A Methodology for Uncertainty in Knowledge-Based Systems by Kurt Weichselberger

By Kurt Weichselberger

In this booklet the resultant use of likelihood idea is proposed for dealing with uncertainty in professional structures. it's proven that equipment violating this advice could have harmful effects (e.g., the Dempster-Shafer rule and the tactic utilized in MYCIN). the need of a few specifications for an accurate combining of doubtful info in specialist platforms is tested and compatible principles are supplied. the chance is considered that period estimates are given rather than precise information regarding chances. For combining info containing period estimates ideas are supplied that are necessary in lots of cases.

Show description

Read Online or Download A Methodology for Uncertainty in Knowledge-Based Systems PDF

Best computers books

Compiler Construction

The complex path happened from March four to fifteen, 1974 and was once organised by way of the Mathematical Institute of the Technical college of Munich and the Leibniz Computing heart of the Bavarian Academy of Sciences, in co-operation with the ecu groups, subsidized by means of the Ministry for study and expertise of the Federal Republic of Germany and via the ecu learn workplace, London.

Logical Foundations of Computer Science: Third International Symposium, LFCS '94 St. Petersburg, Russia, July 11–14, 1994 Proceedings

This quantity offers the refereed papers approved for the foreign symposium Logical Foundations of desktop technological know-how '94, good judgment at St. Petersburg, held in St. Petersburg, Russia in July 1994. The symposium was once the 3rd in a chain of joint efforts of logicians from either the previous Soviet Union and the West.

New Concepts for Parallel Object-Relational Query Processing

Over the last few years, parallel object-relational database administration structures have emerged because the best facts administration know-how out there. those platforms are extensible by means of user-defined information kinds and user-defined performance for the knowledge. This paintings makes a speciality of the effective parallel execution of user-defined performance.

Additional info for A Methodology for Uncertainty in Knowledge-Based Systems

Example text

21), if the initial probabilities lie between their respective limits? Should we restrict ourselves to cases, for which basic probability numbers exist, two further methods can be suggested. 23) shows an important favourable quality which the Dempster-Shafer rule lacks. 21) can be sufficient at all. As was pointed out earlier one should generally distinguish between the use of such rules in statistical inference and their use in diagnostic systems. The characteristic feature of the latter is the combination of pieces of evidence, each of which consists of an observation of one relevant attribute.

We denote for the combined probability distribution P(Ei): Sup~P(Ei) = Ui pes t I n f P(Ei) = Li. PES* These upper and lower limits can be calculated by means of an algorithm, which is described in the Appendix. 22). This will be demonstrated in the following example, which uses the values of the preceding example. 6944. 6667]. [] The result of these comparisons demonstrates that the Dempster-Shafer rule may be a great deal too optimistic. 21) is always possible if the application of the Dempster-Shafer rule is possible.

Ell. ,k. 21). " n~l(Bil) with B (11 being the symbol for concluding from the first source to subset Bil, from the second to Bi2 and so on. For Bil . . . Bit and for Bil N ... Cl Bil --= ¢ an analogous procedure to that used above is obvious. ,Bil are neither identical nor disjunct. Any combination rule for basic probability numbers re(B) affords a decision as how to make allowance for m(B~ l, ), if B~ l> is of this kind. NBil. This seems quite natural, as one would proceed in the same way if 46 experiments were described.

Download PDF sample

Rated 4.02 of 5 – based on 13 votes