Learning automata are reinforcement learners belonging to the class of policy iterators. They have already been shown to exhibit nice convergence properties in a wide range of discrete action game settings. Recently, a new formulation for a continuous action reinforcement learning automata (CARLA) was proposed. In this paper, we study the behavior of these CARLA in continuous action games and propose a novel method for coordinated exploration of the joint-action space. Our method allows a team of independent learners, using CARLA, to find the optimal joint action in common interest settings. We first show that independent agents using CARLA will converge to a local optimum of the continuous action game. We then introduce a method for coordinated exploration which allows the team of agents to find the global optimum of the game. We validate our approach in a number of experiments.

Ontologies provide potential support for knowledge and content management on a P2P platform. Although we can design ontologies beforehand for an application, it is argued that in P2P environments static or predefined ontologies cannot satisfy the ever-changing requirements of all users. So we propose every user should make proposals for what kind of ontology is the most apt to his need. Collecting all these proposals (or votes) helps the drift of ontologies. This paper introduces OntoVote, a scalable distributed vote-collecting mechanism based on application-level broadcast trees, and describes how OntoVote can be applied to ontology drift on a P2P platform by discussing several problems involved in the voting process.

Visuo is an implemented Python program that models visual reasoning. It takes as input a description of a scene in words (e.g. ‘small dog on a sunny street’) and produces estimates of the quantitative magnitudes of the qualitative input (e.g. the size of the dog and the brightness of the street). We claim that reasoners transfer quantitative knowledge to new concepts from distributions of familiar concepts in memory. We also claim that visuospatial magnitudes should be stored as distributions over fuzzy sets. We show that Visuo successfully predicts quantitative knowledge to new concepts.

The knowledge engineering practices of one experienced company are outlined, commenting on differences in emphasis in some of the knowledge elicitation and AI research literatures.

The first International Workshop on Chance Discovery (CD) was held in the hot-springs resort of Matsue, Shimani Prefecture, Japan, on 22 May 2001, as part of the Fifteenth Annual Conference of the Japanese Society for Artificial Intelligence (JSAI-2001). Thirteen presentations were made to the workshop (Ohsawa, 2001), with 25 people attending. The majority of presenters and attendees were from Japan. An edited selection of the papers presented will be included in a volume of papers from the various International Workshops of JSAI-2001, to be published in the Springer series, Advanced Information Processing. A forthcoming special issue of the journal New Generation Computing will also address this topic.

The HealthAgents project aims to provide a decision support system for brain tumour diagnosis using a collaborative network of distributed agents. The goal is that through the aggregation of the small data sets available at individual hospitals, much better decision support classifiers can be created and made available to the hospitals taking part. In this paper, we describe the technicalities of the HealthAgents framework, in particular how the interoperability of the various agents is managed using semantic web technologies. On the broad scale the architecture is based around distributed data-mart agents that provide ontological access to hospitals’ underlying data that has been anonymized and processed from proprietary formats into a canonical format. Classifier producers have agents that gather the global data from participating hospitals such that classifiers can be created and deployed as agents. The design on a microscale has each agent built upon a generic-layered framework that provides the common agent program code, allowing rapid development of agents for the system. We believe that our framework provides a well-engineered, agent-based approach to data sharing in a medical context. It can provide a better basis on which to investigate the effectiveness of new classification techniques for brain tumour diagnosis.

Information services facilitate users to exploit applications over the network and access them from the remote system at the client side. In this paper, we describe the design and development of information services for novelty mining, which allows users to access the novel yet relevant information of a given topic. Several methodologies regarding novelty mining such as novelty scoring, novelty threshold, novelty feedback, and document-to-sentence technique are described. In addition to Web services, mobile information services are also described. Modelling and implementing information services for novelty mining are especially useful for users to reduce their information overload. We describe the challenging issue of decomposing the complex novelty mining application into several smaller and simpler modules, which are later implemented as services on the Web as well as mobile devices. After deploying our information services for novelty mining, test cases are provided to demonstrate the system. Our information services for novelty mining are confirmed to be helpful in increasing the efficiency of enterprise users in gathering novel information from incoming text. By studying the design and development of information services for novelty mining, we can benefit other developers in investigating effective techniques for developing enterprise services for other real-world applications.

System adaptivity is increasingly demanded in high-performance embedded systems, particularly in multimedia system-on-chip (SoC), owing to growing quality-of-service requirements. This paper presents a reactive control model that has been introduced in Gaspard, our framework dedicated to SoC hardware/software co-design. This model aims at expressing adaptivity as well as reconfigurability in systems performing data-intensive computations. It is generic enough to be used for description in the different parts of an embedded system, for example, specification of how different data-intensive algorithms can be chosen according to some computation modes at the functional level; and expression of how hardware components can be selected via the usage of a library of intellectual properties according to execution performances. The transformation of this model toward synchronous languages is also presented, in order to allow an automatic code generation usable for formal verification, based on techniques such as model checking and controller synthesis, as illustrated in the paper. This work, based on Model-Driven Engineering and the standard UML MARTE profile, has been implemented in Gaspard.

We have developed two different art applications as a way of guiding and testing of our technical work in crowd simulation and interface devices. In this paper, we describe both of them as well as the systems used to implement them. ‘Crowds’ is an immersive art installation with stereo projection based on our development of the specification of crowd behavior using XML and images, a proprietary tangible wind interface, and interactive real-time navigation. ‘UnderCrowds’ deals with large crowds, implemented in the graphics processor, and uses a crowd sensor to make the size of the virtual crowd proportional to the number of spectators crossing in front of the said sensor.

The theme of this paper is to review work carried out in the area of applying expert systems and artificial intelligence in electrical power engineering. It surveys expert systems and artificial-intelligence-based algorithms developed for solving decision problems in power network and generator fault diagnosis, reactive power and voltage control, restoration of power supply, determination of load circuits for shedding in under-frequency load shedding schemes, generator scheduling, allocation of loads and circuits, load-flow planning, generation expansion planning and distribution expansion planning.

In their paper “Some Comments on Rule Induction” Bratko and Michie address only three of the seven concerns that I put forward in my original article. Significantly, they have chosen to ignore the deeper and more important philosophical issues which I raised, while seeking to deflect my concerns by proclaiming their own privileged position of “being in the know” as experienced rule inductionists. This is a strategy well-known to philosophers and sociologists of science, unfortunately, it tends to stifle debate and avoids the serious consideration of constructive criticism. It is therefore important to respond to the comments made by Bratko and Michie and to re-focus attention on the central themes of my argument.

Structured objects are items with defined properties that are to be represented in a computer system. Research in Knowledge Representation (KR) and in Database Design (DB) has produced languages for describing structured objects. Although different in the particular means for defining properties, both areas share the goal of representing a part of the world in a structured way. Moreover, the rise of object-centred formalisms in the last decade has significantly influenced the convergence of languages.

In this paper we consider the role case-based reasoning has had in development of computer-aided instruction systems. We survey several case-based teaching systems, each of which is representative of a basic pedagogical principle that motivated its development. Firstly, 15 pedagogical principles are presented that were identified from the analysis of case-based teaching systems. We present some background to the principles, and indicate which systems they are incorporated in. Next, the teaching systems themselves are described, with emphasis on how case-based reasoning has been applied. Finally, we discuss some problems to be addressed when case-based planning is applied to lesson planning within a tutoring system.

Advances in artificial intelligence (AI) have great potential to help address societal challenges that are both collective in nature and present at national or transnational scale. Pressing challenges in healthcare, finance, infrastructure and sustainability, for instance, might all be productively addressed by leveraging and amplifying AI for national-scale collective intelligence. The development and deployment of this kind of AI faces distinctive challenges, both technical and socio-technical. Here, a research strategy for mobilising inter-disciplinary research to address these challenges is detailed and some of the key issues that must be faced are outlined.

The sources of information on the Web relating to Fine Art and in particular to Fine Artists are numerous, heterogeneous and distributed. Data relating to the biographies of an artist, images of their artworks, location of the artworks and exhibition reviews invariably reside in distinct and seemingly unrelated, or at least unlinked, sources. While communication and exchange exists, there is a great deal of independence between major repositories, such as museum, often owing to their ownership or heritage. This increases the individuality in the repository’s own processes and dissemination. It is currently necessary to browse through numerous different websites to obtain information about any one artist, and at this time there is little aggregation of Fine Art Information. This is in contrast to the domain of books and music, where the aggregation and re-grouping of information (usually by author or artist/band name) has become the norm. A Museum API (Application Programming Interface), however, is a tool that can facilitate a similar information service for the domain of Fine Art, by allowing the retrieval and aggregation of Web-based Fine Art Information, whilst at the same time increasing public access to the content of a museum’s collection. In this paper, we present the case for a pragmatic solution to the problems of heterogeneity and distribution of Fine Art Data and this is the first step towards the comprehensive re-presentation of Fine Art Information in a more ‘artist-centric’ way, via accessible Web applications. This paper examines the domain of Fine Art Information on the Web, putting forward the case for more Web services such as generic Museum APIs, highlighting this via a prototype Web application known as the ArtBridge. The generic Museum API is the standardisation mechanism to enable interfacing with specific Museum APIs.