BSc in Computer Science / Бакалавр в Області Комп'ютерних Наук

Course Details

Course Information Package

Course Unit TitleNEURAL NETWORKS
Course Unit CodeACSC402
Course Unit Details
Number of ECTS credits allocated5
Learning Outcomes of the course unitBy the end of the course, the students should be able to:
  1. Define and explain the theoretical foundations of Neural Networks and Genetic Algorithms.
  2. Evaluate the strengths and weaknesses of Neural Networks and Genetic Algorithms and recognize the situations where each technique can be applied successfully.
  3. Point out, explain and propose ways of dealing with the issues involved in the application of Neural Network and Genetic Algorithm techniques.
  4. Compare and evaluate different types of Neural Networks and Genetic Algorithms and identify the most appropriate one for a given problem.
  5. Design, implement and apply a suitable Neural Network or Genetic Algorithm for solving a particular problem and evaluate and report the results appropriately.
Mode of DeliveryFace-to-face
PrerequisitesNONECo-requisitesNONE
Recommended optional program componentsNONE
Course Contents -  Introduction to Neural Networks: Biological motivation; Classification and regression problems; Characteristics of Neural Networks; General structure of neurons.
-  Perceptrons: Computations of a perceptron; Representational interpretation and limitation; perceptron learning algorithm; Gradient descent; Delta learning algorithm.
-  Multilayer Neural Networks: Characteristics; The log-sigmoid, tan-sigmoid and softmax activation functions; The backpropagation training algorithm; Representational power; Issues: local minima and overfitting; Momentum term.
-  Self-Organizing Maps: Unsupervised learning; Competitive learning; Structure of SOM networks; Kohonen’s learning algorithm; Adding bias to learning.
-  Genetic Algorithms: Biological motivation; Search spaces; General structure of a genetic algorithm; Chromosome representation; Fitness function; Selection methods; Genetic operators; Convergence and diversity.
-  Applications of Genetic Algorithms: Combinatorial optimization with GAs; Solving pattern recognition problems with GAs; Hypothesis chromosome representation; Evolving neural network weights.

Recommended and/or required reading:
Textbooks
  • Simon Haykin, Neural Networks: A Comprehensive Foundation, 2nd edition, Prentice Hall, 1998.
  • Melanie Mitchell, An Introduction to Genetic Algorithms, MIT Press, 1998.
References
  • Christopher M. Bishop, Neural Networks for Pattern Recognition, Oxford University Press, 1996.
  • David E. Goldberg, Genetic Algorithms in Search, Optimization, and Machine Learning, Addison-Wesley, 1989.
  • Tom M. Mitchell, Machine Learning, McGraw Hill, 1997.
Planned learning activities and teaching methods The course is delivered through three hours of lectures per week, which include presentation of new material and demonstration of concepts and algorithms. Lectures also include in-class exercises to enhance the material learning process and to assess the student level of understanding and provide feedback accordingly.

Furthermore a lot of work is in done through homework and private study by carrying out the computations of the different techniques for specific inputs and by experimenting in Matlab with the application of these techniques to benchmark datasets. This provides students with practical experience on the ideas and issues discussed in class.

All lecture notes and other material is available to students through the course homepage.

Assessment methods and criteria
Assignments24%
Tests16%
Final Exam60%
Language of instructionEnglish
Work placement(s)NO

 Друк  E-mail