Abstract
In this paper we study the problem of constructing accurate decision tree models from data streams. Data streams are incremental tasks that require incremental, online, and any-time learning algorithms. One of the most successful algorithms for mining data streams is VFDT. In this paper we extend the VFDT system in two directions: the ability to deal with continuous data and the use of more powerful classification techniques at tree leaves. The proposed system, VFDTc, can incorporate and classify new information online, with a single scan of the data, in time constant per example. The most relevant property of our system is the ability to obtain a performance similar to a standard decision tree algorithm even for medium size datasets. This is relevant due to the any-time property. We study the behaviour of VFDTc in different problems and demonstrate its utility in large and medium data sets. Under a bias-variance analysis we observe that VFDTc in comparison to C4.5 is able to reduce the variance component. Copyright 2003 ACM.

Abstract
A network for multiplexing fiber Bragg gratings (FBGs) and intensity-modulated fiber-optic sensors with no need to distinguish between the two kinds of sensor is proposed and experimentally demonstrated. Two FBG sensors and two intensity-modulated sensors are wavelength-division multiplexed; the electrical phase of the output signal is measured as a common parameter for both types of sensor. Furthermore, the intensity sensors become power referenced, and the FBG sensors are interrogated by a low-cost technique. Low cross talk is achieved by use of a tunable optical filter at the detector. (C) 2003 Optical Society of America.

Abstract
In the manufacturing world, globalisation leads to a trend towards the reduction of batches and product life cycle, and the increase of part diversity, which are in conflict with other requirements, such as the cost reduction achieved with higher productivity. Thus, the challenge is to develop flexible, agile and intelligent management and control architectures that satisfy the referred requirements. The holonic manufacturing and the agent-based manufacturing approaches allow a new approach to the manufacturing problem, through concepts such as modularity, decentralisation, autonomy and re-use of control software components. ADACOR, one of the holonic architectures recently proposed, defines a set of autonomous and intelligent holons aiming to improve the performance of control system in industrial scenarios characterised by the frequent occurrence of unexpected disturbances. The formal modeling and validation of the specifications of the ADACOR-holons and of the interactions between these holons to implement the manufacturing control functions is of critical importance. In this paper, a formal methodology is introduced and applied to model the dynamic behaviour of the ADACOR-holon classes.

Abstract
This paper presents techniques for the design and implementation of domain specific languages. Our techniques are based on higher-order attribute grammars. Formal languages are specified in the classical attribute formalism and domain specific languages are embedded in the specification via higher-order attributes. We present a domain specific language for pretty-printing and we show how such language can be easily embedded in the specification of a powerful spreadsheet-like tool. From such specification an incremental implementation is automatically derived and the first results are presented. © 2003 IEEE.

Abstract
The demand for a low cost, portable and accurate instrumentation system for structural monitoring is increasing as civil structures appear to be monitored for their lifetime. Face to conventional solutions, the issue can be handled as to develop an interrogation system based on a CCD spectrometer. A method pointing out to improve the instrumentation characteristics of the whole system was developed The initial low wavelength resolution of the spectrometer (0.2 nm) is enhanced through a sub-pixel resolution algorithm. This algorithm, capable of a approximate to 200 times improvement in the resolution, is presented in this work. Final characteristics are validated within the calibration process. It is made a comparison between the results obtained with the designed system and others reported, using similar models. As the measurement quality depends on the interrogation method and associated wavelength resolution, future developments are planned They are described and analyzed in the final part of the paper.

Abstract
This paper presents a new method that deals with a supervised learning task usually known as multiple regression. The main distinguishing feature of our technique is the use of a multistrategy approach to this learning task. We use a clustering method to form sub-sets of the training data before the actual regression modeling takes place. This pre-clustering stage creates several training sub-samples containing cases that are "nearby" to each other from the perspective of the multidimensional input space. Supervised learning within each of these sub-samples is easier and more accurate as our experiments show. We call the resulting method clustered partial linear regression. Predictions using these models are preceded by a cluster membership query for each test case. The cluster membership probability of a test case is used as a weight in an averaging process that calculates the final prediction. This averaging process involves the predictions of the regression models associated to the clusters for which the test case may belong. We have tested this general multistrategy approach using several regression techniques and we have observed significant accuracy gains in several data sets. We have also compared our method to bagging that also uses an averaging process to obtain predictions. This experiment showed that the two methods are significantly different. Finally, we present a comparison of our method with several state-of-the-art regression methods showing its competitiveness.