Citations: Defining operationality for explanation-based learning

65 citations found. Retrieving documents...

Kel88 Keller, R. M., Defining Operationality for Explanation-Based Learning, Artificial Intelligence 35, (1988), pp 227-241

Home/Search Document Not in Database Summary Related Articles Check

This paper is cited in the following contexts:

First 50 documents Next 50

Probabilistic Hill-Climbing - Cohen, Greiner, Dale (1991) (6 citations) (Correct)

....element will actually encounter. This average case analysis differs from several other approaches as, for example, we do not assume that this distribution of problems will be uniform [Gol79] nor that it will necessarily correspond to any particular collection of benchmark challenge problems [Kel87] 4 Conclusion Applications: Several related papers present specific applications of this palo system. GJ92] illustrates how this approach fits into the framework of explanation based learning systems, and in particular, that how this analysis extends and formalizes [Min88] s utility ....

Richard M. Keller. Defining operationality for explanation-based learning. In Proceedings of AAAI-87, pages 482--87, Seattle, July 1987.

Explaining Scenarios for Information Personalization - Ramakrishnan, Rosson, Carroll (2001) (Correct)

....an instance of politeness. Such an over generalization is however less operational, since it assumes that Linus has some other way of deciding what makes a phrase well mannered. Operationality is thus related to the utility of the induced generalization. 3. 2 Using EBG in Personalization Keller [33] shows how we can think of EBG as a search through a concept description space such as Fig. 5. The operationality consideration is then the objective function used to evaluate entries in the concept description space. The most specific construct simply records the conversation and can only be ....

....complexity on a population of [scenarios that are likely to be encountered] Being too specific when operationalizing explanations will lead to making more distinctions than losing them, contributing to lesser orthogonality (salience, as used in [46] among scenarios. Defining operationality [33] carefully in the personalization context is an area for future research. 5 Discussion This research makes contributions to the state of the art in both personalization systems and scenario based design. For personalization, we have clarified the aspects of requirements specification and ....

[Article contains additional citation context not shown here]

R.M. Keller. Defining Operationality for Explanation-Based Learning. Artificial Intelligence, Vol. 35:pages 227--241, 1988.

Learning as Knowledge Integration - Murray (1995) (2 citations) (Correct)

.... Traditional approaches to explanation based learning require a criterial task to determine when some path through an inference graph should be compiled into a shallow rule: compile the rule when doing so improves performance (e.g. in terms of response time) at that criterial task [Min88, Kel88] Since KI cannot assume a criterial task, it must exploit other criteria to determine when an inference path should be compiled into a shallow rule. The rule macros of the representation language (Section 2.2.2) provide such criteria: compile an inference path into a shallow rule when that rule ....

....system) these learning goals are not defined by specific and relatively narrow expectations about the application goals. The goal concepts that are being learned are not defined by a criterial task as they are with traditional explanation based learning systems (e.g. MKKC86, DM86, Min88, Kel88] 7.4.2 Theory refinement and inductive logic programming FOIL: FOIL improves on traditional approaches to concept acquisition and theory refinement by operating over a clausal representation language that is much more expressive than the attribute vector representation languages as 215 sumed ....

R.M. Keller. Defining operationality for explanation-based learning. Artificial Intelligence, 35:227--241, 1988. 321

Learning Control Knowledge in Models of Expertise - Straatman (1995) (Correct)

....computation. Of these methods, only the third is applicable to knowledge level learning of control knowledge, because it uses a separate and explicit representation of the control knowledge. A number of systems for learning meta level control constructs have been built. Among these are MetaLEX [Keller, 1988], STATIC [Etzioni, 1992] Prodigy EBL [Minton, 1988, Minton, 1990] and LEX [Mitchell et al. 1983] These systems are built on top of meta level problem solvers. The object level of the problem solver consists of domain knowledge and operators, that change the state of the problem solver. The ....

R. M. Keller. Defining operationality for explanation-based learning. Artificial Intelligence, 35:227--241, 1988.

Automated Design Of Knowledge-Lean Heuristics: Learning.. - Ieumwananonthachai (1996) (Correct)

....a well designed automated system for learning heuristics that can solve application problems with a high level of performance and within a reasonable amount of time. In the past, there have been many efforts on designing heuristics using machine learning techniques [1 9] however, most of them [1,2,4 9] assume a world model that relates heuristics and their performance. This type of heuristics is known as knowledge rich heuristics. In this research, we focus on learning for a part of an application problem with knowledge lean heuristics where no such model exists. We address several ....

....of domain knowledge available. For a knowledge rich problem solver with a good world model to relate decisions made by each HDE and performance feedbacks, a better HM can be generated based on the model through credit assignments. This approach is the focus of many studies in machine learning [1, 2, 4 9]. The generation process is much more difficult for knowledge lean problem solvers where such a world model does not exist and credit assignments cannot be used. Weaker, modelfree, domain independent, and syntactic operators for generating new HMs can still be used in this case. Examples of these ....

R. Keller, "Defining operationality for explanation-based learning," in Proc. National Conf. on Artificial Intelligence, Seattle, Washington, AAAI, Inc., June 1987, pp. 482--7.

Learning Function of Devices Using Qualitative Function Formation.. - Far (1992) (Correct)

....whose enabling conditions of their arcs are not yet satisfied are deleted and a conventional simulation program derives landmark values for eachvariable of the remaining processes. The behavior of the system is the record of BFs. A function concept can be expressed in terms of: Operationality [19], i.e. activated processes and their enabling conditions, and repetition cycle denoting a persistence or an order in the trace of the BFs. The repetition cycle or persistence is derived for each of the variables. Note that different cycles can possibly be detected and each cycle may represent a ....

....using behavioral fragments derived by qualitative simulation on the extended qualitative model. ffl Providing solution to some of the functional reasoning problems. Extending the method to account for the associative learning of functions, operationalization and clustering of function concepts [19], learning function by analogy and induction are other problems to be tackled. Application of QFF in functional design [10] and fault diagnosis [11] is currently under investigation [10] ACKNOWLEDGEMENTS This research was conducted at the Computing and Informations System Center, Japan Atomic ....

R.M. Keller, "Defining Operationality for ExplanationBased Learning," in Proc. 6th National Conf. on Artif. Intell. (AAAI'87), Seattle, WA, July 1987, pp. 482-487.

On Rationality and Learning - Doyle (1988) (1 citation) (Correct)

....an important procedure, but to perform well its inputs must be selected rationally and its outputs must be evaluated rationally. In fact, DeJong s [1983] original criteria for explanation based learning concerned utility of the result much more than operationality of the definition s elements. Keller [1987] also recognizes the limitations of formal operational criteria and proposes to redefine operationality to be usability plus utility (by which he seems to mean expected utility) with both usability and utility continuously variable in degree and dynamically changing. This is a step in the right ....

Keller, R. M., 1987. Defining operationality for explanation-based learning, Proc. Sixth Natl. Conf. on Artificial Intelligence, 482-487.

PALO: A Probabilistic Hill-Climbing Algorithm - Greiner (1995) (19 citations) (Correct)

....element will actually address. This average case analysis differs from several other approaches as, for example, we are not assuming that this distribution of problems will be uniform [25] nor that it will necessarily correspond to any particular collection of benchmark challenge problems [52]. N Palo2. A 0 local optimum corresponds exactly to the standard notion of local optimum; hence our ffl local optimum (condition 2 of Theorem 1) generalizes local optimality. This means that palo 1 s output Theta m will (with high probability) be a real local optimum if the difference in ....

R. M. Keller. Defining operationality for explanation-based learning. In Proceedings of AAAI-87, pages 482--87, Seattle, July 1987.

Probabilistic Hill-Climbing: Theory and Applications - Greiner (1992) (4 citations) (Correct)

....element will actually address. This average case analysis differs from several other approaches as, for example, we do not assume that this distribution of problems will be uniform [Gol79] nor that it will necessarily correspond to any particular collection of benchmark challenge problems [Kel87] N PALO2. All three c ff (PE; q) functions discussed in this paper are bounded ; i.e. satisfy 8 PE 2 PE ; q 2 Q: c c ff (PE; q) c 4 See [Bol85, p. 12] N.b. these inequalities holds for essentially arbitrary distributions, not just normal distributions, subject only to the minor ....

R. Keller. Defining operationality for explanation-based learning. In Proceedings of AAAI-87, 1987.

Measuring and Improving the Effectiveness of Representations - Greiner, Elkan (1991) (1 citation) (Correct)

.... it therefore is more likely to provide a good indication of R s true utility than we would get by testing R on worst case queries, a set of concocted queries, a sample drawn randomly from a uniform distribution [ Goldberg, 1979 ] or any particular collection of benchmark challenge problems [ Keller, 1987 ] Comparing representations: If we had an analytic technique for evaluating the utility of representations with respect to a distribution of queries, and knew this distribution of queries, we could directly determine which of two representations was better. In general we have neither analytic ....

R. Keller. Defining operationality for explanation-based learning. In AAAI-87, 1987.

Probably Approximately Optimal Derivation Strategies - Greiner, Orponen (1991) (1 citation) (Correct)

....distribution. Notice this distribution will depend on the particular task our DP is addressing; n.b. we do not assume that it is a uniform distribution over all possible problems [Gol79] nor that it will necessarily correspond to any particular collection of benchmark challenge problems [Kel87] Hence, we are using the same weak distribution free assumption that underlies the current work in PAC learning [Val84] We can translate this query distribution into a set of probability values that specify the chance that each retrieval will succeed. We assume, also, that these success ....

Richard M. Keller. Defining operationality for explanation-based learning. In AAAI87, pages 482--87, Seattle, July 1987.

A Statistical Approach to Solving the EBL Utility Problem - Greiner (1992) (31 citations) (Correct)

....q i is selected. This Pr[ Delta ] reflects the distribution of problems our PE is actually addressing; n.b. it is not likely to be a uniform distribution over all possible problems [Gol79] nor will it necessarily correspond to any particular collection of benchmark challenge problems [Kel87] We can then define the expected cost of a performance element: C[ PE ] def = E[ c(PE; q) X q2Q Pr[ q ] Theta c(PE; q) Our underlying challenge is to find the performance element whose expected cost is minimal. There are, however, two problems with this approach: First, we know to know ....

R. Keller. Defining operationality for explanation-based learning. In AAAI-87, 1987.

Using Goals and Experience to Guide Abduction - Leake (Correct)

....systems take a goal neutral view. Artificial Intelligence investigation of goalbased explanation evaluation has been concentrated in explanation based learning (EBL) research (Mitchell, Keller, Kedar Cabelli, 1986; DeJong Mooney, 1986) and focuses on the task of concept recognition (see (Keller, 1988) for an overview of some of this work) Rich models of usefulness have been developed for the recognition task (e.g. Keller, 1987; Minton, 1988) and other research has considered how the recognition task is motivated by other goals (Kedar Cabelli, 1987) but other tasks have received little ....

Keller, R. (1988). Defining operationality for explanation-based learning. Artificial Intelligence, 35 (2), 227--241.

Explanation-Based Learning of Indirect Speech Act.. - Schulenburg, Pazzani (1989) (Correct)

....Next, the example is generalized by retaining only those features of the example which were necessary to produce the explanation. This generalization characterizes the class of problems that will have the same solution for the same reason as the training example. EBL explicates (or operationalizes (Keller, 1987)) information that is implicitly represented in a system. For example, aces (Pazzani, 1987) is a system that learns diagnosis heuristics (i.e. efficient heuristics that associate faults with symptoms) from a functional device description. In this work, we are using a modified version of the eggs ....

Keller, R. (1987). Defining operationality for explanation-based learning. Proceedings of the National Conference on Artificial Intelligence (482-487). Seattle, WA: Morgan-Kaufmann.

Explanation-Based Generalization for Negation as Failure and.. - Schr�dl (1996) (Correct)