Klaus P. Jantke. Algorithmic learning from incomplete information: Principles and problems. In Jurgen Dassow and Jozef Kelemen, editors, Machines, Languages, and Complexity, volume 381 of Lecture Notes in Computer Science, pages 188--207. SpringerVerlag, 1989.  Home/Search   Document Not in Database   Summary   Related Articles   Check

....procedure as a useful instance of the induction parameter decomposition placeholder in a divide and conquer algorithm schema. The second problem with induction based synthesis is that examples alone are too weak a specification approach. Incompleteness results are indeed abundant [3] 10] 41] [51] [68] 69] It is true that in Machine Learning in general, examples are often all one can hope for. But, as conveyed by Table 1, in synthesis, we usually have the setting of a human specifier who knows the intended relation. So s he probably knows quite a bit more about that relation, but can t ....

Klaus P. Jantke. Algorithmic learning from incomplete information: Principles and problems. In J. Dassow and J. Kelemen (editors), Machines, Languages, and Complexity, pages 188--207. LNCS 381, Springer-Verlag, 1989.

....to be of highest similarity. A sequence of user inputs should result in a sequence of concept changes. In the best case, this may converge to an a priori unknown target similarity concept. The scenario sketched is a typical instance of the general inductive inference scenario (cf. AS83] KW80] Jan89] of learning in the limit from usually incomplete information. In the present paper, we are more particularly faced to problems of language learning (cf. Vid94] The approach is even more particular, as we are interested in two level grammar concepts for generating graphs (see below) There ....

Klaus P. Jantke. Algorithmic learning from incomplete information: Principles and problems. In J. Dassow and J. Kelemen, editors, Machines, Languages, and Complexity, Lecture Notes in Computer Science, pages 188--207. Springer-Verlag, 1989.

....and tuning similarity measures is subject of the present investigation. The specific goal of our research work reported here is to gain a better understanding of the power and limitations of case based learning where stabilization of the acquired knowledge is essential (cf. Gol67] AS83] and [Jan89], e.g. for discussions of the stabilization phenomenon in learning) To allow for precise results which are easy to communicate, we have chosen the problem domain of learning formal languages. There is already a collection of topical results recently published (cf. JL93] SJL94] JL95] and ....

....discussion of almost all the technicalities we need, and [DJ96a] for a similar, but purely learningtheoretic investigation) Gol67] is the seminal paper underlying our learning paradigm invoked. From the large number of introductory and survey papers, the reader is directed to [AS83] or [Jan89], e.g. Here, we intend to introduce and clarify the basic concepts in an informal, but precise way. The target class of formal languages to be learnt is specified via some concept of acceptors: containment decision lists. These are our specific logical case memory systems focussed on throughout ....

Klaus P. Jantke. Algorithmic learning from incomplete information: Principles and problems. In Jurgen Dassow and Jozef Kelemen, editors, Machines, Languages, and Complexity, volume 381 of Lecture Notes in Computer Science, pages 188--207. SpringerVerlag, 1989.

....learning algorithms. However it seems not an easy task to verify that the modifications are indeed helpful. In TIC two prototypical types of learning algorithms were implemented: inductive algorithms and casebased algorithms. For introductions to inductive learning, see (Angluin Smith 1983) and (Jantke 1989); for volkerd informatik.uni leipzig.de jantke dfki.de Copyright c fl2000, American Association for Artificial Intelligence (www.aaai.org) All rights reserved. an overview of case based reasoning, see (Riesbeck Schank 1989) and (Kolodner 1993) We decided to restrict our attention to the ....

Jantke, K. P. 1989. Algorithmic Learning from Incomplete Information: Principles and Problems. In Dassow, J., and Kelemen, J., eds., Machines, Languages, and Complexity, volume 381 of LNCS, 188-- 207. Springer.

....is firmly based on three sources of intensive disciplinary work. First, there is our comprehensive approach towards virtual enterprise management systems (cf. AFHS95] Hin96] Arn96b] and [Arn96c] Second we adopt our former contributions to machine learning for knowledgebased systems (cf. Jan89] Jan95] JL96] Last but not least, the realizability of our conceptual ideas towards machine learning to advance virtual enterprise management systems is substantially based on software technological work on software tools supporting creative design of evolving systems and interfaces (cf. ....

K.P. Jantke. Algorithmic learning from incomplete information: Principles and problems. In J. Dassow and J. Kelemen, editors, Machines, Languages, and Complexity, volume 381 of Lecture Notes in Computer Science, pages 188--207. Springer-Verlag, 1989.

....based on three sources of intensive disciplinary work. First, there is our comprehensive approach towards virtual enterprise management systems (cf. AFHS95] Hin96] Arn96a] Arn96b] and [AU97] Second we adopt our former contributions to machine learning for knowledge based systems (cf. [Jan89], Jan95] and [JL96] Last but not least, the realizability of our conceptual ideas towards machine learning to advance virtual enterprise management systems is substantially based on our software technological work on interactive software tools supporting creative design of evolving interfaces ....

....3: Dragging one pad over another one (a) for activating some method (b) The operational method of the chart tool is activated. It reacts in graphically representing the data stored in the dropped pad. 5. 2 Learning Scenarios Revisited If some learning approach is developed systematically (cf. [Jan89], for some easy introduction) a collection of parameters have to be specified. Here, we focus on four essential questions, only: P1 What are the target objects of learning P2 In which way might incomplete information about some target object arrive P3 What are hypotheses generated by the ....

Klaus P. Jantke. Algorithmic learning from incomplete information: Principles and problems. In Jurgen Dassow and Jozef Kelemen, editors, Machines, Languages, and Complexity, volume 381 of Lecture Notes in Computer Science, pages 188--207. Springer-Verlag, 1989.

....above) is based on TR t (n) Only if it is recognized at time point t that there is no acceptable plan, therapy plan generation starts again upon TR t (n 1) with t (n 1) t. This quite simple basic scenario comes close to the standard scenario of inductive inference (cf. KW80] AS83] Jan89] Ang92] e.g. where sequences of information fragments are fed to an inductive inference machine which, in response, is generating sequences of hypotheses. Interestingly, there is a tradeoff between simplicity and usability of these scenarios. Although the first one is quite simple, it turns ....

Klaus P. Jantke. Algorithmic learning from incomplete information: Principles and problems. In Jurgen Dassow and Jozef Kelemen, editors, Machines, Languages, and Complexity, volume 381 of Lecture Notes in Computer Science, pages 188--207. Springer-Verlag, 1989.

....the behaviour of learning devices on growing information sequences. This motivates the limiting concepts that follow. Learnability is based on the technical concepts defined earlier. The reader may consult similar approaches in several related publications (cf. AS83] and [KW80] for an overview, Jan89] for an easy introduction, and [Jan92] for case based approaches) The following two concepts are both distinguished by the type of admissible information (similar to Definition 1) and the underlying semantics. In it is formalized the idea of collecting cases in a computable manner. Definition ....

Klaus P. Jantke. Algorithmic learning from incomplete information: Principles and problems. In J. Dassow and J. Kelemen, editors, Machines, Languages, and Complexity, Lecture Notes in Computer Science, pages 188--207. SpringerVerlag, 1989.

....incomplete information. Frequently, the information to be processed is assumed in a form (examples, which may be understood as a sequence of cases. This bridges the gap between inductive inference and case based reasoning. AS83] and [KW80] are excellent surveys on inductive inference. [Jan89] is an easy introduction. For the present paper, a brief introduction will do. 2.1 A Peep at Inductive Inference There are several areas of inductive inference. Among them, learning recursive functions is most thoroughly studied. In recursion theoretic inductive inference as invented by Gold in ....

Klaus P. Jantke. Algorithmic learning from incomplete information: Principles and problems. In J. Dassow and J. Kelemen, editors, Machines, Languages, and Complexity, Lecture Notes in Computer Science, pages 188--207. Springer-Verlag, 1989.

....with similarity concepts. Both constituents may be subject to learning. The specific goal of our research work reported here is to gain a better understanding of the power and limitations of case based learning where stabilization of the acquired knowledge is essential (cf. Gol67] AS83] and [Jan89], e.g. for discussions of the stabilization phenomenon in learning) To allow for precise results which are easy to communicate, we have chosen the problem domain of learning formal languages. There is already a collection of topical results recently published (cf. JL95] and [GJLS97] e.g. ....

....formalisms will be introduced almost informally (cf. GJLS97] for a detailled discussion of almost all the technicalities we need) Gol67] is the seminal paper underlying our learning paradigm invoked. From the large number of introductory and survey papers, the reader is directed to [AS83] or [Jan89], e.g. Here, we intend to introduce and clarify the basic concepts in an informal, but precise way. The target class of formal languages to be learnt is specified via some concept of acceptors: containment decision lists. The learning theoretic investigation in [SS92] has drawn our attention to ....

Klaus P. Jantke. Algorithmic learning from incomplete information: Principles and problems. In Jurgen Dassow and Jozef Kelemen, editors, Machines, Languages, and Complexity, volume 381 of Lecture Notes in Computer Science, pages 188--207. Springer-Verlag, 1989.

....and tuning similarity measures is subject of the present investigation. The specific goal of our research work reported here is to gain a better understanding of the power and limitations of case based learning where stabilization of the acquired knowledge is essential (cf. Gol67] AS83] and [Jan89] e.g. for discussions of the stabilization phenomenon in learning) To allow for precise results which are easy to communicate, we have chosen the problem domain of learning formal languages. There is already a collection of topical results recently published (cf. JL93] SJL94] JL95] and ....

....discussion of almost all the technicalities we need, and [DJ96a] for a similar, but purely learningtheoretic investigation) Gol67] is the seminal paper underlying our learning paradigm invoked. From the large number of introductory and survey papers, the reader is directed to [AS83] or [Jan89] e.g. Here, we intend to introduce and clarify the basic concepts in an informal, but precise way. The target class of formal languages to be learnt is specified via some concept of acceptors: containment decision lists. These are our specific logical case memory systems focussed on throughout ....

Klaus P. Jantke. Algorithmic learning from incomplete information: Principles and problems. In Jurgen Dassow and Jozef Kelemen, editors, Machines, Languages, and Complexity, volume 381 of Lecture Notes in Computer Science, pages 188--207. SpringerVerlag, 1989.

....to be of highest similarity. A sequence of user inputs should result in a sequence of concept changes. In the best case, this may converge to an a priori unknown target similarity concept. The scenario sketched is a typical instance of the general inductive inference scenario (cf. AS83] KW80] Jan89] of learning in the limit from usually incomplete information. In the present paper, we are more particularly faced to problems of language learning (cf. Vid94] Klaus P. Jantke Two Level Grammar Concepts for Design 5 The approach is even more particular, as we are interested in two level ....

Klaus P. Jantke. Algorithmic learning from incomplete information: Principles and problems. In J. Dassow and J. Kelemen, editors, Machines, Languages, and Complexity, Lecture Notes in Computer Science, pages 188--207. SpringerVerlag, 1989.

....of learning from incomplete information. This research area has its origins in the early papers [Sol64] and [Gol67] We consider [Gol67] the seminal publication of inductive inference. AS83] and [KW80] are very good surveys. Additionally, there is a large number of easy introductions, like [Jan89] for instance. Therefore, we won t give a new introduction into inductive inference. Instead, we stress some selected aspects which have guided our investigations below. The first aspect is the one of disproving instead of proving formulae. The second one is the problem of consistent vs. ....

Klaus P. Jantke. Algorithmic learning from incomplete information: Principles and problems. In J. Dassow and J. Kelemen, editors, Machines, Languages, and Complexity, Lecture Notes in Computer Science, pages 188--207. Springer-Verlag, 1989.

....Both constituents may be subject to learning. The specific goal of our research work reported here is to gain a better understanding of the power and limitations of case based learning where stabilization of the acquired knowledge is essential (cf. Gold, 1967) Angluin Smith, 1983) and (Jantke, 1989), e.g. for discussions of the stabilization phenomenon in learning) To allow for precise results which are easy to communicate, we have chosen the problem domain of learning formal languages. There is already a collection of topical results recently published (cf. Jantke Lange, 1993) ....

....we need, and (Dotsch Jantke, 1996) for a similar, but purely learning theoretic investigation) Gold, 1967) is the seminal paper underlying our learning paradigm invoked. From the large number of introductory and survey papers, the reader is directed to (Angluin Smith, 1983) or (Jantke, 1989), e.g. Here, we intend to introduce and clarify the basic concepts in an informal, but precise way. The target class of formal languages to be learnt is specified via some concept of acceptors: containment decision lists. The learning theoretic investigation in (Sakakibara Siromoney, 1992) has ....

Jantke, K. P. (1989). Algorithmic learning from incomplete information: Principles and problems. In Dassow, J. & Kelemen, J., (Eds.), Machines, Languages, and Complexity, volume 381 of Lecture Notes in Computer Science, pages 188--207. Springer-Verlag.

....the behaviour of learning devices on growing information sequences. This motivates the following limiting concepts. Learnability is based on the technical concepts defined before. The reader may consult similar approaches in several related publications (cf. AS83] and [KW80] for an overview, Jan89] for an easy introduction, and [Jan92] for case based approaches) The following two concepts are both distinguished by the type of admissible information according to Definition 1 and the underlying semantics. There is formalized the idea of collecting cases in a computable manner. Definition ....

Klaus P. Jantke. Algorithmic learning from incomplete information: Principles and problems. In J. Dassow and J. Kelemen, editors, Machines, Languages, and Complexity, Lecture Notes in Computer Science, pages 188--207. Springer-Verlag, 1989.

....and tuning similarity measures is subject of the present investigation. The specific goal of our research work reported here is to gain a better understanding of the power and limitations of case based learning where stabilization of the acquired knowledge is essential (cf. Gol67] AS83] and [Jan89] e.g. for discussions of the stabilization phenomenon in learning) To allow for precise results which are easy to communicate, we have chosen the problem domain of learning formal languages. There is already a collection of topical results recently published (cf. JL93] SJL94] JL95] and ....

....discussion of almost all the technicalities we need, and [DJ96a] for a similar, but purely learningtheoretic investigation) Gol67] is the seminal paper underlying our learning paradigm invoked. From the large number of introductory and survey papers, the reader is directed to [AS83] or [Jan89] e.g. Here, we intend to introduce and clarify the basic concepts in an informal, but precise way. The target class of formal languages to be learnt is specified via some concept of acceptors: containment decision lists. These are our specific logical case memory systems focussed on throughout ....

Klaus P. Jantke. Algorithmic learning from incomplete information: Principles and problems. In Jurgen Dassow and Jozef Kelemen, editors, Machines, Languages, and Complexity, volume 381 of Lecture Notes in Computer Science, pages 188--207. SpringerVerlag, 1989.

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