Neural Networks 2011-2012
Academic Year 2011–2012
Instructor: Prof. R. De Leone
- Learning Outcomes
- Identify the general characteristics and the computational paradigm for Artificial Neural Networks
- Illustrate the main differences between various Artificial Neural Networks architectures’
- Illustrate the main algorithm for supervisioned and unsupervisioned training
- Utilize specific software for Artificial Neural Networks
- Artificial Neural Networks architectures and computational paradigm
- Characteristics of Artificial Neural Networks and classification
- The biological neuron and the artificial neuron: weights and activation functions.
- Artificial Neural Networks as universal approximators
- Static, dynamic and recurrent networs
- Perceptron and Multilayer Feedforward networks: the back-propagation algorithm
- Learning and generalization
- Unsupervisioned learning
- Support Vector Machines
- Final Exam Oral exam. A project will be assigned to each student.
- J. Hertz, A. Krogh, R.G. Palmer Introduction to the theory of neural computation, Addison-Wesley, 1991
- Further material would be made available during the course.