Feature-based classifiers for design optimization

We present a design optimization method for systems with high-dimensional parameter spaces using inductive decision trees. The essential idea is to map designs into a relatively low-dimensional feature space, and to derive a classifier to search for high-performing design alternatives within this space. Unlike learning classifier systems that were pioneered by Holland and Goldberg, classifiers defined by inductive decision trees were not originally developed for design optimization. In this paper, we explore modifications to such classifiers to make them more effective in the optimization problem. We expand the notions of feature space, generalize the tree construction heuristic beyond the original information-theoretic definitions, increase the reliance on domain expertise, and facilitate the transfer of design knowledge between related systems. There is a relatively small but rapidly growing body of work in the use of inductive trees for engineering design; the method presented herein is complementary to this research effort.     

Organizational determinants of IT adoption in the pharmaceutical distribution sector

The aim of this paper is to propose new organizational factors that might explain the differences in the extent and the speed of IT adoption. With this in mind, we carried out an analysis of 16 cases in the pharmaceutical distribution sector in Spain. The results indicate that there are certain intangible assets that favour the introduction and development of IT. Among these are a frank and fluid communication between departments and members of the organization, low levels of conflict, the explicit support of top management towards IT adoption and learning and creative skills of IT-staff. In addition to these factors, we found others that we propose as catalysts of IT adoption. Among these we might mention the special relationship between the member-clients and the company in the case of cooperative firms.     

Uncertainty Matters: Computer Models at the Science–Policy Interface

The use of computer models offers a general and flexible framework that can help to deal with some of the complexities and difficulties associated with the development of water management plans as prescribed by the Water Framework Directive. However, despite the advantages modelling presents, the integration of information derived from models into policy is far away from being trivial or the norm. Part of the difficulties of this integration is rooted in the lack of confidence policy makers have on the incorporation of modelling information into policy formulation. In this paper we examine the reasons for this apparent lack of confidence and explore how some tools, presently in use, address this problem. We conclude that public confidence in models is highly dependent on the way uncertainties are addressed and suggest possible directions of action to improve the current situation. Four real case studies illustrate how computer models have been used in The Netherlands for carrying out management plans at regional and national scale. We suggest that the solution to integrate modelling information into policy formulation lies on both the modelling and the policy-making communities.     

Two-tier architecture for unconstrained handwritten character recognition

In this paper, we propose an approach that combines the unsupervised and supervised learning techniques for unconstrained handwritten numeral recognition. This approach uses the Kohonen self-organizing neural network for data classification in the first stage and the learning vector quantization (LVQ) model in the second stage to improve classification accuracy. The combined architecture performs better than the Kohonen self-organizing map alone. In the proposed approach, the collection of centroids at different phases of training plays a vital role in the performance of the recognition system. Four experiments have been conducted and experimental results show that the collection of centroids in the middle of the training gives high performance in terms of speed and accuracy. The systems developed also resolve the confusion between handwritten numerals.     

Formal Analysis of Streaming Downloading Protocol for System Upgrading

For a PC-mobile download system which is embedded with streaming download protocol, there are problems that the data cannot be transmitted correctly from the PC to the mobile, or the transmission is unacceptably slow. To solve these problems, we carry out a formal analysis for the protocol with some timing parameters and a given probability of message loss and unordered data using a probabilistic model checking tool PRISM. We introduce a technique to reduce the state space of the system modeling the protocol which is a network of probabilistic timed automata. The experimental results in PRISM give us a clear explanation to the problems, and are helpful in identifying the optimal parameter settings to meet industrial requirements.     

Development of nano indium tin oxide (ITO) grains by alkaline hydrolysis of In(III) and Sn(IV) salts

Indium tin oxide (ITO) nano powders of different compositions (In: Sn = 90: 10, 70: 30 and 50: 50) were prepared by heat treatment (300-450°C) of mixed hydroxides of In(III) and Sn(IV). The hydroxides were obtained by the reaction of aq. NH₃ with mixed aq. solutions of In(NO₃)₃ and SnCl₄. FTIR and TG/DTA studies revealed that powders existed as In(OH)₃H₂O—SnO₃H₂H₂O in the solid state and then they transformed to In₂O₃—SnO₂ via some metastable intermediates after 300°C. Cubic phase of In₂O₃ was identified by XRD for the oxides up to 30% of Sn. Particle size measurements of the solid dispersed in acetone and SEM study for microstructure showed that the oxides were in the nano range (55-75 nm) whereas the size range determined from Debye-Scherrer equation were 11–24 nm.     

A computational study of LP-based heuristic algorithms for two-dimensional guillotine cutting stock problems

In this paper we develop and compare several heuristic methods for solving the general two-dimensional cutting stock problem. We follow the Gilmore-Gomory column generation scheme in which at each iteration a new cutting pattern is obtained as the solution of a subproblem on one stock sheet. For solving this subproblem, in addition to classical dynamic programming, we have developed three heuristic procedures of increasing complexity, based on GRASP and Tabu Search techniques, producing solutions differing in quality and in time requirements. In order to obtain integer solutions from the fractional solutions of the Gilmore-Gomory process, we compare three rounding procedures, rounding up, truncated branch and bound and the solution of a residual problem. We have coded and tested all the combinations of algorithms and rounding procedures. The computational results obtained on a set of randomly generated test problems show their relative efficiency and allow the potential user to choose from among them, according to the available computing time. Rceived: January 9, 2001 / Accepted: December 10, 2001