CSE 478/578 - Soft Computing Techniques for Optimization
Catalog description:
In this course, students will learn about different soft computing algorithms and how to apply them to various optimization problems. They will learn about the different approaches to problem representation and implementation and how to estimate parameters and analyze performance. Students will also have the opportunity to complete a semester-long project in which they apply a soft computing solution to a problem of their choice.
Prerequisites:
CSE 274 and (MTH 231 or MTH 331)
Required topics (approximate weeks allocated):
Common Concepts for Soft Computing in Optimization (3 weeks)
Optimization Models
Optimization Methods
Main Common Concepts for (Meta)heuristics
Constraint Handling
Parameter Tuning
Performance Analysis of (Meta)heuristics
Commercial Software Packages for Metaheuristics
Single-State Methods (3 weeks)
Common Concepts for Single-Solution Methods
Fitness Landscape Analysis
Local Search
Simulated Annealing
Tabu Search
Iterated Local Search
Variable Neighborhood Search
Guided Local Search
Other Single-Solution Based Metaheuristics
Population-Based Methods (3 weeks)
Common Concepts for Population-Based Methods
Evolutionary Algorithms
Common Concepts for Evolutionary Algorithms
Other Evolutionary Algorithms
Scatter Search
Swarm Intelligence
Ant Colony Optimization
Metaheuristics for Multiobjective Optimization (1 week)
Multiobjective Optimization Concepts
Multiobjective Optimization Problems
Main Design Issues of Multiobjective Metaheuristics
Fitness Assignment Strategies
Diversity Preservation
Elitism
Performance Evaluation and Pareto Front Structure
Hybrid Metaheuristics (2 weeks)
Combining Metaheuristics with Mathematical Programming
Combining Metaheuristics with Constraint Programming
Hybrid Metaheuristics with Machine Learning and Data Mining
Hybrid Metaheuristics for Multiobjective Optimization.
Parallel Metaheuristics (time permitting)
Parallel Design of Metaheuristics
Parallel Implementation of Metaheuristics
Learning Outcomes:
Explain the basic principles of soft computing techniques for optimization
Describe available algorithms and evaluate their suitability for diverse optimization problems
Implement single-solution-based and population-based soft computing techniques to solve optimization problems
(Graduate students only) Implement a hybrid method to solve a (multi-objective) optimization problem and analyze experimental results
Department of Computer Science and Software Engineering
