Regularized variational Bayesian learning of echo state networks with delay&sum readout

In this work, a variational Bayesian framework for efficient training of echo state networks (ESNs) with automatic regularization and delay&sum (D&S) readout adaptation is proposed. The algorithm uses a classical batch learning of ESNs. By treating the network echo states as fixed basis functions parameterized with delay parameters, we propose a variational Bayesian ESN training scheme. The variational approach allows for a seamless combination of sparse Bayesian learning ideas and a variational Bayesian space-alternating generalized expectation-maximization (VB-SAGE) algorithm for estimating parameters of superimposed signals. While the former method realizes automatic regularization of ESNs, which also determines which echo states and input signals are relevant for "explaining" the desired signal, the latter method provides a basis for joint estimation of D&S readout parameters. The proposed training algorithm can naturally be extended to ESNs with fixed filter neurons. It also generalizes the recently proposed expectation-maximization-based D&S readout adaptation method. The proposed algorithm was tested on synthetic data prediction tasks as well as on dynamic handwritten character recognition.     

Cost-benefit analysis of an SLA mapping approach for defining standardized Cloud computing goods

Due to the large variety in computing resources and, consequently, the large number of different types of service level agreements (SLAs), computing resource markets face the problem of a low market liquidity. Restricting the number of different resource types to a small set of standardized computing resources seems to be the appropriate solution to counteract this problem. Standardized computing resources are defined through an SLA template. An SLA template defines the structure of an SLA, the service attributes, the names of the service attributes, and the service attribute values. However, since existing research results have only introduced static SLA templates so far, the SLA templates cannot reflect changes in user needs and market structures. To address this shortcoming, we present a novel approach of adaptive SLA matching. This approach adapts SLA templates based on SLA mappings of users. It allows Cloud users to define mappings between a public SLA template, which is available in the Cloud market, and their private SLA templates, which are used for various in-house business processes of the Cloud user. Besides showing how public SLA templates are adapted to the demand of Cloud users, we also analyze the costs and benefits of this approach. Costs are incurred every time a user has to define a new SLA mapping to a public SLA template due to its adaptation. In particular, we investigate how the costs differ with respect to the public SLA template adaptation method. The simulation results show that the use of heuristics within adaptation methods allows balancing the costs and benefits of the SLA mapping approach.     

Estimating spine shape in lateral sleep positions using silhouette-derived body shape models

This study aims at evaluating spinal alignment during sleep by combining personalized human models with mattress indentation measurements. A generic surface model has been developed that can be personalized based on anthropometric parameters derived from silhouette extraction. Shape assessment of the personalized surface models, performed by comparison with 3-D surface scans of the trunk, showed a mean unsigned distance of 9.77 mm between modeled and scanned surface meshes. The surface model is combined with an inner skeleton model, allowing the model to simulate distinct sleep postures. An automatic fitting algorithm sets the appropriate degrees of freedom to position the model on the measured indentation according to the adopted sleep posture. Validation on lateral sleep positions showed good intraclass correlations (0.73–0.88) between estimated and measured angular spinal deformations, indicating that silhouette-derived body shape models provide a valuable tool for the unobtrusive assessment of spinal alignment during sleep.Relevance to industry: A common drawback of the available techniques to assess spinal deformation on bedding systems is that they interfere with the actual sleep process. The current study presents a novel method based on silhouette-derived body shape models in order to estimate spine shape during sleep unobtrusively.     

Overall performance measurement in a supply chain: towards a supplier-prime manufacturer based model

This study deals with the supply chain (SC) overall performance expression. The developed idea concerns more particularly the performance of the manufactured products. Indeed, two companies or more contribute to the manufacturing of products that are generally assembled by the prime manufacturer company. Moreover, in the industrial practice, performance scorecards are defined with regard to each process; and the overall performance is neither expressed for each company, nor for the whole SC. We propose here to identify the SC overall performance to the combination of the performances of the different involved companies in the SC. Thus, in order to obtain a definition of such performance, we choose to focus first on the performance of the prime manufacturer. In this sense, the approach is based on the SCOR model for the handling of the main processes around the considered product manufacturing. The prime manufacturer performance is then defined as the aggregation of its involved processes?? performances. While the prime manufacturer performance is strongly dependent on the suppliers?? performance, we suggest the integration of the impacting supplier performance into the prime manufacturer scorecards. From an operational point of view, the MACBETH methodology is used to coherently express both processes and overall performances. More precisely, the Choquet aggregation integral operator is applied in order to model mutual interactions between processes. Finally, the expression of a bearing??s manufacturer performance illustrates the proposition.     

Maintaining Views in Object-Relational Databases

View materialization is an important way of improving the performance of query processing. When an update occurs to the source data from which a materialized view is derived, the materialized view has to be updated so that it is consistent with the source data. This update process is called view maintenance. The incremental method of view maintenance, which computes the new view using the old view and the update to the source data, is widely preferred to full view recomputation when the update is small in size. In this paper we investigate how to incrementally maintain views in object-relational (OR) databases. The investigation focuses on maintaining views defined in OR-SQL, a language containing the features of object referencing, inheritance, collection, and aggregate functions including user-defined set aggregate functions. We propose an architecture and algorithms for incremental OR viewmaintenance. We implement all algorithms and analyze the performance of them in comparison with full view recomputation. The analysis shows that the algorithms significantly reduce the cost of updating a vieww hen the size of an update to the source data is relatively small. Received 23 May 2000 / Revised 27 March 2001 / Accepted in revised form 30 April 2001 Correspondence and offprint requests to: Jixue Liu, School of Computer and Information Science, University of South Australia, Mawson Lakes, Adelaide SA5084, Australia. Email: jixue.liu@unisa.edu.auau     

Product family modeling for mass customization

With growing reliance on modeling in product development, it is imperative to describe product families in a cohesive way. In particular, mass customization calls for a close integration of product life cycle from customer recognition to delivery and services. This paper proposes a triple-view scheme for modeling product families. Technical challenges are discussed by comparing product family modeling with modeling single products. Individual modeling formalisms for different views are discussed. An example of product family modeling in power supply design is presented to illustrate the feasibility and potential of the proposed approach.     

Distributed data mining on grids: services, tools, and applications

Data mining algorithms are widely used today for the analysis of large corporate and scientific datasets stored in databases and data archives. Industry, science, and commerce fields often need to analyze very large datasets maintained over geographically distributed sites by using the computational power of distributed and parallel systems. The grid can play a significant role in providing an effective computational support for distributed knowledge discovery applications. For the development of data mining applications on grids we designed a system called Knowledge Grid. This paper describes the Knowledge Grid framework and presents the toolset provided by the Knowledge Grid for implementing distributed knowledge discovery. The paper discusses how to design and implement data mining applications by using the Knowledge Grid tools starting from searching grid resources, composing software and data components, and executing the resulting data mining process on a grid. Some performance results are also discussed.     

Soil Water Content and Dry Density by Time Domain Reflectometry

Soil water content and dry density are two important properties for compaction quality control. This paper presents a new method for determining soil water content and dry density using a single time domain reflectometry test, which is an improvement over that designated by ASTM D6780. This new method is based on simultaneous measurement of apparent dielectric constant and bulk electrical conductivity on the same soil sample. Calibration equations correlate these two parameters with soil gravimetric water content and dry density, which are simultaneously solved after adjusting field-measured conductivity to a standard conductivity. The method compensates for temperature effects. The test process takes about 3 min and all calculations are automated. Testing may be done in situ using a special probe that provides sufficient sampling volume or in a compaction mold adapted to form a probe. Laboratory and field tests results show this one-step method is a fast, accurate, and safe method for construction quality control.     

Solving the location-routing problem with simultaneous pickup and delivery by simulated annealing

The study concerns the location-routing problem with simultaneous pickup and delivery (LRPSPD) in which the pickup and delivery take place at the same time for each customer. The goal is to determine the facility locations and vehicle routes in order to minimise the total system cost as a sum of facility opening cost, vehicle fixed cost and vehicle travel cost. A simulated annealing (SA) heuristic is proposed for the problem and extensive computational experiments are conducted. The results show that the proposed SA effectively solves LRPSPD and outperforms existing exact approaches in terms of solution quality.