16.01.2013 Views

An Introduction to Genetic Algorithms - Boente

An Introduction to Genetic Algorithms - Boente

An Introduction to Genetic Algorithms - Boente

SHOW MORE
SHOW LESS

Create successful ePaper yourself

Turn your PDF publications into a flip-book with our unique Google optimized e-Paper software.

9.<br />

*<br />

10.<br />

*<br />

Compare the results of the experiment in computer exercise 7 with that of using random−mutation hill<br />

climbing <strong>to</strong> search for CA lookup tables <strong>to</strong> solve problem. (See Mitchell, Crutchfield, and<br />

Hraber 1994a for their comparison.)<br />

Perform the same experiment as in computer exercise 7, but use GP on parse−tree representations of<br />

CAs (see thought exercise 1). (This will require writing a program <strong>to</strong> translate between parse tree<br />

representations and CA lookup tables that you can give <strong>to</strong> your CA simula<strong>to</strong>r.) Compare the results of<br />

your experiments with the results you obtained in computer exercise 7 using lookup−table encodings.<br />

Figure 2.29 gives a 19−unit neural network architecture for the "encoder/decoder" problem. The<br />

problem is <strong>to</strong> find a set of weights so that the network will perform the mapping given in table<br />

2.2—that is, for each given input activation pattern, the network should copy the pattern on<strong>to</strong> its<br />

output units. Since there are fewer hidden units than input and output units, the network must learn <strong>to</strong><br />

encode and then decode the input via the hidden units. Each hidden unit j and each output unit j has a<br />

threshold Ãj. If the incoming activation is greater than or equal <strong>to</strong> Ãj, the activation of the unit is set <strong>to</strong><br />

1; otherwise it is set <strong>to</strong> 0. At the first time step, the input units are activated according <strong>to</strong> the input<br />

activation pattern (e.g., 10000000). Then activation spreads from the input units <strong>to</strong> the hidden<br />

Figure 2.29: Network for computer exercise 10. The arrows indicate that each input node is connected <strong>to</strong> each<br />

hidden node, and each hidden node is connected <strong>to</strong> each output node.<br />

Table 2.2: Table for computer exercise 10.<br />

Input Pattern Output Pattern<br />

10000000 10000000<br />

01000000 01000000<br />

00100000 00100000<br />

00010000 00010000<br />

00001000 00001000<br />

00000100 00000100<br />

00000010 00000010<br />

Chapter 2: <strong>Genetic</strong> <strong>Algorithms</strong> in Problem Solving<br />

00000001 00000001<br />

units. The incoming activation of each hidden unit j is given by ia i wi, j, where ai is the activation of input<br />

unit i and wi, j is the weight on the link from unit i <strong>to</strong> unit j. After the hidden units have been activated, they in<br />

63

Hooray! Your file is uploaded and ready to be published.

Saved successfully!

Ooh no, something went wrong!