D Lenat. The nature of heuristics. Artificial Intelligence, 19:189--249, 1982. 103  Home/Search   Document Not in Database   Summary   Related Articles   Check

....contrast to the proof of the Four Colour Theorem, the proof of the Robbins conjecture is short enough that it is possible to check it by hand. Computers can also be used to discover new knowledge. Perhaps the best example of this is the Assistant Mathematician program written by Lenat (e.g. [32]) in the 70 s. This program, which started with the notion of sets and some simple rules to determine what is interesting , managed to discover not only integers, addition, and multiplication, but also prime numbers and several elementary results from number theory, such as the Goldbach ....

D. B. Lenat, The nature of heuristics, Artificial Intelligence 19 (1982) 189--249.

....as such. As late as 1991, for instance, Lenat and Feigenbaum, responding to Smith s [42] critique of the context insensitive notion of meaning adopted for CYC, had this to say in rebuttal: Use dependent meaning does not imply that we have to abandon the computational framework of logic [25]. The implied moral was to stick to the universalized meanings of logical terms and sentences (as opposed to use dependent meanings of natural language utterances) and to maintain consistency at the cost of context. This turned out to be an illusive approach to the question of meaning. It took ....

Lenat, D. B., and Feigenbaum, E. A. On the Thresholds of Knowledge. Artificial Intelligence 47, 1-3, Jan. 1991, pages 185-250 (1991)

....corresponding element in the output space Y in unsupervised learning. The motivation for an unsupervised learning algorithm is to look for regularities in the training examples fx i g M i=1 ae X [148] Unsupervised learning algorithms can be divided into general discovery systems (e.g. AM [93], BACON [91] and those which perform 4 clustering [148] In the clustering problem [81, 63] the goal is to group or cluster the data into sets of like points. One hopes to obtain clusters revealing some sort of high level characterization of the points belonging to individual clusters. ....

D. B. Lenat. The ubiquity of discovery. Artificial Intelligence, 9:257--285, 1977.

....computerised decision support. 2. 2 KNOWLEDGE BASED SYSTEMS Our second disciplinary context for decision support systems is artificial intelligence (AI) The goals of AI are to study human intelligence and to build computational models that can simulate intelligent behaviour [Nilsson 1974, Lenat 1975, Newell and Simon 1976] Artificial intelligence is both theoretical research, i.e. the study of intelligent behaviour, and technology, i.e. knowledge engineering where models are implemented as computer programs, knowledge based systems (KBS) Knowledge based systems may also be called expert ....

Lenat D B, The ubiquity of discovery. Artificial Intelligence 9, 3, 1975, 257-285.

....virtual environments used in computer simulations. The sensor signals might be noisy. The robot wheels can slip on the floor while driving. There might be new objects in the robot environment. The robot has to deal with them: avoid them as obstacle or use them for specific actions. Classical AI [39] tries to describe the world in terms of symbolic relations. While this is a powerful way of handling abstract problems (e.g. scheduling work plans, optimizing parameters, etc. it will be cumbersome to create a detailed model of a noisy environment. Modeling all the small perturbations and ....

D. B. Lenat. Artificial intelligence. Scientific American, 273(3):80--83, September 1995.

....machine can be organised so as to exhibit general intelligent action is a complete unknown, there is little reason to take any figure seriously. If the project fails this may, as Guha and Lenat say, give us an indication about whether the symbolic paradigm is flawed and if so, how (Guha and Lenat 1990: 57) On the other hand it may not. The project may simply collapse under the sheer difficulty of the task, leaving us little the wiser about the truth or falsity of its fundamental assumptions. Believing as he does that at the present stage of the project the ontological, epistemological, and ....

....the sheer difficulty of the ontological, logical and epistemological problems that he has taken on. 1. Ontology Lenat records that the first five years of the project were spent primarily on the problem of devising an adequate ontology with which to underpin CYC s representational framework (Lenat and Feigenbaum 1991: 220). This not inconsiderable effort has, in my view, hardly scratched the surface. As Brian Cantwell Smith puts it, It s not so much that [Lenat et al. think that ontology is already solved, as that they propose, in a relatively modest time period, to accomplish what others spend lives on (Smith ....

[Article contains additional citation context not shown here]

Lenat, D.B., Feigenbaum, E.A. 1991a. 'On the Thresholds of Knowledge'. Artificial Intelligence, 47, pp.185-250.

.... learning is called supervised learning [Quinlan, 1986b; Michalski, 1983; Mitchell, 1982; Kibler and Aha, 1987; Rumelhart, Hinton, and Williams, 1986] in contrast to unsupervised learning where training examples are unlabeled or their memberships are unknown [Fisher, 1987; Feigenbaum, 1961; Lenat, 1977; Langley, Simon, and Bradshaw, 1987b] Supervised learning can solve two types of problem: classification problems in which labels are categorical [Quinlan, 1983; Breiman, Friedman, Olshen, and Stone, 1984] and regression problems in which labels are continuous [Scheffe, 1959; Breiman et al. ....

D.B. Lenat, The ubiquity of discovery. Artificial Intelligence, 9, 257-285.

....in the size and complexity of the computer system. At this time, software solutions to leading edge industrial applications are nearing the boundaries of present system engineering methodologies and seemingly insurmountable limitations to current generation intelligent systems are being observed [48, 59, 70]. These limitations have caused a fundamental rethink of the paradigms used to tackle large applications [22, 72] Some researchers have advocated the building of systems which embody vast amounts of common sense knowledge [31] others propose that developing sharable and reusable libraries of ....

D. B. Lenat and E. A. Feigenbaum, On the Thresholds of Knowledge, Artificial Intelligence 47 (1991) 185-250.

....the same regularities that were discovered in the training examples. The result of an unsupervised learning process is a set of class descriptions, one for each discovered class, that together cover all objects. Unsupervised learning is subdivided into discovery and clustering. ffl In discovery [20], algorithms extract implicit information (e.g genetic algorithms) ffl Clustering algorithms (see an overview in [14] identify structures or clusters present in the data. A cluster is defined as a set of similar objects. It is the most 8 commonly used technique in unsupervised learning. ....

D.B. Lenat. The ubiquity of discovery. Artificial Intelligence, 9:257--285, 1977.

....details of AM and to (Lenat and Brown 1984; Ritchie and Hanna 1984; Haase 1990; Shen 1990) for discussions of AM. AM s framework is general, initially being used in AM to discover quite a few new (to AM) mathematical concepts, such as prime numbers and Goldbach s conjecture. Further work by Lenat (Lenat 1990) demonstrated the framework s generality by adapting it to perform threedimensional VLSI design and to find winning strategies in warship fleet design (Lenat 1982B) The agenda and justification based framework has also been adapted to perform discovery in the domain of Conway numbers (Sims ....

....discovery type must also be provided. 6. 4 AM s malaise AM suffered from an unrecoverable malaise: after making all of its wonderful discoveries, the concepts AM created became less and less interesting and AM soon stopped because all of the tasks on its agenda were too uninteresting (Haase 1990; Lenat 1990). Investigation led Lenat to conjecture that the concepts AM was creating were becoming too different from its initial concepts, and that AM s heuristics, which were designed for its initial concepts, were not as applicable to the newer concepts. HAMB, or any program implementing our adapted ....

Lenat, D. B. 1982B. The Nature of Heuristics. Artificial Intelligence 19(2): 189-249.

....or from their experiences. Too often, useful knowledge is omitted the adequacy problem. A critical question is how they can know if the knowledge obtained is sufficient for a given domain. Perfect knowledge is impossible to obtain, but sufficient knowledge is essential for a KBDS to succeed [6, 8]. In a moderately complex domain, it is difficult to find out what s missing, based on a handful of test cases. Increasing the number of cases studied can be time consuming and is often impractical. Hence, a method is sought to alleviate the adequacy problem without increasing the number of cases ....

D.B. Lenat and E.A. Feigenbaum. On the thresholds of knowledge. Artificial Intelligence, 47(1-3), January 1991.

....be multiple potential segmentations. These are some of the very reasons why Bongard problems are so hard in the first place, and why progress has been so slow in this arena. Related analyses in the AI literature are, for instance, the incisive view presented in [28] of the much hyped project CYC [18], or the critical look provided in [3] of the structure mapping engine [5] Some of the issues discussed here have been previously brought by [12,14] On the side of philosophy, alternatives to metaphysical realism which discard the idea of a metaphysically external object have been proposed ....

....There is individuality only research, in what might be refereed to as the unity world [19] This is a world devoid of sounds, smells, and imagery; a world clear cut into unambiguous bounded objects and nothing else. Some projects that are representative of this arena are chess playing programs, CYC[18], expert systems, and RF4. On the other hand, there is particularity only research, in what might be referred to as the wave world [19] Some of the representative projects here are projects that take, for instance, an 22 05 00 25 image as input and obtain other images as a result, as in stereo ....

D.B. Lenat, and E.A. Feigenbaum, On the thresholds of knowledge, Artificial Intelligence 47 (1991) 185-250.

....to achieving the general goal of intelligent computing systems. We will refer to these as the symbolic computation approach and the neural networks approach. In very general terms, those pursuing the symbolic computation approach, whether from the logicist [294] or the knowledge is all there is [235] schools, have been concerned with developing advanced computing systems which can perform complex problem solving tasks (e.g. chess playing, theorem proving, medical diagnosis, chemical structure analysis, constraint satisfaction, rule based deduction) without particular regard to how those ....

D B Lenat and E A Feigenbaum. On the thresholds of knowledge. Artificial Intelligence, 47(1-3):185--250, January 1991.

....and relations are needed for the purpose. The items in a conceptualization represent the distinctions for stating and solving the problem. Reformulation changes distinctions: it restructures our knowledge of the world in terms of new conceptual elements. Traditional accounts of reformulation[Kor80, Mos81, Mar76a, New65, Len82] have only provided syntactic methods without the accompanying semantics. This thesis gives meaning to shifts in formulation by examining the shift in conceptualization it entails. By equating reformulation with reconceptualization we provide a clean Type 1 [Mar76b] theory of reformulation; i.e. ....

D.B. Lenat. The nature of heuristics. Artificial Intelligence, 19:189--249, 1982.

....from many different domains. Often, the major design improvements are obtained when the designer deviates from the standard approaches. It is difficult (if not impossible) to include this kind of reasoning into automatic tools. Artificial intelligence has striven to build up knowledge bases [ 123] for many years, but it is still not possible to build up knowledge bases for a large spectrum of different fields. Exactly this interdisciplinary knowledge is the crucial element in system level design. Only the successful combination of ideas from many different fields can make the design of the ....

D. B. Lenat. Artificial intelligence. Scientific American, 273(3):80--83, September 1995.

....metrics and methods for evaluating the contribution of prior knowledge to knowledge based systems. By prior knowledge we mean the knowledge one has available in an ontology or knowledge base prior to developing a knowledge based system. Several large ontologies have been developed including Cyc (Lenat,1995), Sensus (Knight, 1994) Ontolingua (Farquhar, 1996) All these systems contain hierarchies of knowledge. At the upper levels, one finds knowledge that is general to many applications, such as knowledge about movement, animate agents, space, causality, mental states, and so on. The lower levels ....

Lenat D. 1995. Artificial Intelligence. Scientific American.

....comprise immense problems, and much research remains to be done, there has been substantial progress toward each. Lenat has perhaps done the most work toward the first goal, developing common sense reasoning systems. A brief description of his latest effort related to information retrieval is in [3]. His system works with text, allowing a request for somebody wet to retrieve an annotated shot such as Garcia finishing a marathon. The second goal, based on experiments by Minka and Picard, appears to be best satisfied by a sys1 tem that works with a society of models [4] The society of ....

D. B. Lenat, "Artificial intelligence," Scientific American, pp. 80--82, Sept. 1995.

....human knowledge. The implication is that simulated cognition cannot be achieved in a system lacking the vagaries of human cognition. Smith answers his twelve questions in terms of (1) the traditional approach of predicate logic, 2) the CYC system proposed by Dougles B. Lenat and Edward Feigenbaum [Lenat and Feigenbaum, 1991], and (3) his admittedly ill defined minimum standard for an AI system, called embedded computation, or EC [Smith, 1991, page 259] The questions appear below, with Smith s answers provided in square brackets, followed by preliminary answers for SNePS, to be reviewed in the conclusion. 1. Does ....

D. B. Lenat and E. A. Feigenbaum. On the thresholds of knowledge. Artificial Intelligence, 47:200--250, 1991.

....streams) and numerous other research problems. Natural language queries and common sense systems also play a significant role; for example, Lenat s CYC common sense reasoning system can take a request such as find someone wet, and return an image with a label such as man finishing a marathon [1]. The computer vision solutions will work best if wisely integrated with solutions from these other domains. In [2] I overviewed the latest digital library research issues for the image processing community to address. The emphasis for that community is on finding models for simultaneous ....

D. B. Lenat, "Artificial intelligence," Scientific American, pp. 80--82, Sept. 1995.

....of research goals and methods, as well as serve as standards to judge other researchers works. Following are some representative opinions: Intelligence is the power to rapidly find an adequate solution in what appears a priori (to observers) to be an immense search space. Lenat and Feigenbaum, [17]) Artificial intelligence is the study of complex information processing problems that often have their roots in some aspect of biological information processing. The goal of the subject is to identify interesting and solvable information processing problems, and solve them. Marr, 18] AI ....

D. Lenat and E. Feigenbaum. On the thresholds of knowledge. Artificial Intelligence, 47:185--250, 1991.

No context found.

D Lenat. The nature of heuristics. Artificial Intelligence, 19:189--249, 1982. 103

No context found.

Douglas B. Lenat. The nature of heuristics. Artificial Intelligence, pages 189--249, 1982.

No context found.

Douglas B. Lenat. The nature of heuristics. Artificial Intelligence, 19:189--249, 1982.

No context found.

Lenat, D. B. "Artificial Intelligence." Scientific American (September 1995).

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