K J Hammond. Case-Based Planning: Viewing Planning as a Memory Task. Academic Press Inc., 1989.  Home/Search   Document Not in Database   Summary   Related Articles   Check

....cognitive reasons. Thus formulating learning goals, asking questions, focussing attention, and pursuing learning actions are essential components of our learning model. Since the need to learn often arises from a reasoning failure, credit or blame assignment also plays a central role in learning [Hammond, 1989; Minsky, 1985; Ram Cox, 1993; Schank, 1982; Stroulia, Shankar, Goel, Penberthy, 1992; Sussman, 1975; Weintraub, 1991] Experiential learning: Learning is an incremental process of theory formation, theory revision, and conceptual change, which occurs as the reasoner accumulates experience in ....

....to know to improve its ability to process similar situations in the future. This is essentially a case based or experience based approach, which relies on the assumption that it is worth learning about one s experiences since one is likely to have similar experiences in the future (see, e.g. [Hammond, 1989; Kolodner, 1993; Kolodner Simpson, 1984; Ram, 1993; Schank, 1982] Opportunistic learning: A corollary of the active and experiential nature of goal driven learning is that learning is opportunistic. Often, a desired piece of knowledge will not be immediately available in the input, and so ....

[Article contains additional citation context not shown here]

K.J. Hammond. Case-Based Planning: Viewing Planning as a Memory Task. Perspectives in Artificial Intelligence. Academic Press, Boston, MA, 1989.

....used to supply control knowledge. System Gen CBP M I I DK CK SiN CHEF MI CBP NaCoDAE Prodigy Analogy SHOP SIPE II We first discuss the features shown columns 2 5 of Table 3. SHOP [Nau et al. 1999] as is typical of generative planners, requires a complete domain theory. CHEF [Hammond, 1989] and DIAL [Leake et al. 1997] are case based, but do not have a generative component, and thus need a large case base to perform well across a wide variety of problems. Prodigy Analogy [Veloso, 1994] DerSNLP [Ihrig Kambhampati, 1994] and Paris [Bergmann Wilke, 1995] integrate generative ....

Hammond, K. Case-based planning: Viewing planning as a memory task. Boston, MA: Academic Press, 1989.

.... have been mostly inductive approaches, such the work in instance based learning (Aha et al. 1991) memory based (Stanfill and Waltz, 1986) or examplat based (Porter et al. 1990) Only some of the work has been applied to planning, usually in the context of analogy or case based reasoning (Hammond, 1989, Hanks and Weld, 1995, Kambhampati, 1989, Kettler et al. 1994, Veloso, 1994a, Veloso, 1994b) Most of this work concern domain specific algorithms (Hammond, 1989) Also, although these approaches demonstrated some useful lazy learning behavior, they did not, as we have, compare lazy and eager ....

.... (Porter et al. 1990) Only some of the work has been applied to planning, usually in the context of analogy or case based reasoning (Hammond, 1989, Hanks and Weld, 1995, Kambhampati, 1989, Kettler et al. 1994, Veloso, 1994a, Veloso, 1994b) Most of this work concern domain specific algorithms (Hammond, 1989). Also, although these approaches demonstrated some useful lazy learning behavior, they did not, as we have, compare lazy and eager learning modes. The two main differences with these approaches are: control rules represent knowledge to control individual decisions, while cases allow controlling ....

Kristian J. Hammond. Case-based Planning: Viewing Planning as a Memory Task. Academic Press, New York, NY, 1989.

....the patient, and 0 means that this plan is either impossible or impractical to administer. Our approach is similar to the one used by researchers in case based reasoning where the solution to a new problem is found by adapting the solution to a similar old problem which has been previously solved [8]. To serve as the case base, we will build a library of treatment plans that have been administered in the past to patients at our institution. Radiation oncologists, physicists, and technicians will be asked to assign complexity scores to those plans. The plans will be stored in the plan library ....

Hammond KJ. Case-based Planning: View- ing Planning as a Memory Task. Boston, MA: Academic Press, 1989.

....robot and ROBBIE, but which could easily be activated. Additional objects in the world such as cars and features such as time constraints on ROBBIE s performance could also be added using the same basic simulator mechanisms. 3. 4 The planner ROBBIE s planner combines a typical case based planner #Hammond, 1989# with a reactive planning execution module #Firby, 1989#. Figure 3.9 outlines the planner s process. The case based planner takes a description of its current situation #the current starting and goal locations#, builds an index describing the situation, and ROBBIE and Its Domain 59 Adapt Case ....

....are then combined and executed as a single plan. Case retrieval: Once a goal has been created, ROBBIE must generate a route plan that describes a path from its starting location to the goal location. The plan is created by a case based planner structurally similar to the planning system CHEF #Hammond, 1989# 2 . The case based planner takes the goal #the new desired location and current situation# and builds an index which describes the situation in terms of 2 Previous case based route planners are discussed in Chapter 2. ROBBIE s Planner 66 its relevant features and how cases are stored in ....

[Article contains additional citation context not shown here]

Hammond, C. #1989#. Case-Based Planning: Viewing Planning as a Memory Task. Academic Press, San Diego.

....can occur, since insufficient time exists for online deliberation. Otherwise, an appropriate planning algorithm may be selected based on how much time remains before failure can occur. 46 For example, as shown in Figure 5 5, the moderate time planning algorithm might be a case based technique [28] for which worst case execution time is a function of case database size, or an anytime planner [15] 74] 75] which includes a careful definition of worst case execution time for development of a minimum quality plan to avoid failure. The full replanning algorithm may be a state46 In the ....

K. Hammond, Case-Based Planning: Viewing Planning as a Memory Task, (Academic Press, New York 1989).

....their target is domain content. Many methods have been proposed for selecting case indices. Some are included in the standard case addition process #CBR 1 #, as in the model based approach of Bhatta and Goel #1995#. Others, however, adjust current indices in response to performance de#ciencies. Hammond #1989# describes a failure driven method for explanation based selection of new indexing features. Likewise, Ihrig and Kambhampati #1997# describe a policy that explains plan replay failures in order to add features to check during future retrievals. These policies are conditional, on line, and make ....

....could involve countless factors#, but simply for why it is better than another component in the current context. Focusing the explanation process in this way is related to the common CBR idea of generating expectations for behavior and explaining only deviations from those expectations #e.g. #Hammond 1989; Ihrig and Kambhampati 1997; Schank 1982##. During future adaptations, this rationale will be provided with other information about the component, and will be used as an additional index when retrieving possible substitutions. Storing generated cases. Adapted cases are placed into the system s ....

K. Hammond. Case-Based Planning: Viewing Planning as a Memory Task. Academic Press, San Diego, 1989.

....knowledge. System Gen CBP M I I DK CK SiN CHEF MI CBP NaCoDAE Prodigy Analogy SHOP SIPE II We first discuss the features shown in columns 25 of Table 3. SHOP [Nau et al. 1999] as is typical of generative planners, requires a complete domain theory. CHEF [Hammond, 1989] and DIAL [Leake et al. 1997] are case based, but do not have a generative component, and thus need a large case base to perform well across a wide variety of problems. Prodigy Analogy [Veloso, 1994] DerSNLP [Ihrig Kambhampati, 1994] and Paris [Bergmann Wilke, 1995] integrate generative and ....

Hammond, K. Case-based planning: Viewing planning as a memory task. Boston, MA: Academic Press, 1989.

....Case based planning is an extension of Sussman s ideas of retrieval and repair and an application of case based reasoning to planning. This approach is also characterized as memory based because the organization of the memory of previous plans is changed during the planning process. CHEF [Ham86a, Ham86b, Ham89] (and WOK [Ham83] is a case based planner that builds new plans out of its memory of old ones. CHEF s domain is Chinese Szechwan cooking and its task is to build new recipes on the basis of users requests. CHEF s input is a set of goals for different tastes, textures, ingredients, and type of ....

....maximum reward. CADDY uses cost as a measurement of plan efficiency, which can be viewed as a negative reward. CADDY tries to find a plan with low estimated costs. 6.2 CADDY and Plan Adaptation We next compare CADDY with several plan adaptation systems. HACKER [Sus75] WOK [Ham83] CHEF [Ham86a, Ham86b, Ham89], PLEXUS [Alt86a, Alt86b, Alt88] GTD paradigm [Sim88a, Sim88b] and CADDY are similar in a number of respects. They are all single agent planning systems; they attempt to use past experiences to overcome current problems encountered; they use a collection of plans to serve as past experiences; ....

Kristian J. Hammond. Case--Based Planning : Viewing Planning as a Memory Task. Academic Press, Inc., San Diego, California, U.S.A., 1989.

....motivated by an implemented domain; in this case, cooking breakfast foods. This domain has a number of nice properties, beyond the fact that it will provide demonstrations of the problems that I want to discuss. First, it has been looked at by other researchers namely Hammond with the CHEF program [17] and Agre in the TOAST program [3] Second, the domain builds on the knowledge bases created for the two handed blocks world. Thus, the information constructed for that domain forms the core of the knowledge base for the cooking domain. Third, this domain presents an interesting example of an ....

....occurred in the plan. As Wilkins [42] has argued, it is much more convenient to use variables in a plan; collecting constraints on the objects that can fill that role in the plan until the object can be bound to a unique world entity. Given the well established use of variables in planning systems [8, 10, 17, 37, 38], this section will concentrate on their implementation in ItPlanS. Cooking often calls for the introduction of objects that are best treated as variables. For example, a recipe that calls for an egg doesn t specify which egg to use: any egg will be sufficient. In fact, a recipe that did specify ....

[Article contains additional citation context not shown here]

Kristian Hammond. Case-Based Planning: Viewing planning as a memory task. Academic Press, 1989.

....The Reasoner processes the example to be solved (which description may be augmented during the deductive problem solving) and its solution found by the DRS. The CB reasoner architecture is an instance of the general architecture of a case based planer adapted for solving the classification task (Hammond 1989; Agre 1995) The crucial points in its operation are the failure prediction, case retrieval and conflict reconciling. Failure Prediction The process of searching a solution by means of the second tier starts only when the current situation is recognized as dangerous , i.e. if there is a ....

Hammond, K. 1989. Case-Based Planning: Viewing Planning as a Memory Task. Academic Press.

.... salience of differences (Koton, 1988) PROTOS employed weights reflecting explanatory significance and prototypicality (Bareiss, 1989) CHEF, PERSUADER, and PRODIGY ANALOGY, in various ways, related the salience of similarities and differences to the existence or resolution of goal conflicts (Hammond, 1989, Sycara, 1987, Veloso, 1992) SWALE related them to expectation violating anomalous outcomes (Kass et al. 1986) and CREANIMATE to functionality (Edelson, 1992) We have made some progress in developing an AI CBR program s ability to make context sensitive determinations of the salience of ....

Hammond, K. (1989) Case-Based Planning -- Viewing Planning as a Memory Task. San Diego, CA: Academic Press.

....attempt to start the car, however, when turning the key fails, the agent does not have another plan to start the car. Therefore, it must find the problem with the existing plan and eliminate it (in this case the potato) People are able to repair their actions , because they have, as Hammond [6] points out, information about the causes of previous failures. By remembering these causes and their solutions, an agent will know more the next time a similar failure occurs. Simply put, if an agent is familiar with the common problems in starting a car (and their solutions) and if a car ....

Kristian Hammond. Case-Based Planning: Viewing planning as a memory task. Academic Press, 1989.

....9 rx [INROOM(bx,rx) INROOM(by,rx) INROOM(ROBOT,rx) 25, page 125] In this case, the precondition PUSHABLE(bx) is an executability condition. This is in sharp contrast with the majority of systems which assume that the agent is universally able to perform its actions. For example, [15, 21, 31, 33] don t encode the conditions under which the action is executable but rather only those conditions that generate the action. Applicability Conditions In his work on NONLIN [30] Tate has defined what he refers to as usewhen conditions. These conditions correspond to what Georgeff has referred to ....

....treatment has nothing to do with circumscription. 17 the action. In fact, they may correct several errors before eliminating the actual cause of the problem. Despite the varied and often difficult domains in which problems occur, people are able to solve them, because they have, as Hammond [15] points out, information about the causes of previous failures. By remembering these causes and their solutions, an agent will know more the next time a similar failure occurs. Simply put, if an agent is familiar with the common problems in starting a car (and their solutions) and if a car ....

[Article contains additional citation context not shown here]

Kristian Hammond. Case-Based Planning: Viewing planning as a memory task. Academic Press, 1989.

....as mission priorities and spacecraft characteristics evolve. Summary and Related Work Iterative algorithms have been applied to a wide range of computer science problems such as traveling salesman (Lin Kernighan, 1973) as well as Artificial Intelligence Planning (Chien DeJong, 1994, Hammond, 1989, Simmons, 1988, Sussman, 1973) Iterative repair algorithms have also been used for a number of scheduling systems. The GERRY GPSS system (Zweben et al., 1994, Deale et al. 1994) uses iterative repair with a global evaluation function and simulated annealing to schedule space shuttle ground ....

K. Hammond, "Case-based Planning: Viewing Planning as a Memory Task," Academic Press, San Diego, 1989.

....to learn adaptation knowledge from the case base. Adaptation knowledge can be learned using other domain knowledge if a system is given general adaptation knowledge rules which when applied result in specific adaptation cases, that can be stored for future use (see DIAL, Leake et al. 1995, and Hammond, 1989). The main advantage of this method is that the domain knowledge used could already be available in an existing knowledge based system. However, its main disadvantage lies in its reliance on a large amount of such domain knowledge. Adaptation knowledge could also be acquired directly from the ....

Hammond K. 1989. Case-Based Planning: Viewing Planning as a Memory Task.

....use indexes which describe the content of a case and reflect the current understanding of the reasoner. Often a case contains the description of a problem and its solution. If, in contrast, the case is a plan which contains a sequence of actions, we refer to the process as case based planning (Hammond 1989). To date, case based reasoning and case based planning have largely been treated as separate research issues. Case based reasoners do not typically interact with the environment and casebased planners do not typically interact with a case library reflecting a (partial) state of the environment. ....

....source case is valid for the transformed source case. Second, the reasoner has to modify the causal model to make it applicable to the target case, and it then has to ensure that the modified causal model is valid for this case. The first stage can be achieved by modifying the transformed plan (Hammond 1989). This stage has the following effect: in executing the modified plan the water pipe circuit is built by connecting the various components and a small test paddle wheel is placed after the valve. The test paddle wheel allows the planner to test whether the water runs through the valve. In the ....

Hammond, K. (1989). Case-Based Planning: Viewing Planning as a Memory Task. New York: 18 Academic Press.

....systems represent this domain knowledge by cases (Kolodner, 1993) They are appropriate for domains where a strong theory does not exist but past experience is accessible. A plan is a specific sequence of steps (or actions) with the aim of a goal achievement. Case Based Planning (CBP) systems (Hammond, 1989; Veloso, 1992) reuse past sequences of actions from past plans to construct new ones. There are some case based models for the creative process. Linda Wills and Janet Kolodner (Wills Kolodner, 1994) consider three steps in creative design: enumeration of several alternative solutions; ....

Hammond, K., (1989). Case-Based Planning: Viewing planning as a memory task, Academic Press, Inc., New York, New York, 1989.

....and those prototypes with which they are familiar. This task is made more diOEcult by the nature of heart sounds, which are low frequency and often low volume. These two aspects of students learning are well captured by the model of human cognition known as case based reasoning (Schank 1982; Hammond 1989). In this model, new experiences are understood by referring to previous experiences stored in memory. To retrieve a relevant experience from memory, the reasoner must possess a similarity metric by which dioeerent cases can be compared. Research reported in Bregman (Bregman 1994) suggests that ....

Hammond, K. J. (1989). Case-Based Planning: Viewing Planning as a Memory Task. Boston, Academic Press, Inc.

....conclusion that case based reasoning (CBR) is the knowledge transfer vehicle that can help us best in grasping and exploiting most pieces of knowledge that is related in some or other way to conflicts in manufacturing systems. CBR is based on remembering past experiences when solving new problems (Hammond 1989, Kolodner 1993) Cases are episodic descriptions of problems, their associated solutions and traces of the solution process (e.g. explanations) Retrieval of cases is governed by some kind of conceptual similarity. Beyond recalling, ideally it can perform the adaptation of previous cases, the ....

Hammond, K. J. (1989). Case-Based Planning-Viewing Planning as a Memory Task. Academic Press.

....by avoiding repetition of planning effort is largely regarded as a promising approach. Earliest research on planning from experience is the motion planning of the robot named Shakey [6] which is based on the STRIPS planning system [7] Other systems based on this approach include PLEXUS [1] CHEF [8] and PRIAR [9] The most serious bottlenecks of case based planners are retrieval and modification of a good case. Nebel and Koehler [13] argue that it is not possible to achieve a provable efficiency gain of plan reuse over plan generation in the worst case. They show that if planning from ....

....without taking advantage of other planning strategies. In fact, other techniques can be integrated with the presented approach to create a more flexible and efficient planning agent. For instance, the plan library can be indexed in various ways such as goals they achieve or problems they avoid [8], and the search for a case can be based on a combination of criteria. Stored cases can also be generalized such that each stored plan can match a wider range of problems [12] ....

K.J. Hammond, Case-Based Planning: Viewing Planning as a Memory Task (Academic Press, San Diego, CA, 1989).

....the common strategy under various configurations of system parameters such as memory size and communication restrictions. In the future, we see our research being extended in the following directions. If there is no strong domain theory, case based learning for planning and meta level control [10, 28] is useful and also applicable to learning coordination rules; the past positive control for the similar situation is modified and employed to guide the coordination controls for the current problem solving. Another direction is to understand appropriate coordination actions by statistical ....

K. J. Hammond, Case-Based Planning --- Viewing Planning as a Memory Task, Academic Press, 1989.

.... self design [2] this basic paradigm was later partially implemented as an approach to meta level control for a single agent interpretation system [10,11] and more recently completely implemented in [14] This approach also has strong similarities to the approach developed by Hammond [9] for modifying cases in his casebase of recipes based on failures detected when the recipe was used with substitute ingredients. which can be reviewed on line by the learning component. 5 In future work, we will add an additional phase that takes past rules in conjunction with new rules that we ....

Hammond, K. J., "CaseBased Planning: Viewing Planning as a Memory Task," Academic Press, 1989.

....approach has the benefit of letting users discover tradeoffs in the data because users can watch the pattern of the retrieved data change as values are manipulated. As discussed earlier, the closest precedent for our use of knowledge based methods in retrieval comes from casebased reasoning (CBR) (Hammond, 1989; Kolodner, 1993; Riesbeck Schank, 1989) A case based reasoning system solves new problems by retrieving old problems likely to have similar solutions. Researchers working on the retrieval of CBR cases have concentrated on developing knowledge based methods for precise, efficient retrieval of ....

Hammond, K. 1989. Case-based Planning: Viewing Planning as a Memory Task. Academic Press. Perspectives in AI Series, Boston, MA.

....for the current situation. Perhaps the reasoner chose the wrong strategy to, say, verify a hypothesis, or looked in the wrong place to gather evidence. Identifying the cause of a reasoning failure (or the cause of an expected success) is known as the blame (or credit) assignment problem (e.g. Hammond, 1989; Minsky, 1963; Weintraub, 1991) Traditional approaches in machine learning have assumed that the knowledge to be learned has already been identified by an external agent. In Winston s (1975) arch learning program, for example, the user decided that the concept of an arch was a useful concept to ....

....the outcome of the task. 3 the reasoner retrieves a structure that later proves inappropriate, it must specialize the indices to this structure so the retrieval will not recur in similar situations. Learning the right indices to organize knowledge in memory is known as index learning (e.g. Hammond, 1989; Bhatta Ram, 1991; Ram, 1992) To learn from such failures, the reasoning system uses a multistrategy learning approach in which it records and analyzes a declarative trace of its own reasoning process using a Trace Meta XP. The data structure holds explicit information concerning the manner ....

[Article contains additional citation context not shown here]

Hammond, K.J. Case-Based Planning: Viewing Planning as a Memory Task, Academic Press, Boston, MA, 1989.

....and so on. An alternative approach that avoids these problems is to hold #xed the plans for G 1 ; G n , and merge into them a new plan for G. The plan for G might be generated from scratch, or it might result from the completion of a skeletal plan, as, for example, in case based planning #Hammond 1989; Kambhampati Hendler 1992; Hanks Weld 1992; Veloso Carbonell 1994#. Our focus in this paper is on the process by which the new plan can be merged into a set of existing plans, so that the resulting set is consistent. The most well developed prior work on plan merging is that of Yang #1997, ....

Hammond, K. J. 1989. Case-Based Planning: Viewing Planning as a Memory Task. New York: Academic Press.

....to transfer from the retrieved analog to the problem at hand. One method of analogical transfer is to directly map the analog to the current problem [Gentner, 1983] This method also forms the basis of much recent work in case based reasoning [Kolodner and Simpson, 1989; Riesbeck and Schank, 1989; Hammond, 1989; Ashley and Rissland, 1988; Alterman, 1988] For example, in our earlier work on case based design, we showed how structure behavior function (SBF) models of physical devices can be used for directly mapping the designs of those devices to new problems [Goel, 1991a] An SBF model of a device ....

K. Hammond. Case-Based Planning: Viewing Planning as a Memory Task. Academic Press, Boston, MA, 1989.

....knowledge (BK) represented using combinations of XPs and scripts, to explain the story and to build a representation for it in its foreground knowledge (FK) When this task fails, a trace of the reasoning that preceded the failure is passed to the learning subsystem. A case based reasoning [5,34,45,83,87] subsystem within the learner uses past cases of introspective reasoning from the BK to explain the comprehension failure and to generate a set of learning goals. These goals, along with the trace, are then passed to a non linear planner [84,96,100] The planner subsequently builds a learning ....

....[16,32] uses three types of self knowledge when learning. Records of variable bindings maintain an implicit trace of system performance, justification structures provide the knowledge of the kinds of cognitive states and events needed to explain the system s behavior, and transformation rules [8,34] describe how the mostly implementation independent knowledge in justification structures corresponds to a particular agent s implementation. In the Meta AQUA system, however, TMXPs maintain reasoning traces explicitly, and most implementation dependent knowledge is avoided. Our approach to using ....

K. J. Hammond, Case-Based Planning: Viewing planning as a memory task (Academic Press, San Diego, CA, 1989).

....collision fronts that give a shortest path from a start point to a goal point. Dependence on speed is captured by including generally narrow time intervals on the move actions. In the context of planning, we view statecharts as scripts or schemas [Tu94] or possibly even previously recorded cases [Ha89], that are selected, modified, and applied as required by the case at hand. A key aspect of applying a statechart is to identify vertices in the accessibility graph with locations used in defining the states. Planning becomes a hierarchical process, and ideally as distributed as possible. Such ....

Hammond, K. J. ,Case-Based Planning: Viewing Planning as a Memory Task. New York: Academic Press, 1989.

....of what most people would intuitively call an everyday activity. From an AI perspective, cooking is a useful choice of domain because it has been employed previously in AI systems, thus facilitating comparison of techniques and examples employed by other research projects (e.g. Agre, 1988; Hammond, 1989a; Hammond et al. 1990] In addition to this, cooking is also is a domain that is commonly viewed as both an everyday activity to which the average person can relate, and also an intricate and involved activity when performed by experts. It thus affords a good vantage point from which to contrast ....

....has argued is one of the the primary characteristics of everyday activities. The AI paradigm most commonly associated with such reasoning mechanisms is known as Case Based Reasoning. Case based reasoning involves recalling and adapting previous experiences and making use of them in new situations [Hammond, 1989a; Reisbeck and Schank, 1989] To do this, planners employing a case based approach make extensive use of episodic memory to store and refer to the previous plans and the successes and failures that have occurred when they were used. Episodic memory is a type of memory that stores specific episodes ....

[Article contains additional citation context not shown here]

Hammond, Kristian J., Case-Based Planning: Viewing Planning as a Memory Task (Boston, MA: Academic Press), 1989. 277 pp.

....of similarities and differences between the current problem situation and retrieved cases. This failing has been recognised by CBR researchers, and several have emphasised the need to couple the two processes. Hammond suggested long ago that adaptability should be the criterion for retrieval [Hammond, 1989], Smyth and Keane et al. have developed an impressive and practical adaptation guided retrieval mechanism [Smyth and Keane, 1995] and Leake and his associates have researched similar issues e.g. Leake et al. 1997] However all these approaches all retain a numeric, metric concept of similarity ....

Hammond, K. J. (1989). Case-Based Planning: Viewing Planning as a Memory Task. Academic Press Inc.

.... (e.g. explanation based learning [DeJong and Mooney, 1986; Mitchell, Keller, and Kedar Cabelli, 1986] learning from explanation failures [Hall, 1988; VanLehn, Jones, and Chi, 1992] and analogical learning (e.g. analogy [Falkenhainer, 1989; Gentner, 1989] case based learning [Carbonell, 1986; Hammond, 1989] Recently, there has been much interest in multistrategy learning, and several approaches have been suggested (see, e.g. Michalski and Tecuci, in press] These approaches fall into four broad categories, which we call strategy selection models, toolbox models, cascade models, and single ....

K. J. Hammond. Case-Based Planning: Viewing Planning as a Memory Task. Perspectives in Artificial Intelligence. Academic Press, Boston, MA, 1989.

....solution, for example, is to replay the original itinerary from Los Angeles and modify the route either from Chicago or Denver. 1.2. 2 Redesign by Reverse Engineering Ullman and Gruber have studied the task of redesign by observing mechanical engineers at work and analyzing their protocols [14][29] Their study has illustrated the role of questions and conjectures in redesign. A question is defined as an interrogation by the subject about any aspect of design, and a conjecture is defined as a conclusion or an assumption about the design inferred by the subject from incomplete ....

K. Hammond, "Case-Based Planning: Viewing Planning as a Memory Task", Academic Press, 1989.

....does not record a rationale for the component as a whole (which could involve countless factors) but simply for why it is the better component in the current context. Focusing the explanation process in this way is related to the common idea in CBR of aiming explanations at expectation failures (Hammond, 1989; Leake, 1992; Schank, 1982) During future adaptations, this rationale will be provided with other information about the component, and it can also be used as an additional index when retrieving possible substitutions. Adapted cases are placed into the system s hierarchies of cases at the point ....

K. Hammond. Case-Based Planning: Viewing Planning as a Memory Task. Academic Press, San Diego, 1989.

....planning is well suited to workflow management, given the similarity between processes and task networks. HTN planning can be used to compose long range processes from smaller units, using the power of lookahead analysis to ensure the viability of the constructed process. ffl Case based planning (Hammond 1989; Veloso 1992) generates new plans for a given situation and task by retrieving solutions for similar problems from a previously defined case library, and then adapting them to meet the requirements of the current situation. As such, case based planning methods provide a way to build on experience ....

Hammond, K. 1989. Case-Based Planning: Viewing Planning as a Memory Task. Academic Press.

....engineering effort is the automatic learning of adaptation knowledge. Until now there are only few investigations in learning adaptation knowledge. Some approaches for learning adaptation knowledge can be found in DIAL (David Leake, 1993; Leake, 1995b; Leake, 1995a) or CHEF (Hammond, 1986; Hammond, 1989). These systems use knowledge intensive derivational analogy approaches (Carbonell, 1986; Veloso and Carbonell, 1993) to learn adaptation knowledge. Knowledge intensive means that these approaches require a lot of background and problem solving knowledge. For example in DIAL and CHEF 1 ....

Hammond, K. J. (1989). Case-Based Planning: Viewing Planning as a Memory Task. Academic Press, Boston, Massachusetts.

.... case, revises the retrieved case according the result required by the application, and retains the problem as a new case (Aamodt and Plaza, 1994) Although CBR has been applied to planning before, most CBR planners rely heavily on domain knowledge to manage the inherent complexity (Alterman, 1988; Hammond, 1989; Marks et al. 1988; Turner, 1988) Manual construction and maintenance of such a knowledge intensive system is both costly and difficult. An alternative is empirical learning, where the system accepts examples as input and produces concept descriptions (Michalski, 1983; Quinlan, 1986) Most ....

Hammond, K. J. (1989). Case-based planning: viewing planning as a memory task, Academic Press, Inc., San Diego, CA.

.... of a base case to solve a target problem rather than on the recognition of potential candidates from within a case base (Mostow, 1989) On the other hand work classed as CBR has tended to concern itself with relatively larger numbers of semantically similar cases from a single domain (e.g. CHEF Hammond, 1989). These cases also tend to embody a significant degree of structure. Consequently the difficulty of adaptation is alleviated somewhat and instead the retrieval process is a more immediate problem than in AR. So CBR systems are generally associated with single domain applications whereas AR is more ....

Hammond K. J. (1989) Case-Based Planning: Viewing Planning As A Memory Task, Boston: Academic Press.

....Figure 2 shows an instruction sheet for the servicing of a rear derailers bearing. Therefore it is natural to represent cases as a set of actions connected through temporal relations. The explicit use of the notion of actions was until now restricted to case based planning systems such as (Hammond, 1989; Veloso, 1992) but it seems that for a semantic retrieval this has to be used in diagnostic systems to. Furthermore the representation of the tools used in the actions as well as the different kinds of agents performing the actions is useful. Servicing reports and failure reports are special ....

....sequential information was developed in (Devanbu Litman, 1991) and is based on the inclusion of the languages recognised by finite state automata in order to determine if one sequence of actions subsumes another. This technique would suffice to represent and classify the plans used by CHEF (Hammond, 1989). 3.6. Overhead of Developing the Ontology One of the goals of CBR was to reduce the amount of knowledge engineering needed to build a running system. CBR is often sold as: you simply take the cases you already have, throw it into the CBR system and you have a running CBR system . The amount of ....

Hammond, K. J. (1989). Case-Based Planning - Viewing Planning as a Memory Task. Academic Press.

....problem solving cases to capture contexts. This latter difference provides a natural way to represent contexts which should facilitate knowledge acquisition and which potentially provides a tie to established machine learning approaches such as case based reasoning [e.g. Kolodner et al. . 1985; Hammond, 1989] An Approach to Context sensitive Reasoning Our model of the reasoning process process is diagrammed in Figure 1. Ignoring for the moment the role of contextual information, the process initially begins by the agent focusing its attention on a goal (or set of goals) on which to begin work. A ....

Hammond, K. J. (1989). Case-Based Planning: Viewing Planning as a Memory Task. Academic Press, Inc. (Harcourt Brace Jovanovich, Publishers), New York.

.... [41, 62] ffl Search space decomposition, e.g. 94, 109, 157] operator decomposition, e.g. 86] and competence level decomposition, e.g. 10] ffl Reactive planning via off line computation, e.g. 47, 55, 138] or use of deictic representations, e.g. 6] ffl Off line and on line learning, e.g. [45, 65, 105] Many of these approaches are partial in one way or another. For example, the search performed by a planner may be incomplete, it may produce parts of plans only (e.g. next action models) or it may rely on unproven or default assumptions. Progressive Horizon Planning (PHP) is an incremental ....

Hammond, K. J., Case-Based Planning: Viewing Planning as a Memory Task. Academic Press, New York, 1989.

No context found.

K J Hammond. Case-Based Planning: Viewing Planning as a Memory Task. Academic Press Inc., 1989.

No context found.

K. Hammond. Case-Based Planning: Viewing Planning as a Memory Task. Academic Press, 1989.

No context found.

K. J. Hammond. Case-Based Planning: Viewing planning as a memory task (Academic Press, San Diego, CA, 1989).

No context found.

Hammond, K. J. (1989). Case-based planning : viewing planning as a memory task (Perspectives in artificial intelligence v. 1 . Boston: Academic Press.

No context found.

Hammond, K. J. (1989). Case-based planning: Viewing planning as a memory task. Vol. 1. of Perspectives in artificial intelligence. San Diego, CA: Academic Press.

No context found.

Hammond, Kristian J. (1989a). Case-based planning: viewing planning as a memory task.

No context found.

Hammond, K. Case Based Planning: Viewing Planning as a Memory Task . Proceedings Case Based Reasoning Workshop Publ. Morgan Kaufmann. May (1988)

No context found.

Hammond, K. J. (1989a). Case-Based Planning: Viewing Planning as a Memory Task. Academic Press, Boston, MA.

No context found.

Hammond, K. J. Case-Based Planning: Viewing Planning as a Memory Task. Boston, MA: Academic Press; 1989.

First 50 documents  Next 50

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