Lebowitz, M. 1986a. Concept learning in a rich input domain: generalization-based memory. In Michalski, R.; Carbonell, J.; and Mitchell, T., editors, Machine Learning: An Artificial Intelligence Approach, volume II. Morgan Kaufmann. 193--214.  Home/Search   Document Not in Database   Summary   Related Articles   Check

....that this system provides a useful enhancement to any symbolic reasoning system because it generates and utilizes meaningful patterns from symbolic data in its concept creation process. The system requires no training to generate these results, and unlike other incremental systems such as UNIMEM [Leb86] does not require any supplied description of the input domain. In a separate research project [Ber01] we are working on a method for developing a grounded spatial organization of autonomously generated concepts. We would like to merge that system s controlled top down design [Ber02] with this ....

M. Lebowitz. Concept learning in a rich input domain: Generalization-based memory. In Machine Learning: An AI Approach, Volume II, pages 193 -- 213. Morgan Kaufman, 1986.

....groups called clusters, each cluster containing in some sense similar documents. Several clustering methods and techniques have been proposed so far, such as, for example, based on similarity measures [van Rijsbergen, 1979; Salton and McGill, 1983] neighborhoods [Voorhees, 1985] hierarchies [Lebowitz, 1987; Willett, 1988; Fisher and McKusick, 1989; Crawford et al. 1991; Tanaka et al. 1999] on matrix theory (diagonalisation, singular value decomposition [Deerwester et al. 1990] Retrieval is performed based on a cluster representative which may but need not be one of the cluster members. If ....

Lebowitz, M. (1987). Concept learning in a rich input domain: Generalization-based memory. In Carbonell, J.G., Michalski, R.S., and Mitchell, T.M. (eds.) Machine Learning, An Artificial Intelligence Approach, Vol. II., Morgan Kaufmann, 193-214.

....learning that allows the state change to be explained in the future via recognition. If an explanation cannot be created with the existing set of schemata, the system tries to create a new schema with theory driven learning. If all else fails, OCCAM attempts an empirical method derived from UNIMEM (Lebowitz, 1986). To understand the role of explanation based learning in OCCAM, consider how explanation based learning applies to the following example from Summerlin (1985) Hydrogen peroxide, iodine, and liquid detergent are added to water producing a large quantity of soap bubbles. Here, we assume there is ....

Lebowitz, M. (1986). Concept Learning in an rich input domain: Generalization-based Memory. In Michalski, R., Carbonell, J., & Mitchell, T. (Ed.), Machine Learning, An artificial Intelligence Approach, Vol 2. Los Altos, Ca.: Morgan Kaufman Publishers.

....in order to reach a certain degree of feasibility. Consequently, we highlight the assumptions each method makes in order to cope with the complexity issue. A large number of methods has been considered in the litterature. From them, less recent methods such as EPAM [2] COBWEB [3] UNIMEM [4], CLUSTER 2 [5] and concept analysis [6] are limited to unstructured descriptions such as 2 attribute value pairs, and are therefore inadequate for richer knowledge representations such as needed with structured objects. More recent methods have been proposed to address richer representations. ....

....attributes. Then CLUSTER 2 can be used. In order to proceed as such, the user must know in advance which attributes are clustering relevant, and which are not. Background knowledge on the clustering criterias is essential. Generally, such conditions are not satisfied. Lebowitz s RESEARCHER [4] system claims to deal with complex structural descriptions, but very little details are given. All the examples found in the litterature are based on structured objects which share a strong structural similarity (as with frame systems) Lebowitz does point to the difficulty of the matching ....

M. Lebowitz, "Concept Learning in a Rich Input Domain: Generalization Based Memory", R. Michalski, J. Carbonell, and T. Mitchell, Eds., Machine Learning: An A.I. Approach, Morgan Kaufmann, 193-214, 1986. 13

....make a paradigm shift and alter the model to account for as many exceptions as possible. This threshold is typically a certain number of failures (e.g. 2) and has the effect requiring a certain weight of evidence before change can be justified. Examples of systems using this approach are Unimem [24], Alfred [25] and by Emde [26] 4.2 Techniques for Dealing with Systematic Errors The task of dealing with systematic errors is related to that of dealing with random errors, in that it concerns the problem of finding an adequate model of observations. However, 2 This system is an example of a ....

M. Lebowitz. Concept learning in a rich input domain : generalization-based memory. In J. G. Carbonell, R. S. Michalski, and T. M. Mitchell, editors, Machine Learning, vol. 2, Tioga, Palo Alto, Ca, 1986.

....precondition. The existing classification algorithms are based upon two main strategies: based upon closeness or based upon sparseness (or separation) among the data. Some of the approaches in the literature of creation of classes from examples are: Feigenbaum s EPAM [5] Lebowitz s UNIMEM [6, 7] , Smyth s ITRULE [8] Fisher s COBWEB [9, 10] Gennari s CLASSIT [11] Sammuts s MARVIN [12] Michalski s INDUCE [13] Langley s BACON [14] Quinlan s ID3 [15] There exists a terminology confusion in the term classification , some authors use it as the arrangement or sorting of elements in ....

M. Lebowitz. Concept learning in a rich input domain: Generalization based memory. In R. S. Michalski, J. G. Carbonell, and Mitchell T. M., editors, Machine Learning : and Artificial Intelligence Approach, California, 1983. Morgan Kaufmann.

....14; I Gamma dont Gamma avoid Gamma it) 5 AN APPLICATION EXAMPLE 13 F irstWeek(cas Gamma 14; fried Gamma egg; cooked Gamma vegetables; toasted Gamma butter; 5. 2 A concept Concepts are learnt by the system by conceptual clustering with an algorithm close to that of GBM [Lebowitz 86] So the experimental network is a shared features network linking cases and concepts. An example of a concept is the following : Concept Gamma 28(x) Is Gamma a(concept Gamma 28; concept Gamma 24) Is Gamma a(concept Gamma 28; concept Gamma 11) Abstr(concept Gamma 28; concept Gamma ....

Lebowitz Michael, ((Concept Learning in a Rich Input Domain : Generalization-Based Memory)), In : Machine Learning : An Artificial Intelligence Approach, Vol 2. R.S. Michalski, J.G. Carbonell and T.M. Mitchell. (Edts.), Morgan Kaufmann, Los Altos, CA, 1986.

....features and finding their values, on the other hand, requires domain knowledge and is timeconsuming. On the whole, symptom and initiative fields are surface, while task and condition fields are deep. Case features are also classified as predictive or non predictive, as in CYRUS [11] and UNIMEM [13]. A distinctive feature of CHATKB is that cases were assessed for similarity and merged if possible during knowledge acquisition. During case generation (mapping free format CHAT cases to UPERIT templates) predictive fields were used to identify similarity, and thus to fill in values for deep ....

M. Lebowitz. Concept learning in a rich input domain: Generalization-based memory. In Ryszard S. Michalski, Jaime G. Carbonell, and Tom M. Mitchell, editors, Machine Learning: An Artificial Intelligence Approach, volume 2. Morgan Kaufmann Publishers, Inc., 1986.

....a set I of instances to present sequentially, and their descriptions di : I = di . 2. Find conceptual classes Cj that cluster these instances in categories. 3. Find an intensional definition Dj for each category, that characterises it. 4. Find a hierarchy H that organises theses classes. GBM [Lebowitz 86] is an incremental concept learning system which is particularly interesting for the present research because its methodology can also be considered as case based. As a matter of fact, this author proposed an implementation of the theory of dynamic memory, IPP [Lebowitz 83] which was a true ....

....outcome of human categorization evolution [Houd e 92] and the tendency to impose over constraining structures on the examples plays a leading role [Woo Kyoung and Medin 92] even if these structures are far more complex than those proposed in machine learning. 2. 4 Learning In the GBM system [Lebowitz 86] for example, learning occurs at the level of each element, or feature, of the description of a concept. By the means of a counter associated to each feature of a concept, each one of them may be confirmed, by incrementing, or unconfirmed, by decrementing, during the search for a new instance ....

[Article contains additional citation context not shown here]

Lebowitz Michael, ((Concept Learning in a Rich Input Domain : GeneralizationBased Memory)), In : Machine Learning : An Artificial Intelligence Approach, Vol 2. R.S. Michalski, J.G. Carbonell et T.M. Mitchell. (Edts.), Morgan Kaufmann, Los Altos, CA, 1986.

....and differences between examples in order to form generalizations; we refer to these as similarity based. Similarity based learning analyses data more or less syntactically, with little use of semantics. A few examples of such schemes can be found in Winston (1972) Michalski (1980) and Lebowitz (1986). In contrast, knowledge based methods use prior knowledge often called background knowledge in the form of a domain theory that guides the interpretation of new examples. If the domain theory is complete, of course, there is no new knowledge to learn: the theory already contains a full ....

Lebowitz, M. 1986. Concept learning in a rich input domain: Generalization-Based Memory.

....: they are static. An interesting challenge is that such a system can add classes during processing, by a dynamic abstraction from the cases. Machine learning inferences, such as an inductive method, permit to construct structures similar to classes. For example, incremental concept learning [6] can be used to learn concepts. These concepts can be defined in such a way that they correspond to the definition of classes. When concepts are used to gather, and index cases, such cases are instances of these classes, and then instanciation is a form of indexation [4] The relations between ....

Lebowitz Michael, "Concept Learning in a Rich Input Domain : Generalization-Based Memory", In : Machine Learning : An Artificial Intelligence Approach, Vol 2. R.S. Michalski, J.G. Carbonell and T.M. Mitchell. (Edts.), Morgan Kaufmann, Los Altos, CA, 1986.

No context found.

Lebowitz, M. 1986a. Concept learning in a rich input domain: generalization-based memory. In Michalski, R.; Carbonell, J.; and Mitchell, T., editors, Machine Learning: An Artificial Intelligence Approach, volume II. Morgan Kaufmann. 193--214.

No context found.

M. Lebowitz. Concept learning in a rich input domain: Generalization-based memory. In R.S. Michalski, J.G. Carbonell, and T.M. Mitchell, editors, Machine Learning: An AI Approach, Volume II, pages 193--214. Morgan Kaufmann, 1986.

No context found.

M. Lebowitz. Concept learning in a Rich Input Domain: Generalization-based memory, in: B. Boulay. Advances in artificial intelligence II. Elsevier Science Publishers B. V., 1986.

No context found.

Michael Lebowitz. Concept learning in a rich input domain: Generalization-based memory. In Ryszard Michalski, Jaime Carbonell, and Tom Mitchell, editors, Machine Learning II, pages 193--214. Morgan Kaufman, 1986. 115

No context found.

Lebowitz, M. 1986a. Concept learning in a rich input domain: generalization-based memory. In Michalski, R.; Carbonell, J.; and Mitchell, T., editors, Machine Learning: An Artificial Intelligence Approach, volume II. Morgan Kaufmann. 193--214.

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