Regularised mixture density networks for modelling wind direction

This technical report contains all technical information and results from experiments where Mixture Density Networks (MDN) using an RBF network and fixed kernel means and variances were used to infer the wind direction from satellite data from the ersII weather satellite. The regularisation is based on the evidence framework and three different approximations were used to estimate the regularisation parameter. The results were compared with the results by `early stopping'.

Bayesian methods for neural networks

Bayesian techniques have been developed over many years in a range of different fields, but have only recently been applied to the problem of learning in neural networks. As well as providing a consistent framework for statistical pattern recognition, the Bayesian approach offers a number of practical advantages including a potential solution to the problem of over-fitting. This chapter aims to provide an introductory overview of the application of Bayesian methods to neural networks. It assumes the reader is familiar with standard feed-forward network models and how to train them using conventional techniques.

Combining spatially distributed predictions from neural networks

In this report we discuss the problem of combining spatially-distributed predictions from neural networks. An example of this problem is the prediction of a wind vector-field from remote-sensing data by combining bottom-up predictions (wind vector predictions on a pixel-by-pixel basis) with prior knowledge about wind-field configurations. This task can be achieved using the scaled-likelihood method, which has been used by Morgan and Bourlard (1995) and Smyth (1994), in the context of Hidden Markov modelling

Marking high-technology market evolution through the foci of market stories:the case of local area networks

Previous research suggests that changing consumer and producer knowledge structures play a role in market evolution and that the sociocognitive processes of product markets are revealed in the sensemaking stories of market actors that are rebroadcasted in commercial publications. In this article, the authors lend further support to the story-based nature of market sensemaking and the use of the sociocognitive approach in explaining the evolution of high-technology markets. They examine the content (i.e., subject matter or topic) and volume (i.e., the number) of market stories and the extent to which content and volume of market stories evolve as a technology emerges. Data were obtained from a content analysis of 10,412 article abstracts, published in key trade journals, pertaining to Local Area Network (LAN) technologies and spanning the period 1981 to 2000. Hypotheses concerning the evolving nature (content and volume) of market stories in technology evolution are tested. The analysis identified four categories of market stories - technical, product availability, product adoption, and product discontinuation. The findings show that the emerging technology passes initially through a 'technical-intensive' phase whereby technology related stories dominate, through a 'supply-push' phase, in which stories presenting products embracing the technology tend to exceed technical stories while there is a rise in the number of product adoption reference stories, to a 'product-focus' phase, with stories predominantly focusing on product availability. Overall story volume declines when a technology matures as the need for sensemaking reduces. When stories about product discontinuation surface, these signal the decline of current technology. New technologies that fail to maintain the 'product-focus' stage also reflect limited market acceptance. The article also discusses the theoretical and managerial implications of the study's findings. © 2002 Elsevier Science Inc. All rights reserved.

Regularisation of mixture density networks

Mixture Density Networks are a principled method to model conditional probability density functions which are non-Gaussian. This is achieved by modelling the conditional distribution for each pattern with a Gaussian Mixture Model for which the parameters are generated by a neural network. This thesis presents a novel method to introduce regularisation in this context for the special case where the mean and variance of the spherical Gaussian Kernels in the mixtures are fixed to predetermined values. Guidelines for how these parameters can be initialised are given, and it is shown how to apply the evidence framework to mixture density networks to achieve regularisation. This also provides an objective stopping criteria that can replace the `early stopping' methods that have previously been used. If the neural network used is an RBF network with fixed centres this opens up new opportunities for improved initialisation of the network weights, which are exploited to start training relatively close to the optimum. The new method is demonstrated on two data sets. The first is a simple synthetic data set while the second is a real life data set, namely satellite scatterometer data used to infer the wind speed and wind direction near the ocean surface. For both data sets the regularisation method performs well in comparison with earlier published results. Ideas on how the constraint on the kernels may be relaxed to allow fully adaptable kernels are presented.

Neural networks: A principled perspective

Introductory accounts of artificial neural networks often rely for motivation on analogies with models of information processing in biological networks. One limitation of such an approach is that it offers little guidance on how to find optimal algorithms, or how to verify the correct performance of neural network systems. A central goal of this paper is to draw attention to a quite different viewpoint in which neural networks are seen as algorithms for statistical pattern recognition based on a principled, i.e. theoretically well-founded, framework. We illustrate the concept of a principled viewpoint by considering a specific issue concerned with the interpretation of the outputs of a trained network. Finally, we discuss the relevance of such an approach to the issue of the validation and verification of neural network systems.

Energy efficient architectures and techniques for green radio access networks

Energy consumption in wireless networks, and in particular in cellular mobile networks, is now of major concern in respect of their potential adverse impact upon the environment and their escalating operating energy costs. The recent phenomenal growth of data services in cellular mobile networks has exacerbated the energy consumption issue and is forcing researchers to address how to design future wireless networks that take into account energy consumption constraints. One fundamental approach to reduce energy consumption of wireless networks is to adopt new radio access architectures and radio techniques. The Mobile VCE (MVCE) Green Radio project, established in 2009, is considering such new architectural and technical approaches. This paper reports highlights the key research issues pursued in the MVCE Green Radio project.

Service differentiation for the bandwidth request scheme in fixed IEEE 802.16 networks

IEEE 802.16 standards have been developed as one of the technical solutions for broadband wireless access systems. It has high data rate, large network coverage, flexible QoS schemes and cheap network deployment. Various flexible mechanisms related to QoS provisioning have been specified for uplink traffic at the medium access control (MAC) layer in the standards. Among the mechanisms, contention based bandwidth request scheme can be used to indicate bandwidth demands to the base station for the non-real-time polling and besteffort services. These two services are used for most application with unknown traffic characteristics. Due to the diverse QoS requirements of those applications, service differentiation (SD) is anticipated over the contention based bandwidth request scheme. In this paper we investigate the SD with the bandwidth request scheme by means of assigning different channel access parameters and bandwidth allocation priorities. The effectiveness of the differentiation schemes are evaluated by simulations. It is observed that the initial backoff window can be efficient in SD, and if combined with the bandwidth allocation priority, the SD performances will be better. ©2008 IEEE.

Technology transfer through international manufacturing networks:valuing technology from an owner’s perspective

Technology is a key part of organisational knowledge and gives its owners their distinctive capabilities and competitive advantages. However, to best use these assets technology often needs to be transferred and shared with others through a form of technology collaboration. This raises the important question of how technology should be valued when it is being transferred. Technology valuation has become a critical issue in most transfer transactions. Transfer arrangements and terms of payment have a significant effect on the generation and sharing of joint benefits in commercial, technical and strategic aspects. In this paper the concept of “owner's value” is explored by highlighting its structure and components and assessing the importance of factors affecting value. The influence on technology valuation of the transfer arrangement, the associated terms of payment and the interaction between the shared benefits, cost and risks are discussed.

2R and 3R optical regeneration in 40 Gbit/s WDM terrestrial networks

We analyse a 2R regenerator using nonlinear-optical-loop-mirror and a 3R regenerator employing nonlinearly-enhanced amplitude modulator in 40Gbit/s WDM networks based on standard fibre (SMF). Characterization of one- (600km of SMF) and two-step regeneration is presented.

Learning and change in interorganizational networks:the case for network learning and network change

The ALBA 2002 Call for Papers asks the question ‘How do organizational learning and knowledge management contribute to organizational innovation and change?’. Intuitively, we would argue, the answer should be relatively straightforward as links between learning and change, and knowledge management and innovation, have long been commonly assumed to exist. On the basis of this assumption, theories of learning tend to focus ‘within organizations’, and assume a transfer of learning from individual to organization which in turn leads to change. However, empirically, we find these links are more difficult to articulate. Organizations exist in complex embedded economic, political, social and institutional systems, hence organizational change (or innovation) may be influenced by learning in this wider context. Based on our research in this wider interorganizational setting, we first make the case for the notion of network learning that we then explore to develop our appreciation of change in interorganizational networks, and how it may be facilitated. The paper begins with a brief review of lite rature on learning in the organizational and interorganizational context which locates our stance on organizational learning versus the learning organization, and social, distributed versus technical, centred views of organizational learning and knowledge. Developing from the view that organizational learning is “a normal, if problematic, process in every organization” (Easterby-Smith, 1997: 1109), we introduce the notion of network learning: learning by a group of organizations as a group. We argue this is also a normal, if problematic, process in organizational relationships (as distinct from interorganizational learning), which has particular implications for network change. Part two of the paper develops our analysis, drawing on empirical data from two studies of learning. The first study addresses the issue of learning to collaborate between industrial customers and suppliers, leading to the case for network learning. The second, larger scale study goes on to develop this theme, examining learning around several major change issues in a healthcare service provider network. The learning processes and outcomes around the introduction of a particularly controversial and expensive technology are described, providing a rich and contrasting case with the first study. In part three, we then discuss the implications of this work for change, and for facilitating change. Conclusions from the first study identify potential interventions designed to facilitate individual and organizational learning within the customer organization to develop individual and organizational ‘capacity to collaborate’. Translated to the network example, we observe that network change entails learning at all levels – network, organization, group and individual. However, presenting findings in terms of interventions is less meaningful in an interorganizational network setting given: the differences in authority structures; the less formalised nature of the network setting; and the importance of evaluating performance at the network rather than organizational level. Academics challenge both the idea of managing change and of managing networks. Nevertheless practitioners are faced with the issue of understanding and in fluencing change in the network setting. Thus we conclude that a network learning perspective is an important development in our understanding of organizational learning, capability and change, locating this in the wider context in which organizations are embedded. This in turn helps to develop our appreciation of facilitating change in interorganizational networks, both in terms of change issues (such as introducing a new technology), and change orientation and capability.

Gain switched multi-carrier transmitter and pilot tone based receiver for long reach access networks

A novel and cost effective long reach PON downlink scenario is proposed employing a multi-carrier transmitter and pilot tone aided direct detection at the receiver. Error free performance with QPSK and 50km transmission is presented. © 2012 OSA.

Design of fast core node processor for packet forwarding without header modification in optical networks

We present a design of a fast all-optical core-node processor that performs packet-forwarding in optical networks without header-modification. The design is based on bit-serial architecture using TOADs as logic-gates that perform modulo-arithmetic to forward packets.

Improved adaptivity and robustness in decentralised multi-camera networks

In this paper we present increased adaptivity and robustness in distributed object tracking by multi-camera networks using a socio-economic mechanism for learning the vision graph. To build-up the vision graph autonomously within a distributed smart-camera network, we use an ant-colony inspired mechanism, which exchanges responsibility for tracking objects using Vickrey auctions. Employing the learnt vision graph allows the system to optimise its communication continuously. Since distributed smart camera networks are prone to uncertainties in individual cameras, such as failures or changes in extrinsic parameters, the vision graph should be sufficiently robust and adaptable during runtime to enable seamless tracking and optimised communication. To better reflect real smart-camera platforms and networks, we consider that communication and handover are not instantaneous, and that cameras may be added, removed or their properties changed during runtime. Using our dynamic socio-economic approach, the network is able to continue tracking objects well, despite all these uncertainties, and in some cases even with improved performance. This demonstrates the adaptivity and robustness of our approach.

The potential for networks with capacities exceeding the nonlinear-Shannon limit

We examine the network benefits of various forms of nonlinearity compensation, showing that total network capacities that are more than double the capacity of a network with reaches determined by the Nonlinear-Shannon limit © 2015 OSA.