By Steven Minton

The skill to benefit from event is a primary requirement for intelligence. some of the most uncomplicated features of human intelligence is that folks can examine from challenge fixing, so they turn into more proficient at fixing difficulties in a given area as they achieve event. This publication investigates how pcs should be programmed so they can also examine from event. particularly, the purpose is to take a truly normal, yet inefficient, challenge fixing approach and teach it on a suite of difficulties from a given area, in order that it could possibly remodel itself right into a really expert, effective challenge solver for that area. on a knowledge-intensive lately there was significant development made studying technique, explanation-based studying (EBL), that brings us toward this chance. As confirmed during this booklet, EBL can be utilized to research an issue fixing episode to be able to collect keep an eye on wisdom. keep an eye on wisdom publications the matter solver's seek by means of indicating the easiest possible choices to pursue at each one selection element. An EBL approach can produce area particular keep an eye on wisdom by means of explaining why the alternatives made in the course of an issue fixing episode have been, or weren't, appropriate.

Show description

Read Online or Download Learning Search Control Knowledge: An Explanation-Based Approach PDF

Similar international books

Mobile Information Systems II: IFIP International Working Conference on Mobile Information Systems, MOBIS 2005, Leeds, UK, December 6-7, 2005 (IFIP International Federation for Information Processing)

Cellular details structures II offers a set of study at the making plans, research, layout, building, amendment, implementation, usage, review, and administration of cellular details structures. The articles specialize in the results of this study on the planet of trade, and deal with technical matters and constraints on cellular details structures functionalities and layout.

International Assessment of Research and Development in Simulation-Based Engineering and Science

Simulation-Based Engineering and technological know-how (Sbe&S) cuts throughout disciplines, displaying large promise in parts from hurricane prediction and weather modeling to knowing the mind and the habit of diverse different advanced platforms. during this groundbreaking quantity, 9 distinctive leaders check the newest examine traits, because of fifty two web site visits in Europe and Asia and enormous quantities of hours of specialist interviews, and speak about the results in their findings for the U.S. executive.

Interactive Theorem Proving: First International Conference, ITP 2010, Edinburgh, UK, July 11-14, 2010. Proceedings

This e-book constitutes the refereed court cases of the 1st overseas convention on Interactive Theorem proving, ITP 2010, held in Edinburgh, united kingdom, in July 2010. The 33 revised complete papers provided have been conscientiously reviewed and chosen from seventy four submissions. The papers are equipped in subject matters akin to counterexample iteration, hybrid method verification, translations from one formalism to a different, and cooperation among instruments.

Extra resources for Learning Search Control Knowledge: An Explanation-Based Approach

Sample text

Thus, since matching can be expensive, the utility of learned rules will always be a question worth considering. LEARN[NGSEARCHCONTROLKNOWLEDGE 24 The basis of my argument rests on the fact that matching even a simple conjunctive set of preconditions is NP-complete [28, p. 233]. , matching arbitrary first order sentences is PSPACE-complete). To see that matching conjunctive preconditions is NP-complete, consider that we can encode an arbitrary graph using conjunctive preconditions. For example, figure 2-4 shows a small directed graph and a set of preconditions that encode that graph.

A state cycle indicates that no progress has been made, and therefore if a state cycle is detected the child node is terminated and declared a failure. As indicated above, after applying an operator the candidate goals for the child node are determined by rematching the preconditions (or the top-level goalexpression) that created the goals at the parent node. This is illustrated in the search tree (figure 3-3) for the scheduling example described earlier in this chapter. After the inference rule IS-CLAMPABLE is applied, the goal at Node8 is generated by popping the goal stack, and in effect, resetting the conditions that existed at Node6 without going through the usual decision cycle.

The macrooperator that is learned indicates that a path between room-xl and room-x6 exists if room-xl is connected to room-x2 (with room-x2 not equal to room-xl), and roomx2 is connected to room-x3 and so on. This macro-operator expresses the "hamiltonian path" condition for six-node graphs. The general hamiltonian path problem for n-node graphs is another NP-complete graph matching problem. 25 ANAL)'ZING TIlE UTILITY PROBLEM Indeed, whenever MORRIS would attempt to match this macro-operator, the matcher would, in effect explore all non-cyclic paths of length six in the graph.

Download PDF sample

Rated 4.36 of 5 – based on 33 votes