RC Logo
               TEQIP III National Short Term Course on

                            "Recent Advances in Applied Optimization"

                                               (December 24 - 28, 2018)

RC Logo

The course would provide knowledge in optimization techniques based on mathematical programming (MP) as well as computational intelligence techniques (CIT) so as to solve optimization problems in the context of the availability of a well-defined mathematical model and also in the context of the model being a black box. In Mathematical programming Techniques, the participants will learn to solve linear programming problems (LP), non-linear programming problems (NLP), mixed integer linear programming problems (MILP) and also mixed-integer non-linear programming problems (MINLPs). They will also be given exposure to state-of-the-art optimization tools in this domain through hands-on training sessions. In computational intelligence techniques, the course will cover the evolutionary as well as swarm optimization techniques. In particular, the focus will be on conventional techniques such as Genetic Algorithm (GA), Particle Swarm Optimization (PSO), Differential Evolution (DE) and Artificial Bee Colony (ABC). In addition the participants would be introduced to recently proposed techniques such as Sanitized Teaching Learning Based Optimization (s-TLBO) and Yin-Yang Pair Optimization (YYPO). The candidates would also get knowledge on the performance evaluation of single objective optimization techniques. The course will also lay emphasis on the development of single objective optimization algorithms and will encourage participants to conceive a single objective optimization algorithm on their own.

The course will also cover multi-objective optimization techniques from the perspective of mathematical programming techniques as well as Computational Intelligence Techniques. In Mathematical Programming Techniques, the course will focus on lexicographic optimization, weighted sum method and epsilon constraint method. From the perspective of Computational Intelligence Techniques, the course will introduce the participants to non-dominated sorting using crowding distance and also the epsilon domination method. In addition to the demonstration on the use of these two strategies to convert a single objective optimization technique into a multi-objective optimization techniques, that have been covered in this course (GA, TLBO, YYPO), the participants will be provided an opportunity to extend any of the recently proposed single objective optimization technique to multi-objective optimization problems.

The course will also address various other aspects such as handling of constraints using penalty function approach and also focus on various types of special variables such as integer and semi-continuous variables which are widely encountered in real-life optimization problems. A unique feature of the course, which is not covered in most textbooks, is that it will lay emphasis on appropriate modelling of the problem with respect to mathematical programming and computational intelligence techniques. The participant will be taught on the judicious use of appropriate variables and constraints to enable the development of a suitable model for the two categories of optimization techniques. The course will employ two complex combinatorial optimization problems viz., production planning and job-shop scheduling to demonstrate the appropriate modelling for solving using mathematical programming as well as computational intelligence techniques.

With increasing crisis in several aspects of our everyday life, from basic human necessities such as food and healthcare to advanced requirements in terms of ensuring geo-political security, it has become necessary to optimally utilize the limited resources while simultaneously maximizing the benefits and minimizing the negative consequences of our everyday activities. For example, an investor would possibly want to maximize the profit while the regulations would require the impact on environment to be minimal. Such problems occur not only in research but in fact occur almost in every activity of our life. The primary objective of this course is to help the participants address such common real-life situations by providing a formal optimization knowledge so as to help them to maximize the impact of their work. The hands-on session will give them an exposure to state-of-the-art tools in optimization and enable them to quickly adopt the learnings of the course and employ it in their regular academic, research and administrative activities. The course will also enable them to critically analyze the performance of both single and multi-objective optimization techniques and help them to propose new techniques or harness the advantages of various techniques to develop hybrid techniques for specialized applications. The course is designed to break some of the common myths that the optimization problems need to be postulated in terms of linear or non-linear equalities and inequalities. It is aimed that the course will also enable the faculty participants to offer Bachelors and Masters Projects in Computational Intelligence Techniques for the students in their own institute or at least explore involve formally incorporating optimization in their regular projects.

In many instances, innovations do not become commercially viable in a shorter duration despite their promising performance as these innovations need to compete with state-of-the-art technologies in industries and applications that have been optimized possibly over several decades. Very often there are multiple conflicting objectives that are required to be optimized and an approach of one-size-fits-all can prove to be detrimental. In view of the conflicting nature, it is not possible to determine a single solution but one needs to determine the entire set of trade-off solutions. Such trade-offs in the hands of a user will help them to maximize the use of innovation as per their immediate needs. Thus it has almost become inevitable for an innovator to formally optimize the performance of an innovation thereby making optimization an important tool and its working knowledge very essential. This course will enable the participants to quickly employ various optimization techniques wih the help of easy to use state-of-the-art optimization tools which can be readily incorporate in their current work. A strong emphasis on optimizing using exact mathematical models subject to its availability, and the use of black-box approaches in the absence of rigorous mathematical models will enable the participant to employ optimization under most circumstances. Computational Intelligence Techniques for optimization is one of the important areas of research and the course will enable the participants to not only extend currently available single objective optimization techniques to handle multiple conflicting objectives but will also enable them to propose novel optimization techniques.