Abstract not found

From laser-scanned data to feature human model: a system based on

By analyzing the relationships among chance, weight of evidence and degree ofbelief, it is shown that the assertion \chances are special cases of belief functions " and the assertion \Dempster's rule can be used to combine belief functions based on distinct bodies of evidence " together lead to an inconsistency in Dempster-Shafer theory. To solve this problem, some fundamental postulates of the theory must be rejected. A new approach for uncertainty management is introduced, which shares many intuitive ideas with D-S theory, while avoiding this problem. 1

This paper is about the philosophical and methodological foundation of artificial intelligence (AI). After discussing what is a good "working definition", "intelligence" is defined as "the ability for an information processing system to adapt to its environment with insufficient knowledge and resources". Applying the definition to a reasoning system, we get the major components of Non-Axiomatic Reasoning System (NARS), which isasymbolic logic implemented in a computer system, and has many interesting properties that are closely related to intelligence. The definition also clari es the difference and relationship between AI and other disciplines, such as computer science. Finally, the definition is compared with other popular definitions of intelligence, and its advantages are argued.

Psychological evidence shows that probability theoryisnotaproper descriptive model of intuitive human judgment. Instead, some heuristics have been proposed as such a descriptive model. This paper argues that probability theory has limitations even as a normative model. A new normative model of judgment under uncertainty is designed under the assumption that the system's knowledge and resources are insu cient with respect to the questions that the system needs to answer. The proposed heuristics in human reasoning can also be observed inthis new model, and can be justi ed according to the assumption.

With conicting evidence, a reasoning system derives uncertain conclusions. If the system is open to new evidence, it faces additionally a higher-order uncertainty, because the rst-order uncertainty evaluations are uncertain themselves | they can be changed by future evidence. A new measurement, condence, is introduced for this higher-order uncertainty. It is de- ned in terms of the amount of available evidence, and interpreted and processed as the relative stability of the rst-order uncertainty evaluation. Its relation with other approaches of \reasoning with uncertainty " is also discussed. Keywords. condence, evidence, frequency interval, revision, inference, deduction, induction, abduction. 1

An experience-grounded semantics is introduced for an intelligent reasoning system, which is adaptive, and works with insufficient knowledge and resources. According to this semantics, truth and meaning are defined with respect to the experience of the system — the truth value of a statement indicates the amount of available evidence, and the meaning of a term indicates its experienced relations with other terms. The major difference between experience-grounded semantics and modeltheoretic semantics is that the former does not assume the sufficiency of knowledge and resources. This approach provides new ideas to the solution of some important problems in cognitive science.

We develop a general abductive description of testing models. We find that this testing process is fundamentally slow and cannot be conducted exhaustively. Consequently, we argue that the usual case for model testing is nonexhaustive testing; i.e. some subset of the possible tests are chosen and executed. Note that if the tests result in model refinement, then different tests can result in different models. This leads to the hypothesis that different individuals form different "opinions" (i.e. models) about the world as a result of the different examples they push through their models. We prefer this symbolic explanation for situated semantics to non-symbolic proposals (e.g. neural).

The reference class problem in probability theory and the multiple inheritances (extensions) problem in non-monotonic logics can be referred to as special cases of con icting beliefs. The current solution accepted in the two domains is the speci city priority principle. By analyzing an example, several factors (ignored by the principle) are found to be relevant to the priority of a reference class. A new approach, Non-Axiomatic Reasoning System (NARS), is discussed, where these factors are all taken into account. It is argued that the solution provided by NARS is better than the solutions provided by probability theory and non-monotonic logics. 1

Model-theoretic semantics is inappropriate for adaptive systems working with insufficient knowledge and resources. An experience-grounded semantics is introduced in this paper, using NARS, an intelligent reasoning system, as a concrete example. In NARS, the truth value of a sentence indicates the amount of available evidence, and the meaning of a term indicates its experienced relationship with other terms. Accordingly, both truth value and meaning are dynamic and subjective. This approach provides new ideas to the solution of some important problems in artificial intelligence. 1 Introduction Semantics studies how the items in a language are related to the environment in which the language is used. Concretely, semantics is the theory of meaning and truth. To ask questions like "What is the meaning of a term?" and "What is the truth value of a sentence?", what we are looking for are the principles that determining meaning and truth in general, rather than the meaning of a specific wo...