BBM476 Artificial Neural Networks

Announcements

• The language of this course is English.
• This course does not cover the topics and applications of deep learning.
• Students taking this course must have some background on partial derivatives.
• Course notes are here in ppt.
• Least Squares Material ishere in pdf.
• Multiple dataset learning presentation is here in pptx.
• Linear Algebra course from Prof. Gilbert Strang
• Differential Equations course from Prof. Arthur Mattuck

The Goal of the Course

• To teach the black box models based on numerical data or experience
• To teach the notion of adaptive systems
• To teach -some of- the approaches in intelligent systems research
• To use a CAD software and CAD based simulation

Capabilities to be Gained

• Understanding the types of neural nets and the principles of how they work
• Understanding the principles of adaptive neural systems
• Understanding the notion of learning and the ways to obtain it
• Gaining the capability of CAD based studies

Instructor

• Prof. Dr. Mehmet Önder Efe

Grading

• Midterm 40%
• Final 60%

Resources

• M.Ö. Efe ve O. Kaynak, Yapay Sinir Ağları ve Uygulamaları, Boğaziçi Üniversitesi Yayınları, ISBN: 975-518-223-3, (No:696, pp.141), Birinci baskı: Ekim 2000.
• Haykin, S., Neural Networks, Macmillan College Printing Company, New Jersey, 1994.
• Bishop, C. M., Neural Networks for Pattern Recognition, Oxford University Press, 1995.
• Jang, J.-S. R., C.-T. Sun, E. Mizutani, Neuro-Fuzzy and Soft Computing, PTR Prentice-Hall, 1997.
• Narendra, K. S. and K. Parthasarathy, “Identification and Control of Dynamical Systems Using Neural Networks”, IEEE Transactions on Neural Networks, Vol. 1, No. 1, pp. 4-27, 1990.
• Ivan Nunes da Silva, Danilo Hernane Spatti, Rogerio Andrade Flauzino, Luisa Helena Bartocci Liboni, Silas Franco dos Reis Alves (auth.), Artificial Neural Networks : A Practical Course [1 ed.], Springer International Publishing, 2017.

Contents

• Historical Perspective/Tarihsel Perspektif
• Systems/Sistemler
  • Non-intelligent systems
  • Continuous and discrete system models
• Neuron and Its Analytic Model
  • Inner product as a similarity measure
  • Activation functions
  • Differentiability
  • Parameterization and operational aspects
  • Learning
• Least Mean Squares Algorithm
• Hopfiels Neural Network
• Perceptron Learning Algorithms
• Multilayer Perceptron (MLP)
  • Derivation of the learning algorithm
  • Error backpropagation
  • Memorization and generalization
  • Intervals and normalization
• Radial Basis Function Neural Nets
• Dynamical Neural Nets
• Feedback Nets
• Second Order Training Algorithms
  • Levenberg-Marquardt algorithm
  • Gauss-Newton algorithm
• Derivative Free Optimization
  • Particle Swarm Optimization (PSO) algorithm
• Federated Learning
• Applications of Neural Nets
  • Identification
  • Neurcocontrol
  • Noise elimination
  • Adaptive noise cancellation
  • Electronics
  • Medicine
  • Finance
• Stability in Adaptive Systems
• Reinforcement Learning
• Unsupervised Learning
• Multiple Dataset Learning