Contents authors' list error: Reference source not found 9


N.I.Shanchenko INFORMATION INVESTMENT DESIGN SYSTEM



Download 1.06 Mb.
Page10/65
Date07.08.2017
Size1.06 Mb.
1   ...   6   7   8   9   10   11   12   13   ...   65

N.I.Shanchenko INFORMATION INVESTMENT DESIGN SYSTEM

The working of large body of information and complexity of applied algorithms make it necessary to use computers in the investment design process.

The purpose of information investment design system is to support fast and easy data input and investment design parameters calculation. Therefore this system must include such components, as data input and interface subsystems, calculation module, data and model banks.

Common requirement for information investment design system may be formulated as:



  • presence “friend” interface;

  • possibility for using investment design archives;

  • possibility for modeling of different conditions for design carrying out, including material and cash flow models;

  • existence of increased subject space model and calculation procedure banks;

  • possibility for many alternative design and sensitivity analysis;

  • visual representation of design results;

  • possibility for design background specifying.

The process of design description input is expedient to realize in the form of choice from available models such as:

  • tax models;

  • inflation models;

  • sale budget models;

  • production plan models;

  • cash and material flow models;

  • finance models.

To support most flexibility in the subject space modeling it is recommended to use corporate imitate model in the form of dynamic model of material and cash flows, considered as event chain.

In conclusion it may be noted, that information investment design system application allows to supply such qualities for design process, as accuracy and validity, detail and visual results representation, timely and alternate design.




I.N. Glukhikh IMITATION OF DECISION MAKING BY AN AIR TRAFFIC CONTROLLER IN INTELLIGENT TRAINING SYSTEMS: INTERACTIVE MODE FOR KNOWLEDGEBASE MAKING

The providing of safe distance between airplanes in potentially dangerous situations is a very important task of air traffic control. This problem is solvied by an air traffic controller (ATC). For training of ATC it is reasonable to use intelligent solving systems (ISS), which allow to make an automatic decision in potentially dangerous situations for demonstrations of ways of conflict prevention between airplanes. The term «decision» here is referred to a new program of flight, which allows to avoid a conflict with other planes.

There is a possible approach to decision-making, at which educating of ISS can be realized in interactive mode with constructing a base of situations examples and decisions examples (BSD). Decision in current situations can be made as the result of transformations of decision examples from BDS. The formal presentation of situation examples and decision examples is based on notion of events multivector. Any situation is considered as a projection of general situation into its coordinate system, and it is possible to find a formal transformation, which allows to get a current situation from the example of situation. It is possible to take a decision of current situation from the example of decision by means of these transformations as well. The base models of transformations of decision examples with different restrictions are designed.

For checking serviceability of models programm module of shaping and checking the examples is designed. Shaping the examples is executed in the real-time mode of imitation by commands, which ATC uses in practice.




P.I.Sosnin CONSTRUCTIVE PREDICATION AS A TASK OF PATTERN RECOGNITION

The use of precedent {di } begins from the generation of a protocol of interaction prj with an actual or a potential situation Stk , in which one is going to act. The protocol prj registers the results of interaction in a certain language that are obtained with the help of a final set of means of observation and/or measurement. The aims of interaction are detection and identification of a connected collection of objects included into Stk . The typical form of registration is the text description Stk of the present situation.

The interpretation of each clause Clm of the text Tl is offered as hypothesis fixing an outcome of pattern recognition, detected during the interaction with Stk .

Let's present the interpretation for simple clauses of texts, analysing their perception in terms of base problems of automated pattern recognition.



1.The choice of an informative set of attributes. The active interaction of a concrete person with his surroundings results in his generating and mastering of a unique (personal) system of operational definitions S(gn). His organs of sense and measuring instruments he knows and uses lie in the basis of this system. The system S(gn ) finds its representation in the language of the person. Each of the definitions gn of the system is capable of executing the role of an attribute.

The system S(gn) is constantly developed and improved forming a dynamic system S(gn ,t). Due to frequent uses it forms the "library of procedures " that includes not only procedures of operational definitions but serving procedures as well, particularly the procedures of the set-up of the system S(gn ,t) for a concrete situation. The set-up is aimed at selecting a priori set of attributes that corresponds to Stk. This set will be used in the process of interaction with Stk and the value will be assigned to each of the components. The set will receive some information load and the status of primary measuring information. The vector of attributes for the pattern will be included into this set.



2. Forming of a training set. In persons constant interaction with his surroundings the system S(gn,t) is actively used for structuring the surroundings. The results of this structuring are coded and processed, forming a set of applications of the system S(gn,t). These applications can be also used in safe situations and/or in useful (successful) persons reactions. They are connected and coordinated with the system of precedents, forming a tested set of attributes ck(gnj). Each of them points to a certain element of the appropriate situation Stk.

The set ck(gnj) makes the experimental base of sample used by the person for pattern recognition. The typical way of using ck(gnj) in the training mode is taxonomy.



3. Taxonomy. A system of fixed groups (taxons, classes) are formed on the set of operatively determined sets of attributes. Such system is called a system of concepts S(Ni,t). The system of concepts is individual as well. It is developed and improved in accordance with declarative and procedural components, generating the base of names Ni for registering the results of recognition.

The set of names Ni consists of the following subsets: subsets Nik, naming the elements of the set, ck(gnj); names Nil indicating constructions from ck(gnj) according to the laws of indirect and joint measurements; names Nim assigned to abstract objects; names Nin in which some elements from such sets as Nik, Nil, Nim, Nin were used.

Each of the names Ni indicates an appropriate class the current ‘’definition’’ of which is built in accordance with the results of recognition. The set of the previous acts of recognition makes a training sample on the basis of which the system S(Ni, t) is constructed.

4. Preliminary preparation for recognition. Under certain circumstances (Stk) the person includes the mechanisms of recognition into action. It means that he activates the systems S (gn, t) and S (Ni, t). Such usage is registered by the protocol ‘’the descriptions Stk”. This protocol includes the codes of results of object recognition into the text Tl.

The elementary unit of registration is a simple clause Clm in the recorded language. In any natural language such clause consists of a predicate group Gvlm and a noun group Gnlm. Both of them are combined by a predicative connective.

The basic requirement to a clause is the requirement of ‘’conformity to reality ’’, the performance of which indicates the correctness of the connective Gvlm and Gnlm. The correctness should be confirmed by a special test which is called “predication” in linguistics. If the correctness is confirmed then the value of validity is assigned to the clause.

If the initial registration of the clause is understood as a hypothesis hlm that requires substantiation, its function can be reasonably assigned to the act repetition of recognition. Such repetition must be realized through some methods of pattern recognition. If the repetition uses instrumental support, it represents a constructive version of a predicative process.

The repetition should be started from a repeated gathering of the primary measuring information according to the code Clm. First of all it is necessary to initiate the gathering and testing of the primary measuring information registered in the group of the predicate Gvlm. In the process of such actions the set of informative attributes can be changed.

The gathering of the primary measuring information should be accompanied by its processing. Such work should be realized through automated pattern recognition (for example, to reduce information noise and the degree of uncertainty).



5. Approach to the algorithm of recognition. After gathering the primary information about attributes and its processing, we must go to the final step of the class definition. As a base of such actions we must use the validity of concept usage on which indicates the code of the noun group. Not only the confirmation of the code Gn1 but its modification as well can be positive if it indicates the validity of the concept usage. This step can result in changing the code Gnlm as well as the code of the clause Gnlm that is practically proved.

It is valid to use the search and adaptation of the “precedent of the concept usage” in algorithm recognition. Such algorithm can be easily adopted to the case of recognition when the initial code was wrong.



6. Conclusions. The previous comparisons of predication and pattern recognition are easily transformed to clauses of any type (compound clauses consist of simple clauses). Consequently, predicative processes admit their modeling with the help of pattern recognition. Analogies and borrowings from the experience of automated pattern recognition are put into the basis of constructive predication. This experience should be adopted to generating, transforming, processing and using of some textual information in order to get access to the precedents. The parallels with pattern recognition prompt another analogue for interaction with the textual information, the functions of which are capable of executing the automated image analysis.

The analogies and borrowings from the experience of pattern recognition and image analysis open a new and a perspective direction of investigations and developments in interactive human-computer interaction.





Directory: confer
confer -> Tilte : a critical examination of the police relations with bbc
confer -> Eavesdropping on a virtuous circle Richard Whately and the Oriel Noetics. Elena Pasquini Douglas uwa business School
confer -> Simulation and Prediction of Storm Surges, Waves, and Morphological Changes due to Tropical Cyclones by Using a pc-based Integrated Coastal Process Model
confer -> Panel 0511 Disability and Difference I: Post-War Journeys through Disability
confer -> Do remittances have a flip side? A general equilibrium analysis of remittances, labor supply responses and policy options for Jamaica* Maurizio Bussolo and Denis Medvedev
confer -> Conference approval process made easy for acm in cooperation conferences
confer ->  Proceedings of gt2009 asme turbo Expo 2009: Power for Land, Sea and Air Orlando, Florida, USA gt2009-59981 dynamics of premixed h2/CH4 flames under near blowoff conditions
confer -> South Korea’s Economic Future: Industrial Policy, or Economic Democracy?
confer -> Asset rotator Cuff Rehabilitation Course Faculty Dr. Spero Karas
confer -> Acm word Template for sig site

Download 1.06 Mb.

Share with your friends:
1   ...   6   7   8   9   10   11   12   13   ...   65




The database is protected by copyright ©ininet.org 2020
send message

    Main page