Abstract: Instance-based representations have been applied to numerous classification tasks with some success Most of these applications involved predicting a symbolic class based on observed attributes This paper presents an instance-based method for predicting a numeric value based on observed attributes We prove that, given enough instances, if the numeric values are generated by continuous functions with bounded slope, then the predicted values are accurate approximations of the actual values We demonstrate the utility of this approach by comparing it with a standard approach for value prediction The instance-based approach requires neither ad hoc parameters nor background knowledge

Abstract: Ordinal reasoning plays a major role in human cognition. This paper identifies an important class of classification problems of patterns taken from ordinal domains and presents efficient, incremental algorithms for learning the classification rules from examples. We show that by adopting a monotonicity assumption of the output with respect to the input, inconsistencies among examples can be easily detected and the number of possible classification rules substantially reduced. By adopting a conservative classification criterion, the required number of rules further decreases. The monotonicity and conservatism of the classification also enable the resolution of conflicts among inconsistent examples and the graceful handling of don't knows and don't cares during the learning and classification phases. Two typical examples in which the suggested classification model works well are given. The first example is taken from the financial domain and the second from machining. Le raisonnement ordinal joue un role important dans le traitement cognitif de l'information. Cet article traite d'une classe importante de problemes de classification des formes provenant de domaines ordinaux, et presente des algorithmes incrementiels efficaces pour l'apprentissage des regles de classification a partir d'exemples. Les auteurs demontrent qu'en adoptant une hypothese de monotonicite de la sortie en fonction de l'entree, les incoherences dans les exemples peuvent ětre facilement detectees et le nombre de regles de classification possibles considerablement reduit. En choisissant un critere de classification prudent, le nombre requis de regles decroit davantage. La monotonicite et la prudence de la classification permettent egalement la solution de conflits parmi les exemples incoherents, ainsi que le traitement habile des incertitudes et des indifferences durant les phases d'apprentissage et de classification. Deux exemples classiques dans lesquels le modele de classification propose fonctionne sont fournis. Le premier exemple est tire du domaine financier et le second de l'informatique.

Abstract: During incremental concept learning from examples, tentative hypotheses are formed and then modified to form new hypotheses. When there is a choice among hypotheses, bias is used to express a preference. Bias may be expressed by the choice of hypothesis language, it may be implemented as an evaluation function for selecting among hypotheses already generated, or it may consist of screening potential hypotheses prior to hypothesis generation. This paper describes the use of the third method. Bias is represented explicitly both as assumptions that reduce the space of potential hypotheses and as procedures for testing these assumptions. There are advantages gained by using explicit assumptions. One advantage is that the assumptions are meta-level hypotheses that are used to generate future, as well as to select between current, inductive hypotheses. By testing these meta-level hypotheses, a system gains the power to anticipate the form of future hypotheses. Furthermore, rigorous testing of these meta-level hypotheses before using them to generate inductive hypotheses avoids consistency checks of the inductive hypotheses. A second advantage of using explicit assumptions is that bias can be tested using a variety of learning methods. Durant l'apprentissage incremental de concepts a partir d'exemples, des hypotheses de travail sont d'abord elaborees, puis modifiees afin d'en former de nouvelles. Lorsqu'il faut effectuer un choix entre des hypotheses, le parti pris est utilise pour marquer une preference. Il peut ětre exprime par le choix du langage d'hypotheses, ětre utilise comme fonction d'evaluation pour choisir parmi des hypotheses deja produites, ou servir au filtrage d'hypotheses possibles avant la generation d'hypotheses. Cet article decrit l'utilisation de la troisieme methode. Le parti pris est exprime en termes de suppositions qui limitent l'etendue d'hypotheses possibles et de procedures pour verifier ces suppositions. L'utilisation des suppositions explicites presente des avantages. L'un d'eux est que les suppositions sont des hypotheses meta-niveau utilisees pour generer des hypotheses inductives futures, ainsi que pour choisir parmi des hypotheses inductives existantes. En verifiant ces hypotheses meta-niveau, le systeme developpe la capacite d'anticiper la forme d'hypotheses futures. Qui plus est, des essais rigoureux de ces hypotheses meta-niveau avant de les utiliser pour produire des hypotheses inductives eliminent la neessite de verifier ces hypotheses. Un second avantage d'utiliser des assomptions explicites est que le parti pris peut ětre verifiea l'aide d'une variete de methodes d'apprentissage.

Abstract: Compilation is a powerful preprocessing technique in the processing of recursions in knowledge-based systems. This paper develops a method of compiling and optimizing complex function-free linear recursions using a variable connection graph, the V-graph. It shows that a function-free recursion consisting of a linear recursive rule and one or more nonrecursive rules can be compiled to (1) a bounded recursion, in which recursion can be eliminated from the program, or (2) an n-chain recursion, whose compiled formula consists of one chain, when n= 1, or n synchronized compiled chains, when n > 1. The study is based on a classification of linear recursions and a study of the compilation results of each class. Using the variable connection graph, linear recursions are classified into six classes: acyclic paths, unit cycles, uniform cycles, nonuniform cycles, connected components, and their disjoint mixtures. Recursions in each class share some common properties in compilation. Our study presents an organized picture for the compilation of general function-free linear recursions. After compilation, the processing of complex linear recursions becomes essentially the processing of primitive n-chain recursions or bounded recursions to which efficient processing methods are available. La compilation est une technique de preetraitement puissante dans le traitement des recurrences a l'interieur des systemes a base de connaissances. Cet article elabore une methode de compilation et d'optimisation des recurrences lineaires sans fonction a l'aide d'un graphe de connexion variable. Il montre qu'une recurrence sans fonction comprenant une regle recursive lineaire et une ou plusieurs regles non recursives peut ětre compilee a (1) une recurrence liee dans laquelle la recurrence peut ětre eliminee du programme, ou a (2) une recurrence an chaines dont la formule compilee est constituee d'une chaine, lorsque n= 1, ou de n chaines compilees synchronisees lorsque n > 1. Cette etude est fondee sur une classification des recurrences lineaires et une analyse des resultats de compilation de chaque classe. A l'aide du graphe de connexion variable, les recurrences lineaires sont classees en six categories: parcours acycliques, cycles unitaires, cycles uniformes, cycles non uniformes, composantes reliees et leurs melanges non consecutifs. Les recurrences de chaque classe possedent certaines proprietes communes au niveau de la compilation. Cette etude presente un schema structure pour la compilation des recurrences lineaires generales sans fonction. Apres la compilation, le traitement de recurrences lineaires complexes devient essentiellement le traitement de recurrences an chaines primitives pour lesquelles des methodes efficaces existent.

Abstract: The modal logic LL was introduced by Halpern and Rabin as a means of doing qualitative reasoning about likelihood. Here the relationship between LL and probability theory is examined. It is shown that there is a way of translating probability assertions into LL in a sound manner, so that LL in some sense can capture the probabilistic interpretation of likelihood. However, the translation is subtle; several more obvious attempts are shown to lead to inconsistencies. We also extend LL by adding modal operators for knowledge. This allows us to reason about the interaction between knowledge and likelihood. The propositional version of the resulting logic LLK is shown to have a complete axiomatization and to be decidable in exponential time, provably the best possible. La logique modale LL a ete proposee par Halpern et Rabin comme moyen de proceder a un raisonnement qualitatif a propos de la vraisemblance. Dans cet article, la relation entre la logique modale LL et la theorie des probabilites est examinee. Les auteurs demontrent qu'il existe une facon de bien traduire des assertions probabilistiques en logique modale LL de facon a ce que cette derniere puisse saisir l'interpretation probabilistique de la vraisemblance. Cependant, cette traduction est subtile; plusieurs tentatives plus evidentes ont entraine des incoherences. Des operateurs modaux ont ete ajoutes a la logique modale LL afin de permettre un raisonnement sur l'interaction de la connaissance et de la vraisemblance. On a constate que la version propositionnelle de la logique resultante possedait une axiomatisation complete et s'averait un facteur decisif en temps exponentiel.

Abstract: Reasoning about defaults—implications that typically hold, but which may have exceptions—is an important part of commonsense reasoning. We present some parts of a theory of defaults, concentrating on distinctions between various subtle ways in which defaults can be defeated, and on inferences which seem plausible but which are not correct in all cases. To represent this theory in a formal system, it is natural to use the epistemic concept of self-belief. We show how to express the theory by a local translation into autoepistemic logic, which contains the requisite epistemic operators. Le raisonnement a propos des valeurs implicites est une partie importante du raisonnement de sens commun. Les auteurs presentent des extraits d'une theorie des valeurs implicites, en insistant sur les distinctions entre diverses facons subtiles d'annuler des valeurs implicites, et sur les inferences qui semblent plausibles mais qui ne sont pas toujours exactes. Afin de representer cette theorie a l'interieur d'un systeme formel, il est courant d'utiliser le concept epistemique de l'auto-croyance. Les auteurs expliquent comment presenter la theorie par une traduction en logique auto-epistemique contenant les operateurs epistemiques requis.

Abstract: This paper describes a qualitative technique for interpreting graphical data. Given a set of numerical observations regarding the behaviour of a system, its attributes can be determined by plotting the data and qualitatively comparing the shape of the resulting graph with graphs of system behaviour models. Qualitative data modeling incorporates techniques from pattern recognition and qualitative reasoning to characterize observed data, generate hypothetical interpretations, and select models that best fit the shape of the data. Domain-specific knowledge may be used to substantiate or refute the likelihood of hypothesized interpretations. The basic data modeling technique is domain independent and is applicable to a wide range of problems. It is illustrated here in the context of a knowledge-based system for well test interpretation. Cet article decrit une technique qualitative d'interpretation de donnees graphiques. A partir d'un ensemble de donnees numeriques portant sur le comportement d'un systeme, il est possible de determiner ses attributs en presentant les donnees sous forme graphique, puis en effectuant une comparaison qualitative de la courbe resultante avec des graphiques provenant de modeles comportementaux de systemes. La modelisation qualitative fait appel a des techniques de reconnaissance des formes et de raisonnement qualitatif pour caracteriser les donnees observees, generer des interpretations hypothetiques et choisir des modeles parmi les plus representatifs du format des donnees. Il est possible de recourir a des connaissances specifiques du domaine pour justifier ou refuter les probabilites d'interpretations hypothetiques. La technique fondamentale de modelisation est independante du domaine et s'applique a un vaste eventail de problemes. Elle est illustree ici dans le contexte d'un systeme de connaissances pour l'interpretation de tests de puits.

Abstract: Artificial neural networks capable of doing hard learning offer a new way to undertake automatic speech recognition. The Boltzmann machine algorithm and the error back-propagation algorithm have been used to perform speaker normalization. Spectral segments are represented by spectral lines. Speaker-independent recognition of place of articulation for vowels is performed on lines. Performance of the networks is shown to depend on the coding of the input data. Samples were extracted from continuous speech of 38 speakers. The error rate obtained (4.2% error on test set of 72 samples with the Boltzmann machine and 6.9% error with error back-propagation) is better than that of previous experiments, using the same data, with continuous Hidden Markov Models (7.3% error on test set and 3% error on training set). These experiments are part of an attempt to construct a data-driven speech recognition system with multiple neural networks specialized to different tasks. Results are also reported on the recognition performance of other trained networks, such as one trained on the E-set consonants. Les reseaux neuraux artificiels en mesure d'effectuer des apprentissages difficiles offrent une nouvelle facon d'effectuer la reconnaissance automatique de la parole. L'algorithme de la machine de Boltzmann et l'algorithme de propagation arriere d'erreurs ont ete utilises pour effectuer la normalisation du locuteur. Les segments spectraux sont representes par des lignes spectrales. La reconnaissance independante de l'endroit de l'articulation des voyelles par le locuteur est effectuee sur des lignes. La performance des reseaux s'est revelee dependante de l'encodage des donnees d'entree. Des echantillons ont ete tires du discours continu de 38 locuteurs. Le taux d'erreur obtenu (4,2% d'erreur dans les 72 echantillons avec la machine de Boltzmann, 6,9% d'erreur avec l'algorithme de propagation arriere d'erreurs) est inferieur a celui des experiences precedentes, effectuees a l'aide des měmes donnees et avec les modeles continus de Markov (7,3% d'erreur avec la serie d'essai et 3% d'erreur avec la serie de formation). Ces experiences font partie d'un programme en vue d'elaborer un systeme de reconnaissance de la parole guidee par les donnees qui serait dote de reseaux neuraux multiples specialises dans diverses taches. Cet article presente egalement des donnees sur la performance d'autres reseaux formes.

Abstract: In the past, Kripke structures have been used to specify the semantic theory of various modal logics. More recently, modal structures have been developed as an alternative to Kripke structures for providing the semantics of such logics. While these approaches are equivalent in a certain sense, it has been argued that modal structures provide a more appropriate basis for representing the modal notions of knowledge and belief. Since these notions, rather than the traditional notions of necessity and possibility, are of particular interest to artificial intelligence, it is of interest to examine the applicability and versatility of these structures. This paper presents an investigation of modal structures by examining how they may be extended to account for generalizations of Kripke structures. To begin with, we present an alternative formulation of modal structures in terms of trees; this formulation emphasizes the relation between Kripke structures and modal structures, by showing how the latter may be obtained from the former by means of a three-step transformation. Following this, we show how modal structures may be extended to represent generalizations of possible worlds, and to represent generalizations of accessibility between possible worlds. Lastly, we show how modal structures may be used in the case of a full first-order system. In all cases, the extensions are shown to be equivalent to the corresponding extension of Kripke structures. Au cours des annees passees, les structures de Kripke ont ete utilisees pour preciser la theorie semantique de diverses logiques modales. Comme solution de rechange aux structures de Kripke, des structures modales ont recemment eteelaborees pour constituer la semantique de ces logiques. Bien que ces methodes soient equivalentes dans une certaine mesure, certains sont d'avis que les structures modales constituent une base plus appropriee de representation des notions modales de la connaissance et des croyances. Puisque ces notions, contrairement aux notions traditionnelles de necessite et de possibilite, sont d'un interet particulier dans le domaine de l'intelligence artificielle, il est interessant d'examiner l'applicabilittet la polyvalence de ces structures. Cet article presente les rtsultats d'une etude des structures modales qui examine comment celles-ci pourraient ětre elargies pour tenir compte des generalisations des structures de Kripke. Pour commencer, les auteurs presentent une formulation des structures modales sous forme d'arbres; cette formulation met l'accent sur la relation entre les structures de Kripke et les structures modales, en montrant comment ces dernieres peuvent ětre derivees des premieres par le biais d'une transformation en trois etapes. Par la suite, ils demontrent comment des structures modales peuvent etre elargies de maniere a representer des generalisations de mondes possibles et des generalisations d'accessibilite entre des mondes possibles. Enfin, ils illustrent comment des structures modales peuvent etre utilisees dans le cas d'un systeme entier de premier ordre. Dans tous les cas, il est demontre que les extensions sont equivalentes a l'extension des structures de Kripke.

Abstract: This paper describes LEW (learning by watching), an implementation of a novel learning technique, and discusses its application to the learning of plans. LEW is a domain-independent learning system with user-limited autonomy that is designed to provide robust performance in realistic knowledge acquisition tasks in a variety of domains. It partly automates the knowledge acquisition process for different knowledge types, such as concepts, rules, and plans. The inputs to the system, which we call cues, consist of an environmental component and of pairs containing a problem and its solution. Unlike traditional forms of “learning from examples”, in which the system uses the teacher's answer to improve the result of a prior generalization of an example, LEW treats the problem-solution or question-answer instances, i. e., the cues themselves, as the basic units for generalization. Cet article traite de la mise en œuvre d'une nouvelle technique d'apprentissage et discute de son utilite au niveau de l'apprentissage de plans. LEW (learning by watching) est un systeme d'apprentissage independant du domaine a autonomie limitee par l'utilisateur, conp pour offrir une performance solide dans une varete de domaines relies a l'acquisition de connaissances. Il automatise en partie le processus d'acquisition de divers types de connaissances comme les concepts, les regles et les plans. Les donnees d'entree du systeme (appelees des indicateurs) comprennent une com-posante environnementale et des paires constitutees d'un probleme et de sa solution. Contrairement aux formes traditionnelles « d'apprentissage a partir d'exemples », dans lesquelles le systkme utilise la reponse du tuteur pour ameliorer le resultat de la generalisation precetdente d'un exemple, LEW traite les combinaisons probleme-solution et question-reponse, c'est-a-dire les indicateurs, comme les unites de base de la generalisation.

Abstract: Presentation d'une methode generale de recuperation de la figure a partir de l'ombre qui suppose que les surfaces sont lisses et que l'albedo est constant, comme ce serait le cas si une image de Mondrian etait peinte sur une surface courbe lisse. Cette methode combine la methode de recuperation de la figure de Brooks et Horn et celle de la recuperation de l'albedo a l'aide de la relaxation stochastique

Abstract: A definition of extended definite clause grammars and their relationship to unrestricted grammars are presented. A method for translating extended definite clause grammars describing unrestricted grammars into executable prolog programs is given. Three different parsing techniques are presented, and for each a complete presentation of how to incorporate unrestricted grammars in the actual formalism is done. Extended definite clause grammar is a powerful formalism usable for specifying grammars in natural language processing systems. L'auteur presente une definition des grammaires a clauses definies etendues et discute de leur relation avec les grammaires non restreintes. Une methode pour transposer les grammaires a clauses definies etendues decrivant des grammaires non restreintes en programmes en langage prolog est presentee. Trois differentes techniques d'analyse sont proposees; dans chacun des cas, une presentation complete sur la facon d'incorporer des grammaires non restreintes au formalisme actuel est effectuee. La grammaire a clauses definies etendues est un formalisme puissant qui peut servir a precker des grammaires dans les systemes de traitement du langage naturel.

Abstract: Although there are many arguments that logic is an appropriate tool for artificial intelligence, there has been a perceived problem with the monotonicity of classical logic. This paper elaborates on the idea that reasoning should be viewed as theory formation where logic tells us the consequences of our assumptions. The two activities of predicting what is expected to be true and explaining observations are considered in a simple theory formation framework. Properties of each activity are discussed, along with a number of proposals as to what should be predicted or accepted as reasonable explanations. An architecture is proposed to combine explanation and prediction into one coherent framework. Algorithms used to implement the system as well as examples from a running implementation are given.

Abstract: It is commonplace in artificial intelligence to divide an agent’s explicit beliefs into two parts: the beliefs explicitly represented or manifest in memory, and the implicitly represented or constructive beliefs that are repeatedly reconstructed when needed rather than memorized. Many theories of knowledge view the relation between manifest and constructive beliefs as a logical relation, with the manifest beliefs representing the constructive beliefs through a logic of belief. This view, however, limits the ability of a theory to treat incomplete or inconsistent sets of beliefs in useful ways. We argue that a more illuminating view is that belief is the result of rational representation. In this theory, the agent obtains its constructive beliefs by using its manifest beliefs and preferences to rationally (in the sense of decision theory) choose the most useful conclusions indicated by the manifest beliefs.