You have 1.5 hours. Good luck!
1a. Brooks made a few radical claims in his classic paper Intelligence without Representation. Explain one of these claims. Do you think that he was right or wrong? Explain your opinion.
1b. Blank attempted to elevate the ideas of the Turing Test to the level of a formal definition of "intelligence." In his argument, he gave two options:
void crossover() {
int i, j, p, g;
int p1, p2, p3;
for (i = 0; (float)i < CROSSOVERRATE * (float) POPSIZE; i++) {
p1 = random_bias_good(POPSIZE);
p2 = random_bias_good(POPSIZE);
p3 = random_bias_bad(POPSIZE);
g = random(GENESIZE);
p = p1;
for (j = 0; j < GENESIZE; j++) {
if (j > g)
p = p2;
pop[p3][j] = pop[p][j];
}
}
}
2a. Rewrite the code so that it only does selection but not
crossover. Could evolution work after such a change (assuming that the
rest of the gahb.c program remains unchanged)? Explain.2b. Rewrite the code so that it has two crossover points rather than just one. For example, if two parents looked like:
111111111111111111111111111111111111 000000000000000000000000000000000000Selecting two points (the spots marked with ^) gives:
111111110000000000001111111111111111
000000001111111111110000000000000000
^ ^
Notice that this method produces more thoroughly mixed-up
offspring. How will this effect evolution? Explain.
3a. Describe "the law of uphill analysis and downhill invention". What did he mean? Explain.
3b. John Holland claims that genetic algorithms work via an "implicit parallelism." What did he mean? Explain.