International Journal of Pattern Recognition and Artificial Intelligence

Please refer to full tex.

A definition of expert systems is given, its pragmatic demands are cited and its structure is described. The methods and ways of pattern recognition are used in the subsystems DIALOGUE, ANALYTIC and HOMEOSTAT. The recognition algorithms which work on the information to be retained in the data base and knowledge base are described. The problems of recognition appearing under the construction of expert systems are noted.

The concept of K-stationarity of an environment in which a computer system (CS) works is introduced and the wide prevalence of such environments is pointed out. Their property of originating after a long period the problems belonging to a small number K of stable types leads to what is enough for a CS to only recognize the type of current problem. Thereafter the problem is solved by the algorithm previously prepared for that type. The problem of computer understanding of natural language (CUNL) can serve as an illustration. Any address made to a CS is a task, therefore a text analysis must reveal not its sense as in known constructions, but only the data about the task. However, in a K-stationary environment, those tasks can relate only to K known types, and this reduces the CUNL problem to the problem of recognition from a text of a proper algorithm (out of the K present) and of concrete values of its arguments, If a CS operates simultaneously in several subject areas (SA) then the stage of prerecognition of an SA is to be attached to the system. The CUNL systems under exploitation, which allow absolutely free and easy addressing within the SAs present, are described.

Peculiarities of expert systems for image analysis (ESIA) as well as their distinction from traditional expert systems are considered. Two interconnected components of the ESIA which simulate left and right cerebral hemisphere mechanisms of human visual perception are pointed out. The left formal-logical mechanism is simulated by a “conceptual hierarchy”, reflecting a problem area specialists’ knowledge. The right spatial-pattern mechanism is simulated by a “visual hierarchy” and an image processing expertise. An approach to ESIA development on the base of pyramidal-recursive structures as a way for image representation and for establishing a correspondence between elements of the two hierarchies is proposed.

We present a specification of the problem of combining evidence that arises in the approach to consistent object-labelling known as probabilistic relaxation. This specification differs from others in several important respects. Firstly, we ensure internal consistency by distinguishing between directly and indirectly interacting objects. Secondly, we avoid certain problems of interpretation and meaning by regarding the iterative updating of probabilities as a filtering process on the measurements for objects. Finally, we overcome the problem of the exponential complexity of the resulting evidence combining formula by deriving practical support functions of at most polynomial complexity. The computational realisation of the advocated approach to combining evidence is demonstrated for the application of edge labelling. This application includes details of how both the measurement generation process and the world-model may be expressed probabilistically and what the advantages are to be gained from using different forms of support-function.

In propositional logic (zero order) a system of logical rules may be put under the form of a conjunction of disjunction, i.e. a “satisfiability” or SAT-problem. SAT is central to NP-complete problems. Any result obtained on SAT would have consequences for a lot of problems important in artificial intelligence. We deal with the question of the number N of solutions of SAT. Firstly, any system of SAT clauses may be transformed in a system of independent clauses by an exponential process; N may be computed exactly. Secondly, by a statistical approach, results are obtained showing that for a given SAT-instance, it should be possible to find an estimate of N with a margin of confidence in polynomial time. Thirdly, we demonstrate the usefulness of these ideas on large knowledge bases.

The use of spatial knowledge is necessary in a variety of artificial intelligence and expert systems applications. The need is not only in tasks with spatial goals such as image interpretation and robot motion, but also in tasks not involving spatial goals, e.g. diagnosis and language understanding. The paper discusses methods of representing spatial knowledge, with particular focus on the broad categories known as analogical and propositional representations. The problem of neurological localization is considered in some detail as an example of intelligent problem-solving that requires the use of spatial knowledge. Several solutions for the problem are presented: the first uses an analogical representation only, the second uses a propositional representation and the third uses an integrated representation. Conclusions about the different representations for building intelligent systems are drawn.

In this paper, we discuss how rule based expert system shells can be used in pattern recognition for the implementation of algorithms which are procedurally oriented in their nature rather than rule based, and have traditionally been implemented in procedural programming languages. Particular examples include finite state automata, context free parsing, string matching, graph matching and discrete relaxation.

Stacking velocity is a very important parameter in seismic data processing. Until now the determination of stacking velocity has been done manually. This article proposes an automatic algorithm for picking stacking velocity. The algorithm uses artificial intelligence and pattern recognition techniques.

Sometimes programs, for multivariate statistical procedures are included into expert systems. The requirements of accuracy, exactness and reliability for such programs are very high. In this paper a new method for testing algorithms and programs of multivariate statistical procedures—the so-called “exact samples method” is introduced.

The programs of simple linear regression analysis from four most popular standard packages are tested and compared with the help of the new method.

We consider the class of logical decision rules and its applications for the solution of various problems of multivariate statistical analysis: discriminant and regression analysis, and cluster analysis. Some useful properties of the statistical analysis methods using the class under consideration are shown. Particular attention is paid to the possibility of presenting statistical results (unlike all other methods) in a language close to a natural language of logical statements.