06.03.2013 Views

Artificial Intelligence and Soft Computing: Behavioral ... - Arteimi.info

Artificial Intelligence and Soft Computing: Behavioral ... - Arteimi.info

Artificial Intelligence and Soft Computing: Behavioral ... - Arteimi.info

SHOW MORE
SHOW LESS

You also want an ePaper? Increase the reach of your titles

YUMPU automatically turns print PDFs into web optimized ePapers that Google loves.

like p <strong>and</strong> negation of p. The results of the reasoning process considered in<br />

this chapter differ from that of McDermott's logic [25] on the following<br />

considerations. In McDermott's logic, there exists two stable points in a<br />

system represented by p→ ¬q, <strong>and</strong> q→¬p, whereas the present method leads<br />

to only one stable point, involving either p or q, depending on the initial fuzzy<br />

beliefs of p, q <strong>and</strong> their supporting evidences.<br />

Exercises<br />

1. For the FPN given in fig. 10.3, identify the P, Q, P / fm, Q / fm matrices.<br />

Assuming that the relational matrices associated with the transitions to be<br />

the identity matrix <strong>and</strong> an arbitrary belief vector N(0), compute N(2) by<br />

an appropriate forward reasoning model. What guideline should you<br />

suggest to identify the appropriate reasoning model for a given FPN?<br />

2. Identify the cycle in the FPN of fig. 10.6 by using the algorithm for cycle<br />

detection.<br />

3. From the given belief vectors n5 <strong>and</strong> n7 in the FPN of fig. 10.8, determine<br />

the belief vectors n1 <strong>and</strong> n2 by using the backward reasoning algorithm.<br />

Assume that the relational matrices are I.<br />

4. Prove that for a purely cyclic net (P o Q) k = I when k = number of<br />

transitions in the cycle.<br />

5. Given that pre- <strong>and</strong> post-inverse of matrix I is I. Hence show that<br />

reciprocity relations hold perfectly for a purely cyclic net. Also show that<br />

Rfm <strong>and</strong> Rbm for such net = I.<br />

6. Prove logically that a dual net can always be constructed by reversing the<br />

arrowheads in a primal net.<br />

7. Can you devise an alternative formulation of the fuzzy inversion of<br />

matrices? [open ended problem]<br />

8. Does the algorithm for computing fuzzy inverse apply to binary matrices?<br />

If yes, can you use it for diagnostic applications in switching circuits?<br />

[open ended problem] [32]-[33].

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

Saved successfully!

Ooh no, something went wrong!