ÍÃ×ÓÏÈÉú

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:

1. Explain the basic principles of soft computing techniques for optimization
2. Describe available algorithms and evaluate their suitability for diverse optimization problems
3. Implement single-solution-based and population-based soft computing techniques to solve optimization problems
4. (Graduate students only) Implement a hybrid method to solve a (multi-objective) optimization problem and analyze experimental results