Intendierte Lernergebnisse
This lecture familiarizes students with the fundamentals of optimization and neural networks. Selected applications are considered in various fields of engineering including transportation.The general expectation regarding the knowledge to be provided/acquired is as follows:Mastering of the basics of optimization and selected applicationsMastering of some MATLAB Toolboxes (e.g. Linear programming and Quadratic programming toolboxes) and their application in solving linear and nonlinear optimization problems.Mastering of Recurrent Neural Networks and their application in solving linear and nonlinear optimization problems.Mastering of the development of simulation algorithms (based on Recurrent Neural Networks) for the solving of shortest path problems and traveling salesman problems in graph networks.Mastering of the computation based Neural Network: Application for solving concrete case studies of practical interest in engineering.
Lehrmethodik
The slides are available for the entire lecture. These slides are uploaded into the MOODLE system. The entire content of each slide is systematically explained by the lecturer.The Lecturer provides full explanation of how to write numerical codes to solve the exercises proposed in each chapter of the Lecture.Additional examples that are not included in the slides are suggested by the lecturer to allow a good understanding of the information provided.The slides contain exercises with solutions to allow a good understanding of the contents of each chapter. These solutions are systematically explained (during the lecture) by the lecturer.The Slides contain exercises without solutions to be solved by students during the lecture (this is part of oral exam). The students are fully assisted by the Lecturer in order to obtain correct/exact solutions to the proposed exercises. This will help to check whether the students have understood the chapters or not. Several exercises will be proposed by the Lecturer to be solved by students as projects (Homework). This will help to test the self-learning potential of students.
Inhalt/e
The lecture is organized around the following topics:Chapter 1. Basics of optimizationChapter 2. Simulation algorithms for optimizationChapter 3. Dynamic neural networks based simulation of Shortest Path Problems (SPP)Chapter 4. Dynamic neural networks based simulation of Traveling Salesman Problems (TSP)Chapter 5. Introduction to classical Artificial Neural Networks (ANNs)Chapter 6. Application of the classical ANNs for solving selected concrete application examples in engineering
Literatur
Textbooks [1] Martin Treiber, and Arne Kesting, „Traffic Flow Dynamics: Data, Models and Simulation,“ Springer-Verlag, Berlin Heidelberg, ISBN 978-3-642-32460-4, 2013[2]. F. M. Ham and I. Kostanic, „Principles of Neurocomputing for Science , & Engineering,“ New York, NY, USA: McGraw-Hill, 2001.[3] Adam B. Levy, „The Basics of Practical Optimization,“ SIAM, The society of industrial and applied mathematics, ISBN 978-0-898716-79-5, 2009[4] Nocedal J. and Wright S.J., „Numerical Optimization,“ Springer Series in Operations Research, Springer, 636 pp, 1999.[5] Saidur Rahman, „Basics of Graph Theory,“ Springer, ISBN: 978-3-319-49474-6, 2017Journal Papers [1] J. C. Platt and A. H. Barr, “Constrained differential optimization for neural networks,” American Institute of Physics, Tech. Rep. TR- 88-17, pp. 612-621, Apr. 1988.[2] I. G. Tsoulos, D. Gavrilis, and E. Glavas, “Solving differential equations with constructed neural networks,” Neurocomputing, vol. 72, nos. 10–12, pp. 2385–2391, Jun. 2009.[3] J.C. Chedjou, and K. Kyamakya, "A universal concept for robust solving of shortest path problems in dynamically reconfigurable graphs," Mathematical Problems in Engineering, 2015.[4] J.C. Chedjou, and K. Kyamakya, "Benchmarking a recurrent neural network based efficient shortest path problem (SPP) solver concept under difficult dynamic parameter settings conditions," Neurocomputing, Elsevier, pp. 175-209, Vol. 196, 2016.[5] J.C. Chedjou, K. Kyamakya, and N. A. Akwir "An efficient, scalable, and robust neuro-processor-based concept for solving single-cycle traveling salesman problems in complex and dynamically reconfigurable graph networks," IEEE Access, pp. 42297-42324, Vol. 8, 2020.