Link to Feedback Form |
Link to Lecture Slides
Links to Student Projects (partial)
Can a computer be taught to read words aloud,
recognize faces, perform a medical diagnosis,
drive a car, play a game, balance a pole, predict physical phenomena?
The answer to all these is yes. All these applications and others
have been demonstrated using
varieties of the computational model known as "neural networks", the subject of this
The course will develop the theory of a number of neural network models.
Participants will exercise the theory through both pre-developed computer
programs and ones of their own design.
Modeling, simulation, and analysis of artificial neural networks.
Relationship to biological neural networks. Design and optimization of
discrete and continuous neural networks. Backpropagation, and other gradient
descent methods. Hopfield and Boltzmann networks. Unsupervised learning.
Self-organizing feature maps. Applications chosen from function approximation,
signal processing, control, computer graphics, pattern recognition, time-series
analysis. Relationship to fuzzy logic, genetic algorithms, and artificial
Prerequisites: CS 60 and Mathematics 73 and 82, or permission of
the instructor. 3 credit hours.
- Robert Keller
242 Olin (4-5 p.m. MTuW or by appt.), keller@turing, x 18483
There will be some homework and programming
assignments, but no exams. These assignments will constitute about
50% of the grade. The other 50% of the grade is from a substantial
final project involving either a working neural network application or
a research paper. The grade on the project will be determined by the
comprehensiveness and degree to which you explored competing
approaches. The projects will be presented orally.
voluntary oral presentations on textbook material
can also be made during the term. These
can act to cushion your grade. They are very much encouraged, as it
they really help you learn the material at a higher level than you
Please see me if you are interested in
making a presentation.
CS 152 Topic Outline
- Week 1 (read MMR chapter 1)
- Contexts for and Motivation Neural Networks:
Artificial Intelligence |
- Artificial Neural Network overview
- Week 2 (read MMR chapter 2)
Supervised Learning: Single-Layer Networks
- Weeks 3-6 (read MMR chapters 3 and 4)
Supervised Learning: Multi-Layer Networks
- Multi-Layer Perceptrons (MLPs)
- Conjugate Gradient method
- Levenberg-Marquardt (LM) method
- Radial-Basis Networks
- Cascade-Correlation Networks
- Polynomial Networks
- Recurrent Networks
- Time series
- Backpropagation through time
- Finite Impulse Response (FIR) MLP
- Temporal Differences method (TD)
- Weeks 7-8 (read MMR chapter 5)
- Simple Competitive Networks: Winner-take-all | Hamming network
- Learning Vector Quantization (LVQ)
- Counterpropagation Networks (CPN)
- Adaptive Resonance Theory (ART)
- Kohonen Self-Organizing Maps (SOMs)
- Principal Component Analysis networks (PCA)
- Weeks 9-10 (read MMR chapter 6)
- Linear Associative Memory (LAM)
- Hopfield Networks
- Brain-State-in-a-Box (BSB)
- Boltzmann Machines and Simulated Annealing
- Bi-Directional Associative Memory (BAM)
- Week 11 (read MMR chapter 7)
- Neural Network Approaches
- Evolutionary Programming
- Week 12
Fuzzy logic and its connection to NNs
Martin T. Hagan,
Howard B. Demuth,
and Mark Beale,
Neural Network Design,
PWS Publishing Company, Boston, 1996, ISBN 0-534-94332-2.
This book was used in the 1996 offering of the course. It has excellent
tutorial content and is strongly keyed to the use of MATLAB. Matrix
formulations are used throughout.
Neural networks - A comprehensive foundation,
This book was used in the 1995 offering of the course.
It has a strong mathematical and signal-processing orientation.
- Mohamad H. Hassoun,
Fundamentals of artificial neural networks,
MIT Press, 1995.
This is another fairly thorough introduction.
- James A. Anderson,
An introduction to neural networks,
MIT Press, 1995.
This is a more gentle introduction to the topic, by one of the pioneers in the field.
Irwin B. Levitan and Leonard K. Kaczmarek,
Oxford University Press, 1991.
This book focuses on the biology and physics of neurons, if you wish to
know more about this aspect; it will not be emphasized in the course.
Marvin L. Minsky and Seymour Papert,
Perceptrons (expanded addition),
MIT Press, 1988.
The historical importance of this book will be discussed in the course.
Duda and Hart,
Pattern classification and scene analysis,
This book gives a broad look at pattern classification problems, but is not
on neural nets as such.
This is a comprehensive reference by the originator of this concept.
Neural networks and fuzzy systems : a dynamical systems approach to machine intelligence,
Prentice Hall, 1992.
This book compares fuzzy and neural approaches to control problems.
Genetic Algorithms + Data Structures = Evolution Programs,
Springer Verlag, 1996.
This book describes the evolutionary approach, which in some cases can achieve
results similar to neural approaches.
John R. Koza,
MIT Press, 1994.
This book focuses on the evolutionary approach to producing programs.
Christopher G. Langton (ed.),
Artificial Life, an Overview,
This is an early colllection of articles on the topic.
Neuro-Fuzzy and Soft Computing