Presentation is loading. Please wait.

Presentation is loading. Please wait.

Ontologies Reasoning Components Agents Simulations Project Topics: MCI 2007.1 Jacques Robin.

Similar presentations


Presentation on theme: "Ontologies Reasoning Components Agents Simulations Project Topics: MCI 2007.1 Jacques Robin."— Presentation transcript:

1 Ontologies Reasoning Components Agents Simulations Project Topics: MCI 2007.1 Jacques Robin

2 Topics Supervised by Prof. Jacques 1.Developing an ontology and component framework of search algorithms  Top-level ontology classes derived from Russell & Norvig and Dechter  Leaves of the ontology:  to work with a Constraint Handling Rules (CHR)  to include variations of conflict-directed backjumping for complete global search  to include variations of min-conflict for incomplete, local, scalable search  Using UML knowledge representation and transformation rules from UML Components to Java OSGi Components 2.Incrementally developing multi-agent simulation  Starting from simplest simulation of penalty shot  Using UML as knowledge layer representation language  Using Java and CHR as implementation layer representation language

3 Topics Supervised by Prof. Fred  ??

4 > StateSpaceSearch +gSearch(StateSpaceSearchPb):SearchSolution > StateSpaceSearch +gSearch(StateSpaceSearchPb):SearchSolution StateSpaceSearchPb +fullStateFornulation; Boolean +suc(State,AgentAction):State AgentAction +name:String +cost:Real State +full:Boolean +goal:Boolean +initial:Boolean Node +/expanded: Boolean +/root: Boolean +/visited: Integer > ExpandStrategy +choose(Fringe):Node > BtStrategy +bt(Node):Node > PruningHeuristic +prune(Node):Node[*] models 1..* > Cost2GoalHeuristic +estimCost2Goal(Node):Real > ExpandStrategy +choose(Fringe):Node > BtStrategy +bt(Node):Node > PruningHeuristic +prune(Node):Node[*] > Cost2GoalHeuristic +estimCost2Goal(Node):Real Path +/cost:Real SearchSolution NodeSolution PathSolution 2..* * > parent child * Fringe * {ordered} Framework Top-Level

5 Prof. Jacques’ Search Topic SearchAlgo GlobalSearchAlgoLocalSearchAlgo CSPSearchProblem SearchProblem FullStateFormulation SearchProblem PartialStateFormulation SearchProblem FDCSP SearchProblem VariableChoice Heuristic ValueChoice Heuristic Backtracking Heuristic PartialStateFormulation FDCSPSearchProblem FullStateFormulation FDCSPSearchProblem CSPSearchAlgo FDCSP SearchAlgo GlobalFDCSP SearchAlgo LocalFDCSP SearchAlgo CDBJMin-Confllict

6 Tasks 1.Model UML2/OCL2 hierarchy of abstract and concrete specializations of StateSpaceSearchProblem (2 students)  including concrete classes and instances of:  8 queens, CSP backjumping slides map coloring 2.Model UML2/OCL2 hierarchy of abstract and concrete specializations of CoastToGoalHeuristic (except for CSP problem, fully problem dependent) (same 2 students than 1) 3.Model UML2/OCL2 hierarchy of abstract and concrete specializations of ExpandStrategy and PruningHeuristic (Fúlvio) 4.Model UML2/OCL2 hierarchy of abstract and concrete specializations of BtStrategy (Zé Carlos, Renan) 5.Model UML2/OCL2 hierarchy of abstract and concrete specializations of StateSpaceSearch as assembly of abstract and concrete specializations ExpandStrategy, BtStrategy, PruningHeuristic, CoastToGoalHeuristic (Carlos, Alexandre) 6.Model UML2/OCL2 or other technology search visualization GUI (Joabe, speak to Luiz Lacerda, luiz.francisco.lacerda@gmail.com, about his UML2 Profile for GUI Modeling)luiz.francisco.lacerda@gmail.com

7 Tasks 1.Implementation OSGi Java, tests JUnit hierarchy of abstract and concrete specializations of StateSpaceSearchProblem (2 students)  including concrete classes and instances of:  8 queens, CSP backjumping slides map coloring 2.Implementation OSGi Java, tests JUnit hierarchy of abstract and concrete specializations of CoastToGoalHeuristic (except for CSP problem, fully problem dependent) (same 2 students than 1) 3.Implementation OSGi Java, tests JUnit hierarchy of abstract and concrete specializations of ExpandStrategy and PruningHeuristic (Fúlvio) 4.Implementation OSGi Java, tests JUnit hierarchy of abstract and concrete specializations of BtStrategy (Zé Carlos, Renan) 5.Implementation OSGi Java, tests JUnit hierarchy of abstract and concrete specializations of StateSpaceSearch as assembly of abstract and concrete specializations ExpandStrategy, BtStrategy, PruningHeuristic, CoastToGoalHeuristic (Carlos, Alexandre) 6.Implementation OSGi Java, tests JUnit or other technology search visualization GUI (Joabe, speak to Luiz Lacerda, luiz.francisco.lacerda@gmail.com, about his UML2 Profile for GUI Modeling)luiz.francisco.lacerda@gmail.com

8 Scope Search Problems  Priority1:  FullStateStateFormulation, PartialStateFormulation  CSPFullStateStateFormulation, CSPPartialStateFormulation  N-queens as FullStateStateFormulation  N-queens as PartialStateFormulation  8-queens as FullStateStateFormulation  8-queens as PartialStateFormulation  MapColoring as FullStateStateFormulation  R1-R7 MapColoring as FullStateStateFormulation  Priority 2:  PathSolutionProblem  ShortestPathBetween2Cities  Romenia

9 Scope Expand and Pruning Strategies  General StateSpaceSearch  Priority 1:  For FullStateFormulationProblems: Depth-first, Backtracking  For PartialStateFormulationProblems: min-conflict  Priority 2: Uniform cost search, A*  Priority 3: Breadth-first, iterative deepening, RBFS  CSPSearch:  Priority 1:  Variable ordering: Degree Heuristic  Value ordering: Least Constraining Value  Pruning: forward checking  Priority 2:  Variable ordering: Minimum Remaining Value  Arc consistency

10 Scope Backtrack Strategies  Chronological backtracking  Conflict-directed backjumping

11 Time Table  09-13/07: Version 1.0 of first half of model  16-20/07: Version 2.0 of first half of model  30/07-03/08: Version 1.0 of first half implementation and 1.0 of second half of model  13/08-17/08: Version 1.1 of first half implementation and 1.0 of second half implementation and integration tests  22/08: Final report

12 Topic 2 Starting Point: Simplest Possible Multi-Agent Simulation 3 2k 1s, b Y X 123 3 2bk 1s Y X 123 3b 2k 1s Y X 123 action(s,legs,shoot(2)) action(k,legs,move(right)) 3 2k, b 1s Y X 123 action(k,hands,grab(yes)) 3 2k, b 1s Y X 123 gameOver, goal gameOver, nogoal action(s,legs,shoot(3)) action(k,legs,move(right)) actions(k,legs,move(right)) Simulation Agent Shooter Agent Keeper Agent percept action

13 Topic 2 Possible Task Division  Simulation Agent:  Reasoning  Simulation Visualization:  Agent Reasoning Explanation Visualization  Shooter Agent:  Reasoning  Simulation Visualization:  Agent Reasoning Explanation Visualization  Keeper Agent  Reasoning  Simulation Visualization:  Agent Reasoning Explanation Visualization


Download ppt "Ontologies Reasoning Components Agents Simulations Project Topics: MCI 2007.1 Jacques Robin."

Similar presentations


Ads by Google