1 / 21

Intelligent Systems

Intelligent Systems. Lecture 8 Acquisition of knowledge and learning, tools of knowledge engineering. What is knowledge acquisition?. knowledge acquisition without computer Methodic methods of acquisition of knowledge from experts during interview, dialog, questionnaire design and others

clancy
Download Presentation

Intelligent Systems

An Image/Link below is provided (as is) to download presentation Download Policy: Content on the Website is provided to you AS IS for your information and personal use and may not be sold / licensed / shared on other websites without getting consent from its author. Content is provided to you AS IS for your information and personal use only. Download presentation by click this link. While downloading, if for some reason you are not able to download a presentation, the publisher may have deleted the file from their server. During download, if you can't get a presentation, the file might be deleted by the publisher.

E N D

Presentation Transcript

  1. Intelligent Systems Lecture 8 Acquisition of knowledge and learning, tools of knowledge engineering

  2. What is knowledge acquisition? • knowledge acquisition without computer • Methodic methods of acquisition of knowledge from experts during interview, dialog, questionnaire design and others • knowledge acquisition with computer • Data mining or intelligent data analyzing • learning

  3. Learning systems Knowledge level learning (KLL) Symbol level learning (SLL) Analytical Empirical Non-deductive knowledge level learning (NKLL) (induction) Brief Deep Knowledge driven Data driven By examples By set of examples By single example (by explanation) Conceptualization (clusterization, classification) By analogy By discovery Main paradigms of learning and knowledge acquisition in knowledge engineering

  4. What is induction? Let the following facts: Titmouse is a bird, Sparrow is a bird, Crane is a bird, Titmouse is able to fly, Sparrow is able to fly, Crane is able to fly, Then the system having inductive capabilities may be to make conclusion that A bird is able to fly.

  5. Example of method of induction – DSM-method Let ai – potential reason bj – potential consequence M+ - matrix of confirming of hypothesis of causal dependence between ai and bj b1 b2…. bj… bm a1 h11 h12… h1j… h1m a2 h21 h22… h2j… h2m M+ = ai hi1 hi2… hij… him an hn1 hn2… hnj… hnm hij = kij+ / kij, kij+ -number of facts confirming of hypothesis, kij – number of all facts (experiments)

  6. What is needed define to define of induction • Method and form of presentation of facts • Method of inductive inference (algorithm of processing of facts) • What and in which form we want to obtain • Conditions of finishing of inductive inference • Method (algorithm) of extraction of desired rules from results of inductive inference

  7. Classifications of tools of knowledge engineering • Universal algorithmic programming languages • Object-oriented programming languages (C++, Object Pascal, Java) • Logical programming languages (Prolog, Lisp, Smalltalk) • Toolboxes for AI developers (e.g. components for programming of expert systems or neural networks) (OPS 5, NeuralBase) • Universal shells (or development environment) (e.g. Expert Shells) (ESWin, ExSys, G2, CExpert, CLIPS, Kee, Leonardo, NExpert Object, ART*Enterprise, Kappaб, Flex) • problem-specificor domain-specificshells (or development environment), in particular, empty expert systems (EMYCIN)

  8. Expert shell ESWin (Insycom Ltd.) • Consist of: • Language for description of knowledge base • Expert shell for developer • Expert shell for end user • Two kinds of editors of Knowledge Base (KB) • Program utility for view and diagnostics of KB • Program utility for improvement of structure of KB • Solving of task by backward fuzzy inference • Aims to using for development of expert systems for diagnostics, identification, support of making of decisions

  9. Sources of facts • Dialog with user • Databases by SQL-query forming automatically during dialog • External special programs been developed in case that possibilities of ESWin is not enough for solving of task • For example, as external program may be any neural network

  10. Knowledge base • Consist of: • TITLE = ‹name of Expert System› • COMPANY = ‹name of organization - owner of ES› • Frame named as Goal with names of tasks solved by expert system • Other frames describing of domain • Rules for solving of tasks • Descriptions of linguistic variables (if ones are used in expert system)

  11. Structure of frame FRAME (‹ type of the frame ›) = ‹ a name of the frame › PARENT: ‹ a name of the frame - parent › OWNER: <a name of the frame - owner> ‹ A name of slot 1 › ….. …. ‹ A name of slot i › (‹ type of slot ›) [‹ a question of slot ›?] {‹ the comment of slot ›}: (‹ value 1 ›; ‹ value 2 ›; …; ‹ value m ›) … ‹ A name of slot n › ….. ENDF • Types of frames: • Class • Instance • Template • Types of slots: • Symbol • Number • LV – linguistic variable

  12. Structure of rules Relations in conditions can be: EQ or = Equally; GT or > It is more; LT or < It is less; NE or <> Not equally; IN Two frames are connected by the relation "part of" (there is a connection through slot OWNER). Relations in the conclusions can be: EQ or = Equally (creation of the fact - slot in a frame-instance); IN Inclusion in the frame-owner (creation of connection - slot OWNER in the frame-instance); DL Remove of slot in a frame-instance; EX Execute of the external program; FR Output of a frame-instance; GO Execute of a rule; MS Output of the message to the screen; GR Output to the screen of a graphic file (formats *.gif, *.avi or *.htm). RULE ‹number› ‹condition 1› ‹condition 2› … ‹condition m› DO ‹conclusion 1› ‹conclusion 2› … ‹conclusion n› ENDR

  13. Linguistic variables • Parameters describing of ones: • Name • Set of symbolic values • For every symbolic values • Minimal numeric value • Maximal numeric value • Number of values of membership function • Set of values of membership function

  14. Example of KB in ESWin (fragment)

  15. Expert shell ESWin for developers

  16. Expert shell ESWin for developers

  17. Editor of knowledge base

  18. Editor of KB (editing of linguistic variables)

  19. Other editor of KB

  20. G2 (Gensym inc. ) Gensym's underlying product technology is built around its powerful G2 rule engine platform that drives critical business and operations decisions. Unlike other rule engines, G2 works in real time and is proven in the most demanding mission-critical applications. G2 combines real-time operational data, business policies, and expert knowledge to automate decisions and actions.

  21. Products based on platform G2 • G2 Optegrity for Manufacturing - for predictive abnormal condition management applications in the process manufacturing industries. • G2 NeurOn-Line for Manufacturing - for predicting, controlling, and optimizing the product quality of complex manufacturing processes through the use of real-time neural network and rule-based technology. • G2 ReThink for Business Processes - for rule-driven, real-time business process management. • G2 e-SCOR for Supply Chains - for rules-driven, supply-chain modeling and simulation. • G2 webSCOR.com for Supply Chains - for collaborative supply-chain design via the Web. • G2 Integrity for Networks - for maximizing the availability and service quality of telecommunication networks.

More Related