S. Kambhampati. On the relations between intelligent backtracking and explanation-based learning in planning and constraint satisfactions. Artijcial Intelligence, 105, 1998.  Home/Search   Document Details and Download   Summary   Related Articles   Check

....and to improve the algorithm so that it uses the extra information given in an OCL model. For example, domain invariants typically found in an OCL model often read as mutex constraints on a pair of substates. Finally, improvements to the basic algorithm such as dependency directed backtracking [3] have not been implemented but there is no reason to expect that they would not be equally applicable to our version of the algorithm. ....

S. Kambhampati. On the relations between intelligent backtracking and explanation-based learning in planning and constraint satisfactions. Artijcial Intelligence, 105, 1998.

....and to improve the algorithm so that it uses the extra information given in an OCL model. For example, domain invariants typically found in an OCL model often read as mutex constraints on a pair of substates. Finally, improvements to the basic algorithm such as dependency directed backtracking [3] have not been implemented but there is no reason to expect that they would not be equally applicable to our version of the algorithm. ....

S. Kambhampati. On the relations between intelligent backtracking and explanation-based learning in planning and constraint satisfactions. Artificial Intelligence, 105, 1998.

....the algorithm so that it uses yet more of the extra information given in an OCL model. For example, domain invariants typically found in an OCL model often read as mutex constraints on a pair of substates. Finally, improvements to the basic algorithm such as dependency directed backtracking [4] have not been implemented but there is no reason to expect that they would not be equally applicable to our version of the algorithm. ....

S. Kambhampati, `On the relations between intelligent backtracking and explanation-based learning in planning and constraint satisfactions ', Artificial Intelligence, 105, (1998).

....Complete details of our experiments can be found in [7] 7. 1 EBL and nogood learning The most important extension to the solver is the incorporation of EBL, which helps the solver to explain the failures it has encountered during search, and use those explanations to avoid similar failures later [15]. The nogoods are stored as partial variable value assignments, with the semantics that any assignment that subsumes a nogood cannot be refined into a solution. Extending GAC CBJ to support EBL is reasonably straightforward as the conflict directed backtracking already provides most of the ....

S. Kambhampati. On the relation between intelligent backtracking and failure-driven explanation-based learning in constraint satisfaction and planning. Artificial Intelligence, page Spring, 1999. 25

....and to improve the algorithm so that it uses the extra information given in an OCL model. For example, domain invariants typically found in an OCL model often read as mutex constraints on a pair of substates. Finally, improvements to the basic algorithm such as dependency directed backtracking (Kambhampati 1998) have not been implemented but there is no reason to expect that they would not be equally applicable to our version of the algorithm. ....

Kambhampati, S. 1998. On the relations between intelligent backtracking and explanation-based learning in planning and constraint satisfactions. Artificial Intelligence 105.

....(at v s) at v d) Figure 7: Parameterized specification of the action of driving a vehicle from a source location to a destination. based on Kambhampati s earlier work on the relationship between traditional planning based speedup methods (e.g. explanation based learning) and CSP methods [50]. 2.3.2 Closed World Assumption The closed world assumption says that any proposition not explicitly known to be true in the initial state can be presumed false. A simple way of implementing the closed world assumption in Graphplan would be to explicitly close the zeroth level of the planning ....

S. Kambhampati. On the relations between intelligent backtracking and failure-driven explanation based learning in constraint satisfaction and planning. Department of Computer Science and Engineering TR-97-018, Arizona State University, 1998. To appear in Artificial Intelligence.

....at the point of failure, within a layer, are actually memoized. More goal sets are likely to contain the smaller memoized subset than would be likely to contain the complete original failing goal set. This therefore allows us to prune search branches earlier. This method is a weak version of Kambhampati s (1998, 1999) EBL (Explanation Based Learning) modifications. EBL allows the identification of the subset of a goal set that is really responsible for its failure to yield a plan. Memoization of smaller sets increases the efficiency of the planner by reducing the overhead necessary in identifying failing goal ....

Kambhampati, S. (1999). On the Relations Between Intelligent Backtracking and Explanation Based Learning in Planning and CSP. Artificial Intelligence, 105 (1-2).

....Complete details of our experiments can be found in [8] 7. 1 EBL and nogood learning The most important extension to the solver is the incorporation of EBL, which helps the solver explain the failures it has encountered during search, and use those explanations to avoid those failures later [15] . The nogoods are stored as partial variable value assignments, with the semantics that any assignment that subsumes a nogood cannot be refined into a solution. Extending GAC CBJ to support EBL is reasonably straightfoward as the conflict directed backtracking already provides most of the ....

S. Kambhampati. On the relation between intelligent backtracking and failure-driven explanation-based learning in constraint satisfaction and planning In Artificial Intelligence, Spring 1999.

....fails, the reason of the failure has to be created and communicated to the planner. The three steps are detailed next. Explanation Generation Generating failure explanation for the scheduler can be done in a straightforward fashion by using the explanation based backtracking techniques [19,18]. Specifically, if we employ a conflict directed backjumping strategy [44,19] to guide the solution of the scheduling CSP, in the event the CSP cannot be solved, the conflict set at the root of the search tree shows the subset of variables of the scheduling CSP that are causing the failure. ....

....to the planner. The three steps are detailed next. Explanation Generation Generating failure explanation for the scheduler can be done in a straightforward fashion by using the explanation based backtracking techniques [19,18] Specifically, if we employ a conflict directed backjumping strategy [44,19] to guide the solution of the scheduling CSP, in the event the CSP cannot be solved, the conflict set at the root of the search tree shows the subset of variables of the scheduling CSP that are causing the failure. Explanation Translation After we get the failure explanations from the scheduler ....

Kambhampati, S. On the Relations between Intelligent Backtracking and Failure-Driven Explanation-based Learning in Constraint Satisfaction and Planning. Artificial Intelligence. 1999.

Online articles have much greater impact   More about CiteSeer.IST   Add search form to your site   Submit documents   Feedback  

CiteSeer.IST - Copyright Penn State and NEC