Prosser, P.: 1993a, `Domain Filtering Can Degrade Intelligent Backtrack Search'. In: Proceedings IJCAI'93. Chambry, France, pp. 262--267.  Home/Search   Document Not in Database   Summary   Related Articles   Check

....result in an additional search e#ort to find a solution. From a practical perspective, unrestricted nogood recording is not necessarily a bad approach. It is interesting to observe that the uselessness of CBJ related algorithms w.r.t. CB related algorithms has been experienced in the past [14] [3] CBJ, when applied jointly with a domain filtering procedure (e.g. MAC) and an accurate variable ordering heuristic, has been considered an expensive approach that almost always slows down the search, even if it saves a few constraint checks . Table 3 gives the results for the number of ....

P. Prosser. Domain filtering can degrade intelligent backjumping search. In Proceedings of the International Joint Conference on Artificial Intelligence, pages 262-- 267, 1993.

....we cannot assign v to the variable x. To keep track of all nogoods that eliminate a value in this fashion, let us use N x to represent the set of nogoods that are of the form p hx; vi. This process of collecting nogoods is in fact much like that used by sophisticated versions of backjumping [10, 31]. The difference between dynamic backtracking and backjumping lies in how the backtracking is done. In each case, the algorithm finds that for some variable, say x, there are no values that have not been eliminated. This means that in order to instantiate x, we must undo some variable assignment ....

Patrick Prosser. Domain filtering can degrade intelligent backtracking search. In Proceedings of the Thirteenth International Joint Conference on Artificial Intelligence, pages 262--267, 1993.

....constraint processing; the number of search steps needed to find a solution was also increased. Several researchers have demonstrated unexpected effects when combining two or more techniques: backtracking has been shown to interact badly with the normally beneficial techniques of domain pruning [6], arc consistency preprocessing [7] and removal of inconsistent or redundant domain values or subproblems [2] The effects of combining techniques can be surprising and unpredictable, so it should not be a complete surprise that local search and symmetry breaking do not always combine well. ....

P. Prosser. Domain Filtering Can Degrade Intelligent Backjumping Search. Proceedings of the Thirteenth International Joint Conference on Artificial Intelligence, Chambery, France, Morgan Kaufmann 1993, pp. 262--267.

....such that k,C # D # c . The concept of conflict is crucial for decision repair. It is used to focus the search on judicious (and dynamic) neighborhoods. Conflicts have been widely used in AI in different contexts and under different names, such as nogoods as in [45] or conflict sets as in [41]. See Appendix A for a presentation of a conventional way to compute conflicts. 3.3. tabu decision repair Decision repair is a generic algorithm for which instances are obtained by specializing several parameters: the nature and behavior of # the storage structure; the neighboring ....

P. Prosser, Domain filtering can degrade intelligent backtracking, in: Proc. IJCAI-93, Chambry, France, 1993, pp. 262--267.

....study of constraint satisfaction problems, and several important lessons have emerged. These lessons concern the relationships between the number of constraints in a problem, the probability that the problem has a solution, and the typical difficulty of such a problem for a search algorithm [3, 5, 7, 15, 16, 18, 19, 22]. Suppose we randomly generate constraint satisfaction problems of some particular type (e.g. graph coloring) The probability of there being a solution will depend on the distribution used to generate these problems. If the problems only have a few constraints, then they are very likely to have ....

P. Prosser. Domain filtering can degrade intelligent backtracking search. In Proceedings of the Thirteenth International Joint Conference on Artificial Intelligence, pages 262--267, 1993.

....assist to FC. One final point is that EFC is identical to the algorithm FC BM described by Prosser in [9] In particular, Prosser identified that when backing up from a DWO the gains of Prune: Backup could be achieved. However, the algorithm he described uti 6 Prosser s FC D2C algorithm [4] uses a special test to recognize the case when a value can be pruned back to level 0. This special case is achieved automatically in EFC EFC . 7 It is not hard to define EFC for n ary CSPs, and it should be feasible to extend all of the new n ary versions of FC defined in [5] using these ideas. ....

....7 of the routine with line v.CF : flevelg. CFFC is generally faster that CFFC as computing the conflict sets for all values pruned by forward checking can take more time than it saves. We have also found that on some problems the extra pruning performed by CFFC can seriously degrade backjumping [4]. 13 Since 0 might be the maximum of a conflict set, this process automatically achieves the special case permanent pruning of Prosser s CBJ DkC algorithm [4] Fahiem Bacchus Finally we can define our fifth algorithm CFMAC. In this routine instead of doing forward checking we enforce ....

[Article contains additional citation context not shown here]

Patrick Prosser. Domain filtering can degrade intelligent backtracking search. In Procceedings of the International Joint Conference on Artifical Intelligence (IJCAI), pages 262--267, 1993.

....of value to each variable such that no constraints are violated [20] In this paper, the discussion is confined to binary CSPs only. Forward checking (FC) is considered as one of the most efficient search methods to solve CSPs. Enhancements have been proposed in order to make it more efficient [15, 16]. One recent result of this effort is minimal forward checking (MFC) 4] and a similar idea has been proposed in [22] MFC is interesting since given the same variable and value orderings, it always performs fewer or the same number of compatibility checks than that of FC. In this paper, we ....

....small n and p1. Out of the 32,400 CSPs that we tested, there are only 43 cases that MFC outperformed MFC BM2 CBJ with a very small amount. We identify two possible reasons accounting for this. The first is the harmful interaction between domain filtering and intelligent backtracking as reported in [16]. The second is the harmful interaction between domain filtering and the delayed evaluation in MFC BM2 CBJ. Although the latter phenomenon has not been reported in the literature according to our knowledge, we have to omit it in this paper due to space limitation. A discussion of the topic can be ....

Prosser, P. Domain filtering can degrade intelligent backtracking search. In Proceedings International Joint Conference on Artificial Intelligence, 262-267, 1993.

....our test problems. Another counterintuitive demonstration that pruning values can increase search effort, was obtained recently by Prosser. He showed that pruning values can degrade performance for algorithms that employ intelligent backtracking (though the actual exhibited effects were small) [4]. However, even Prosser concluded that: We should now assume that increased consistency, or the removal of redundancies, can only guarantee a reduction in search effort if that search is unintelligent (such as a chronological backtracker) Forward checking is a chronological backtracker. ....

Prosser P.: Domain filtering can degrade intelligent backtracking search. Proceedings of the Thirteenth International Joint Conference on Artificial Intelligence (1993) 262--267.

....domain f1; 2; 3; 4; 5; 6; 7; 8g. The constraint graph for this problem is connected. The tightness of the problem is 0.06. 3 A complete listing of the specification of the EHP is given in appendix A. 38 4.3. 1 Benchmarking The forward checking algorithm with dynamic variable ordering (fc dvo) [29], which always chooses variables with the smallest current domain, cannot solve the problem after 711 million consistency checks. However, the same algorithm augmented with conflict directed back jumping method (fccbj dvo) 30] finds a solution with 9588 consistency checks. PROCLANN solves the ....

P. Prosser. Domain filtering can degrade intelligent backtracking search. In Proceedings of the International Joint Conference on Artificial Intelligence, pages 262--267, 1993.

....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 ....

Prosser, P. Domain Filtering can Degrade Intelligent Backjumping Search. Proc. IJCAI-93, 262-267. 1993.

....dependency directed backtracking and EBL have been dormant in recent years, there are several reasons to expect a resurgence of interest these topics. Much of the early work in CSP has been on systematic search algorithms, within which EDB and EBL play a role. A variety of empirical studies (c.f. (Prosser, 1993; Frost Dechter, 1994c, 1994a) have consistently shown that EDB and EBL techniques are often part of the winning constraint satisfaction search algorithms. Although the emphasis shifted to non systematic search strategies such as GSAT (Selman, Levesque, Mitchel, 1992) in the recent past, ....

....nodes and provide them an explanation (if not a minimal explanation) of failure in CSP, even this can be computationally expensive in 14. Similar points are raised by Minton and Etzioni (Etzioni Minton, 1992) This phenomenon is also similar to the bridging effect that Prosser talks about (Prosser, 1993) 15. Dechter (Dechter, 1990) uses multiple failure explanations for leaf node failures but she never propagates them upwards or simplifies them (since she is getting the minimal explanations to begin with) So, the implicit disjunction involved in multiple failure explanation is not a problem ....

[Article contains additional citation context not shown here]

Prosser, P. (1993). Domain filtering can degrade intelligent backtracking search. In Proc. IJCAI-93.

....characterised then it should be possible to go on and claim that a particular combination is the best choice, for example, when tackling hard problems . There has also been some recent work on the pitfalls of particular algorithms and on adverse interactions between some algorithms and heuristics [9], 10] 11] This work has given more evidence to the notion that there is much that we do not yet fully understand the interactions between different algorithms and heuristics, where they perform best, and which one is the best, over the infinite range of possible CSPs. These new findings, ....

....that there is much that we do not yet fully understand the interactions between different algorithms and heuristics, where they perform best, and which one is the best, over the infinite range of possible CSPs. These new findings, however, should be treated with caution. For example, Prosser [9] has identified potential problems with the combination of intelligent back trackers and consistency techniques in the form of directional k consistency (DKC) whereby the phenomena of the Long Jump and the Bridge manifest themselves. The problem with this finding is how significant is it over ....

[Article contains additional citation context not shown here]

Prosser, P., Domain Filtering can Degrade Intelligent Backjumping Search, Proceedings 13th International Joint Conference on Artificial Intelligence, 1993, 263-267

....exceptionally hard problems (EHP s) 31] which consists of 50 variables, each with domain f1; 2; 3; 4; 5; 6; 7; 8g. The constraint graph for this problem is connected. The tightness of the problem is 0.06. 4.3. 1 Benchmarking The forward checking algorithm with dynamic variable ordering (fc dvo) [29], which always chooses variables with the smallest current domain, cannot solve the problem after 711 million consistency checks. However, the same algorithm augmented with conflict directed back jumping method (fccbj dvo) 30] finds a solution with 9588 consistency checks. PROCLANN solves the ....

P. Prosser. Domain filtering can degrade intelligent backtracking search. In Proceedings of the International Joint Conference on Artificial Intelligence, pages 262--267, 1993.

....then it should be possible to go on and claim that a particular combination is the best choice, for example, when tackling hard problems . There has also been some recent work on the pitfalls of particular algorithms and on adverse interactions between some algorithms and heuristics [Prosser 93b] Baker 94] Sabin CSM 210 page 3 Freuder 94] This work has given more evidence to the notion that there is much that we do not yet fully understand the interactions between different algorithms and heuristics, where they perform best, and which one is the best, over the infinite range ....

....however, should be treated with caution. For example, Prosser has identified potential problems with the combination of intelligent back trackers and consistency techniques in the form of directional k consistency (DKC) whereby the phenomena of the Long Jump and the Bridge manifest themselves [Prosser 93b] The problem with this finding is how significant is it over the range of CSPs, and how can someone make an informed decision on the use of an intelligent backtracker with consistency techniques without full knowledge of the prevalence and significance of it. In this paper we describe a series ....

[Article contains additional citation context not shown here]

Prosser, P., 1993, Domain Filtering can Degrade Intelligent Backjumping Search, Proceedings 13th International Joint Conference on Artificial Intelligence, 263-267

No context found.

P. Prosser, Domain filtering can degrade intelligent backtracking search. To appear in Proc IJCAI-93

....for all. But sometimes it s not easy to forget. We resorted to a detailed analysis of the two problems, tracing every feature of the algorithm as it progressed through the search space. A visual analysis took days, and resulted in an explanation of the phenomenon and an ability to replicate it [Prosser 1993a] Don t reject the obvious Several times we have looked at some result and rejected or not considered an obvious interpretation. For example, we plotted graphs of the score for GSAT decaying with the number of moves performed, but did not consider the obvious possibility of exponential ....

P. Prosser. 1993a. Domain filtering can degrade intelligent backtracking search. In Proceedings of IJCAI-93, 262--267.

....uninstantiated variables are made consistent, to some degree, with respect to that instantiation. Chronological backtracking (BT) backmarking (BM) backjumping (BJ) conflict directed backjumping (CBJ) and dynamic backtracking (DB) might all be considered as algorithms that check backwards [11, 5, 6, 13, 10], whereas forward checking (FC) and maintaining arc consistency (MAC) might be considered as algorithms that check forwards [12, 17] This study investigates only forward checking algorithms, and in particular FC CBJ. Algorithm FC instantiates variable v i with a value x i and removes from the ....

P. Prosser. Domain filtering can degrade intelligent backtracking search. In Proc. IJCAI-93, pages 262--267, 1993.

No context found.

Prosser, P.: 1993a, `Domain Filtering Can Degrade Intelligent Backtrack Search'. In: Proceedings IJCAI'93. Chambry, France, pp. 262--267.

No context found.

P. Prosser. Domain filtering can degrade intelligent backjumping search. In Proceedings of the International Joint Conference on Artificial Intelligence, pages 262-- 267, August 1993.

No context found.

P. Prosser. Domain filtering can degrade intelligent backtracking search. In Proc. IJCAI-93, 1993.

No context found.

Prosser, P. (1993a). Domain filtering can degrade intelligent backtracking search. In Proceedings of the Thirteenth International Joint Conference on Artificial Intelligence, pp. 262--267, Chamb`ery, France.

No context found.

Prosser, P. 1993a. Domain filtering can degrade intelligent backtracking search. Dans Proc. of the 13 th IJCAI, 262--267. 8, 9 26

No context found.

Prosser, P. (1993a) Domain filtering can degrade intelligent backtracking search. Proc. International Joint Conference on Artificial Intellience pp. 262--267.

No context found.

Prosser, P. (1993) Domain filtering can degrade intelligent backtracking search. Proc. of IJCAI'93, Chamb éry (France), pp. 262-267.

No context found.

Prosser, P. 1993b. Domain filtering can degrade intelligent backtracking search. In Proceedings of the Thirteenth International Joint Conference on Artificial Intelligence, 262--267.

First 50 documents

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