Robust initialisation of Gaussian radial basis function networks using partitioned k-means clustering

Radial basis function networks can be trained quickly using linear optimisation once centres and other associated parameters have been initialised. The authors propose a small adjustment to a well accepted initialisation algorithm which improves the network accuracy over a range of problems. The algorithm is described and results are presented.

Digital signal processing of optical monitoring by means of a Kalman filter

A Kalman filter algorithm has been applied to interpret the optical reflectance excursions during vacuum deposition of infrared coatings and multilayer thin-film filters. The application has been described in detail elsewhere and this paper now reports on-line experience for estimating deposition rate and thickness. The estimation proved sufficiently reliable to firstly 'navigate' regular manufacture (as controlled by a skilled operator) and to subsequently reproduce the skill without interpretation or intervention whilst maintaining exemplary product quality. Optical control by means of this Kalman filter application is therefore considered suitable as a basis for the automated manufacture of infrared coatings and multilayer thin-film filters.

How different are GM food accepters and rejecters really? A means-end chains application to yogurt in Germany

The purpose of the paper is to identify and describe differences in cognitive structures between consumer segments with differing levels of acceptance of genetically modified (GM) food. Among a sample of 60 mothers three segments are distinguished with respect to purchase intentions for GM yogurt: non-buyers, maybe-buyers and likely-buyers. A homogeneity test for the elicited laddering data suggests merging maybe- and likely-buyers, yielding two segments termed accepters and rejecters. Still, overlap between the segments’ cognitive structures is considerable, in particular with respect to a health focus in the evaluation of perceived consequences and ambivalence in technology assessment. Distinct differences are found in the assessment of benefits offered by GM food and the importance of values driving product evaluation and thus purchase decisions.

Means, motive and opportunity? Disentangling client influence on performance measurement appraisals

This paper investigates the extent to which clients were able to influence performance measurement appraisals during the downturn in commercial property markets that began in the UK during the second half of 2007. The sharp change in market sentiment produced speculation that different client categories were attempting to influence their appraisers in different ways. In particular, it was recognised that the requirement for open-ended funds to meet redemptions gave them strong incentives to ensure that their asset values were marked down to market. Using data supplied by Investment Property Databank, we demonstrate that, indeed, unlisted open ended funds experienced sharper drops in capital values than other fund types in the second half of 2007, after the market turning point. These differences are statistically significant and cannot simply be explained by differences in portfolio composition. Client influence on appraisal forms one possible explanation of the results observed: the different pressures on fund managers resulting in different appraisal outcomes.

Improvement of link failure restoration utilising multi-protocol label switching (MPLS) as a means to maintain quality of service (QOS)

This paper proposes a practical approach to the enhancement of Quality of Service (QoS) routing by means of providing alternative or repair paths in the event of a breakage of a working path. The proposed scheme guarantees that every Protected Node (PN) is connected to a multi-repair path such that no further failure or breakage of single or double repair paths can cause any simultaneous loss of connectivity between an ingress node and an egress node. Links to be protected in an MPLS network are predefined and a Label Switched path (LSP) request involves the establishment of a working path. The use of multi-protection paths permits the formation of numerous protection paths allowing greater flexibility. Our analysis examined several methods including single, double and multi-repair routes and the prioritization of signals along the protected paths to improve the Quality of Service (QoS), throughput, reduce the cost of the protection path placement, delay, congestion and collision. Results obtained indicated that creating multi-repair paths and prioritizing packets reduces delay and increases throughput in which case the delays at the ingress/egress LSPs were low compared to when the signals had not been classified. Therefore the proposed scheme provided a means to improve the QoS in path restoration in MPLS using available network resources. Prioritizing the packets in the data plane has revealed that the amount of traffic transmitted using a medium and low priority Label Switch Paths (LSPs) does not have any impact on the explicit rate of the high priority LSP in which case the problem of a knock-on effect is eliminated.

Epidemic K-Means clustering

The K-Means algorithm for cluster analysis is one of the most influential and popular data mining methods. Its straightforward parallel formulation is well suited for distributed memory systems with reliable interconnection networks. However, in large-scale geographically distributed systems the straightforward parallel algorithm can be rendered useless by a single communication failure or high latency in communication paths. This work proposes a fully decentralised algorithm (Epidemic K-Means) which does not require global communication and is intrinsically fault tolerant. The proposed distributed K-Means algorithm provides a clustering solution which can approximate the solution of an ideal centralised algorithm over the aggregated data as closely as desired. A comparative performance analysis is carried out against the state of the art distributed K-Means algorithms based on sampling methods. The experimental analysis confirms that the proposed algorithm is a practical and accurate distributed K-Means implementation for networked systems of very large and extreme scale.

Fault tolerant decentralised K-Means clustering for asynchronous large-scale networks

The K-Means algorithm for cluster analysis is one of the most influential and popular data mining methods. Its straightforward parallel formulation is well suited for distributed memory systems with reliable interconnection networks, such as massively parallel processors and clusters of workstations. However, in large-scale geographically distributed systems the straightforward parallel algorithm can be rendered useless by a single communication failure or high latency in communication paths. The lack of scalable and fault tolerant global communication and synchronisation methods in large-scale systems has hindered the adoption of the K-Means algorithm for applications in large networked systems such as wireless sensor networks, peer-to-peer systems and mobile ad hoc networks. This work proposes a fully distributed K-Means algorithm (EpidemicK-Means) which does not require global communication and is intrinsically fault tolerant. The proposed distributed K-Means algorithm provides a clustering solution which can approximate the solution of an ideal centralised algorithm over the aggregated data as closely as desired. A comparative performance analysis is carried out against the state of the art sampling methods and shows that the proposed method overcomes the limitations of the sampling-based approaches for skewed clusters distributions. The experimental analysis confirms that the proposed algorithm is very accurate and fault tolerant under unreliable network conditions (message loss and node failures) and is suitable for asynchronous networks of very large and extreme scale.

Toward monitoring the tropospheric temperature by means of a general circulation model

The recent global tropospheric temperature trend can be reproduced by climate models that are forced only by observed sea surface temperature (SST) anomalies. In this study, simulations with the Hamburg climate model (ECHAM) are compared to temperatures from microwave sounding units (MSU) and to reanalyses from the European Centre for Medium-Range Weather Forecasts. There is overall agreement of observed and simulated tropospheric temperature anomalies in many regions, in particular in the tropics and over the oceans, which lack conventional observing systems. This provides the opportunity to link physically different quantities, such as surface observations or analyses (SST) and satellite soundings (MSU) by means of a general circulation model. The proposed method can indicate inconsistencies between MSU temperatures and SSTs and has apparently done so. Differences between observed and simulated tropospheric temperature anomalies can partly be attributed to stratospheric aerosol variations due to major volcanic eruptions.