Presentation is loading. Please wait.

Presentation is loading. Please wait.

D OSHISHA U NIVERSITY 13 November 20151 XML-based Genetic Programming Framework: Design Philosophy, Implementation and Applications.

Similar presentations


Presentation on theme: "D OSHISHA U NIVERSITY 13 November 20151 XML-based Genetic Programming Framework: Design Philosophy, Implementation and Applications."— Presentation transcript:

1 D OSHISHA U NIVERSITY 13 November 20151 XML-based Genetic Programming Framework: Design Philosophy, Implementation and Applications

2 D OSHISHA U NIVERSITY 13 November 20152 Outline 1.Introduction 2. Objective 3. Proposed approach 4. Verification results 5. Applications 6. Conclusion

3 D OSHISHA U NIVERSITY 13 November 20153 1. Introduction: the Problem The Needs A) Promptly developed software models of the evolved artifacts B) Fast running offline (phylogenetic) learning via simulated evolution (e.g. GP) A) Promptly developed software models of the evolved artifacts B) Fast running offline (phylogenetic) learning via simulated evolution (e.g. GP)

4 D OSHISHA U NIVERSITY 13 November 20154 The Needs A) Promptly developed software models of the evolved artifacts B) Fast running offline (phylogenetic) learning via simulated evolution (e.g. GP) A) Promptly developed software models of the evolved artifacts B) Fast running offline (phylogenetic) learning via simulated evolution (e.g. GP) The Reality A) Slow development time of evolutionary systems (specific semantics) B) Notoriously poor performance of GP (populations, generations, independent runs) A) Slow development time of evolutionary systems (specific semantics) B) Notoriously poor performance of GP (populations, generations, independent runs) Discrepancy, Gap 1. Introduction: the Problem

5 D OSHISHA U NIVERSITY 13 November 20155 Discrepancy, Gap The Needs A) Promptly developed software agents B) Fast running offline (phylogenetic) learning via simulated evolution (e.g. GP) A) Promptly developed software agents B) Fast running offline (phylogenetic) learning via simulated evolution (e.g. GP) The Reality A) Slow development time of evolutionary systems (specific semantics) B) Notoriously poor performance of GP (populations, generations, independent runs) A) Slow development time of evolutionary systems (specific semantics) B) Notoriously poor performance of GP (populations, generations, independent runs) A) Quicker development time GP B) Better performance characteristics of GP A) Quicker development time GP B) Better performance characteristics of GP The Reality 2. The Objective

6 D OSHISHA U NIVERSITY 13 November 20156 3. The Approach Quicker development time of GP ? History of “Reuse of Software Blocks” in Software Engineering: loops, procedures, functions (incl. recursions), modules (units), objects, component objects History of “Reuse of Software Blocks” in Software Engineering: loops, procedures, functions (incl. recursions), modules (units), objects, component objects Component objects (CO): appears to be an object of the IDE which incorporates them, binary standard (language-independent) Component objects (CO): appears to be an object of the IDE which incorporates them, binary standard (language-independent)

7 D OSHISHA U NIVERSITY 13 November 20157 3. The Approach Focusing on representation of genetic programs: A) Standard DOM-parsing tree and XML text. B) CO: DOM-parser with built-in API for dealing with genetic programs.

8 D OSHISHA U NIVERSITY 13 November 20158 3. The Approach Advantages A) Significant reduction of the time consumption of software engineering of GP using build-in API for creating and manipulating genetic programs.

9 D OSHISHA U NIVERSITY 13 November 20159 3. The Approach Issue: How to represent the allowed syntax (i.e. to reduce the search space) of GP? In the program source of GP-system (modifications by expert, recompilation, etc…) ? As an external text with well-known format? Employing XML facilitates the second choice. Issue: How to represent the allowed syntax (i.e. to reduce the search space) of GP? In the program source of GP-system (modifications by expert, recompilation, etc…) ? As an external text with well-known format? Employing XML facilitates the second choice.

10 D OSHISHA U NIVERSITY 13 November 201510 3. The Approach B) Increase of efficiency of execution of XGP: Reducing the computational effort as a result of generic support for the idea of pruning the solution space via strongly typed GP. How: XML-schema as a standard, generic way to represent the syntax of XGP. B) Increase of efficiency of execution of XGP: Reducing the computational effort as a result of generic support for the idea of pruning the solution space via strongly typed GP. How: XML-schema as a standard, generic way to represent the syntax of XGP. Advantages

11 D OSHISHA U NIVERSITY 13 November 201511 3. The Approach Relationship between tree nodes in XGP, Data types associated with tree nodes Relationship between tree nodes in XGP, Data types associated with tree nodes Fragment of XML Schema

12 D OSHISHA U NIVERSITY 13 November 201512 3. The Approach B) Increase of efficiency of execution of XGP - parallelism: Improving the computational performance: XML representation of both the schema and the genetic programs is a feasible format for migration of agents in parallel, distributed computer architectures. B) Increase of efficiency of execution of XGP - parallelism: Improving the computational performance: XML representation of both the schema and the genetic programs is a feasible format for migration of agents in parallel, distributed computer architectures. Advantages In-memory tree structures of GP cannot be transferred between computing units in parallel architectures.

13 D OSHISHA U NIVERSITY 13 November 201513 3. The Approach Memory Structure (DOM) Text (XML) Straightforward Mapping

14 D OSHISHA U NIVERSITY 13 November 201514 3. The Approach GP Manager (selection, crossover, and mutation) Domain Independent (only XML Schema need to be updated) Simulation Boards (evaluation) Domain-specific Structure of XGP-framework Implications: Reuse of GP Manager across the applications, Parallel Simulation Boards

15 D OSHISHA U NIVERSITY 13 November 201515 Parallel Implementation via Boss-Workers Model Example – Evolution of Behavior of Agents in MAS GP Manager (selection, crossover, and mutation) GP Manager (selection, crossover, and mutation) Simulation Boards (evaluation) Simulation Boards (evaluation) Genetic program (XML) Fitness 3. The Approach

16 D OSHISHA U NIVERSITY 13 November 201516 4. Verification Results Development time for the initial prototype of XGP (from scratch): several [person*days]

17 D OSHISHA U NIVERSITY 13 November 201517 4. Verification Results Porting time (employing XGP for already developed simulation board): less than one hour XML Schema File

18 D OSHISHA U NIVERSITY 13 November 201518 4. Verification Results Computational Effort of XGP: Reducing the search Space (XML Schema) Probability of Success for Evolution of XGP with (STGP) and without (LP, LPA) strong types

19 D OSHISHA U NIVERSITY 13 November 201519 GP Manager Domain Neutral MAS Simulation Board Domain Specific 5. Applications Evolution of Agents Behavior in MAS

20 D OSHISHA U NIVERSITY 13 November 201520 XML representation of GP 5. Applications Evolution of Agents Behavior in MAS

21 D OSHISHA U NIVERSITY 13 November 201521 GP Manager Domain Neutral Simulation Board Domain Specific 5. Applications DOM representation of GP Evolution of Locomotion of Snakebot

22 D OSHISHA U NIVERSITY 13 November 201522 GP Manager Domain Neutral Simulation Board Domain Specific 5. Applications XML representation of GP Evolution of Neural Networks

23 D OSHISHA U NIVERSITY 13 November 201523 Car (1/24 Scale Model) Remote Control (agent’s actions) Camera (perceptions of the agent) PC (driving agent) Control Loop, 100ms 5. Applications Evolution of Driving Agent

24 D OSHISHA U NIVERSITY 13 November 201524 GP Manager Domain Neutral Simulation Board Domain Specific 5. Applications DOM representation of GP Evolution of Driving Agent

25 D OSHISHA U NIVERSITY 13 November 201525 GP Manager Domain Neutral Simulation Board Domain Specific 5. Applications DOM representation of GP Interactive Evolution of Postures of Aibo Robot

26 D OSHISHA U NIVERSITY 13 November 201526 GP Manager Domain Neutral Simulation Board Domain Specific 5. Applications DOM representation of GP Interactive Evolution of Room Colors

27 D OSHISHA U NIVERSITY 13 November 201527 GP Manager Domain Neutral Simulation Board Domain Specific 5. Applications Evolution of Human-Relation Networks

28 D OSHISHA U NIVERSITY 13 November 201528 6. Conclusion A)Reduced Development Time Managing genetic program via standard DOM parsers with built-in API Proposed DOM/XML-Based Portable Genetic Representation in XGP B) Easy Porting to New Applications Reusing the very General, Domain-Independent GP Manager, Modifying the XML-schema only.

29 D OSHISHA U NIVERSITY 13 November 201529 6. Conclusion Proposed DOM/XML-Based Portable Genetic Representation in XGP C) Improved Execution Time of XGP Reducing Computational Effort: Limiting solution space using strongly typed GP and offering generic support via XML schema, Improving Computational Performance: Generic support of distributed (web-compliant) implementation of GP. Drawbacks? Fitness evaluation – parsing of XML/DOM tree and navigating among the nodes…


Download ppt "D OSHISHA U NIVERSITY 13 November 20151 XML-based Genetic Programming Framework: Design Philosophy, Implementation and Applications."

Similar presentations


Ads by Google