’s proficiency in applying AI methods to solve engineering problems. The grades reflect students’ ability to integrate theoretical knowledge with practical skills.

Programme

The programme for our upcoming course is structured to maximize learning and hands-on experience. Each day is divided into two primary sessions, as indicated in the table below. The mornings are dedicated to theoretical understanding, wherein participants will engage with core concepts and principles. This will be followed by a break, allowing time for reflection and networking. Post lunch, the afternoons are committed to hands-on practice, where participants will apply their morning lessons to real-world scenarios and data sets.

Monday 26th August Tuesday 27th August Wednesday 28th August Thursday 29th August Friday 30th August

9:00-12:00

Theoretical Theoretical Theoretical Theoretical Theoretical
12:00-13:00 Break Break Break Break Break
13:00-16:00 Hands-on Hands-on Hands-on Hands-on Hands-on

Suggested Readings

The following readings are recommended for further exploration of the topics covered in the course:

Y. S. Abu-Mostafa, M. Magdon-Ismail, and H.T Lin. Learning from data. Vol. 4. AMLBook New York, 2012.

C.M. Bishop. “Neural Networks for Pattern Recognition”. In: Clarendon Press google schola 2 (1995), pp. 223–228.

C. Borgelt, F. H ̈oppner, and F. Klawonn. Guide to intelligent data analysis. Vol. 128. Springer, 2010.

T. Hastie, R. Tibshirani, and J. H. Friedman. The elements of statistical learning: data mining, inference, and prediction. Vol. 2. Springer, 2009.

N. Japkowicz and M. Shah. Evaluating learning algorithms: a classification perspective. Cambridge University Press, 2011.

D. C. Montgomery and G. C. Runger. Applied statistics and probability for engineers. John wiley & sons, 2010.

H. Pham. Springer handbook of engineering statistics. Springer Nature, 2023.

V. Vapnik. The nature of statistical learning theory. Springer science & business media, 1999.