Information Technology Reference
In-Depth Information
5.5 Exercises
1. Find specifi c behavior for some of the IN models based on diff erential
equations for some particular values of the parameters.
2. Design a specifi c learning rule for a parallel distributed processing (PDP) IN.
3. Use a specifi c learning algorithm to design a specifi c PDP IN.
4. Defi ne IN basic elements such as shape-space and a
nity measures for
specifi c problems.
5. Ask some specifi c questions about particular diff erential equation models,
such as under what conditions is stability reached? What is the behavior of
the system in the absence of antigen? What are the attractors of the dynamical
system? Also, answer specifi c questions such as the number of equilibrium
points and limit cycle.
6. For AIN discrete models, also suggest some minor modifi cations to some
of the models and fi nd out how such changes modify the behavior of the
network.
7. D e s i g n s o m e s p e c i fi c AINs to solve toy problems such as
a. Optimization of real value functions on 1D, 2D, or 3D
b. Discrete optimization problems
c. Small classifi cation problems
5.6 General Questions
1. What is the general model of an IN, also including B cell-T cell
cooperation?
2. How to model an AIN structure? (A straightforward way representing it as a
graph in which nodes will correspond to antibodies)
3. How to characterize the diff erent states of an antibody?
4. How to model an antigen?
5. How to specify the concept of an IN state?
6. How to model, in a general fashion, the process(es) that occur after the
interaction with either antigens or antibodies (immune response) in such a
way that the model conveys all diff erent IN models? Describe the adaptation
process that occurs?
7. Like neural networks, an IN dynamics should be characterized by a learn-
ing (adaptation) process that is typically a result of a stimulation process by
a set of antigens. How does an antibody send signals to other antibodies as a
result of the IN dynamics? How to express the total stimulation received by
an antibody?
8. Explain diff erent variations of clonal selection.
9. Distinguish between fuzzy AIS and a nonfuzzy version of AIN.
10. Describe the GAIN algorithm.
Search WWH ::
Custom Search