B-spline neural network based digital baseband predistorter solution using the inverse of De Boor algorithm

In this paper a new nonlinear digital baseband predistorter design is introduced based on direct learning, together with a new Wiener system modeling approach for the high power amplifiers (HPA) based on the B-spline neural network. The contribution is twofold. Firstly, by assuming that the nonlinearity in the HPA is mainly dependent on the input signal amplitude the complex valued nonlinear static function is represented by two real valued B-spline neural networks, one for the amplitude distortion and another for the phase shift. The Gauss-Newton algorithm is applied for the parameter estimation, in which the De Boor recursion is employed to calculate both the B-spline curve and the first order derivatives. Secondly, we derive the predistorter algorithm calculating the inverse of the complex valued nonlinear static function according to B-spline neural network based Wiener models. The inverse of the amplitude and phase shift distortion are then computed and compensated using the identified phase shift model. Numerical examples have been employed to demonstrate the efficacy of the proposed approaches.

The plastic self organising map

A novel extension to Kohonen's self-organising map, called the plastic self organising map (PSOM), is presented. PSOM is unlike any other network because it only has one phase of operation. The PSOM does not go through a training cycle before testing, like the SOM does and its variants. Each pattern is thus treated identically for all time. The algorithm uses a graph structure to represent data and can add or remove neurons to learn dynamic nonstationary pattern sets. The network is tested on a real world radar application and an artificial nonstationary problem.

Dealing with complexity: an overview of the neural network approach

The problem of complexity is particularly relevant to the field of control engineering, since many engineering problems are inherently complex. The inherent complexity is such that straightforward computational problem solutions often produce very poor results. Although parallel processing can alleviate the problem to some extent, it is artificial neural networks (in various forms) which have recently proved particularly effective, even in dealing with the causes of the problem itself. This paper presents an overview of the current neural network research being undertaken. Such research aims to solve the complex problems found in many areas of science and engineering today.

Multi-layer radial basis function networks: an extension to the radial basis function

This paper presents the initial research carried out into a new neural network called the multilayer radial basis function network (MRBF). The network extends the radial basis function (RBF) in a similar way to that in which the multilayer perceptron extends the perceptron. It is hoped that by connecting RBFs together in a layered fashion, an equivalent increase in ability can be gained, as is gained from using MLPs instead of single perceptrons. The results of a practical comparison between individual RBFs and MRBF's are also given.

Identification of nonlinear systems with a dynamic recurrent neural network

Two approaches are presented to calculate the weights for a Dynamic Recurrent Neural Network (DRNN) in order to identify the input-output dynamics of a class of nonlinear systems. The number of states of the identified network is constrained to be the same as the number of states of the plant.

Neural network basis function center selection using cluster analysis

This paper deals with the selection of centres for radial basis function (RBF) networks. A novel mean-tracking clustering algorithm is described as a way in which centers can be chosen based on a batch of collected data. A direct comparison is made between the mean-tracking algorithm and k-means clustering and it is shown how mean-tracking clustering is significantly better in terms of achieving an RBF network which performs accurate function modelling.

Relative order defines a topology for recurrent networks

This paper uses techniques from control theory in the analysis of trained recurrent neural networks. Differential geometry is used as a framework, which allows the concept of relative order to be applied to neural networks. Any system possessing finite relative order has a left-inverse. Any recurrent network with finite relative order also has an inverse, which is shown to be a recurrent network.

Neural network based control in food processing

A number of commonly encountered simple neural network types are discussed, with particular attention being paid to their applicability in automation and control when applied to food processing. In the first instance n-tuple networks are considered, these being particularly useful for high speed production checking operations. Subsequently backpropagation networks are discussed, these being useful both in a more familiar feedback control arrangement and also for such things as recipe prediction.

Analysis of protein networks in resting and collagen receptor (GPVI)-stimulated platelet sub-proteomes.

Proteomics approaches have made important contributions to the characterisation of platelet regulatory mechanisms. A common problem encountered with this method, however, is the masking of low-abundance (e.g. signalling) proteins in complex mixtures by highly abundant proteins. In this study, subcellular fractionation of washed human platelets either inactivated or stimulated with the glycoprotein (GP) VI collagen receptor agonist, collagen-related peptide, reduced the complexity of the platelet proteome. The majority of proteins identified by tandem mass spectrometry are involved in signalling. The effect of GPVI stimulation on levels of specific proteins in subcellular compartments was compared and analysed using in silico quantification, and protein associations were predicted using STRING (the search tool for recurring instances of neighbouring genes/proteins). Interestingly, we observed that some proteins that were previously unidentified in platelets including teneurin-1 and Van Gogh-like protein 1, translocated to the membrane upon GPVI stimulation. Newly identified proteins may be involved in GPVI signalling nodes of importance for haemostasis and thrombosis.

Simulated annealing technique for fast learning of SOM networks

The Self-Organizing Map (SOM) is a popular unsupervised neural network able to provide effective clustering and data visualization for multidimensional input datasets. In this paper, we present an application of the simulated annealing procedure to the SOM learning algorithm with the aim to obtain a fast learning and better performances in terms of quantization error. The proposed learning algorithm is called Fast Learning Self-Organized Map, and it does not affect the easiness of the basic learning algorithm of the standard SOM. The proposed learning algorithm also improves the quality of resulting maps by providing better clustering quality and topology preservation of input multi-dimensional data. Several experiments are used to compare the proposed approach with the original algorithm and some of its modification and speed-up techniques.

Understanding the early stages of the innovation diffusion process: awareness, influence and communication networks

The themes of awareness and influence within the innovation diffusion process are addressed. The innovation diffusion process is typically represented as stages, yet awareness and influence are somewhat under-represented in the literature. Awareness and influence are situated within the contextual setting of individual actors but also within the broader institutional forces. Understanding how actors become aware of an innovation and then how their opinion is influenced is important for creating a more innovation-active UK construction sector. Social network analysis is proposed as one technique for mapping how awareness and influence occur and what they look like as a network. Empirical data are gathered using two modes of enquiry. This is done through a pilot study consisting of chartered professionals and then through a case study organization as it attempted to diffuse an innovation. The analysis demonstrates significant variations across actors’ awareness and influence networks. It is argued that social network analysis can complement other research methods in order to present a richer picture of how actors become aware of innovations and where they draw their influences regarding adopting innovations. In summarizing the findings, a framework for understanding awareness and influence associated with innovation within the UK construction sector is presented. Finally, with the UK construction sector continually being encouraged to be innovative, understanding and managing an actor’s awareness and influence network will be beneficial. The overarching conclusion thus describes the need not only to build research capacity in this area but also to push the boundaries related to the research methods employed.

Modular non-computational-connectionist-hybrid neural network approach to robotic systems

Spiking neural networks are usually limited in their applications due to their complex mathematical models and the lack of intuitive learning algorithms. In this paper, a simpler, novel neural network derived from a leaky integrate and fire neuron model, the ‘cavalcade’ neuron, is presented. A simulation for the neural network has been developed and two basic learning algorithms implemented within the environment. These algorithms successfully learn some basic temporal and instantaneous problems. Inspiration for neural network structures from these experiments are then taken and applied to process sensor information so as to successfully control a mobile robot.

NeMo: a platform for neural modelling of spiking neurons using GPUs

Simulating spiking neural networks is of great interest to scientists wanting to model the functioning of the brain. However, large-scale models are expensive to simulate due to the number and interconnectedness of neurons in the brain. Furthermore, where such simulations are used in an embodied setting, the simulation must be real-time in order to be useful. In this paper we present NeMo, a platform for such simulations which achieves high performance through the use of highly parallel commodity hardware in the form of graphics processing units (GPUs). NeMo makes use of the Izhikevich neuron model which provides a range of realistic spiking dynamics while being computationally efficient. Our GPU kernel can deliver up to 400 million spikes per second. This corresponds to a real-time simulation of around 40 000 neurons under biologically plausible conditions with 1000 synapses per neuron and a mean firing rate of 10 Hz.

Fractional order modeling of large three-dimensional RC networks

An incidence matrix analysis is used to model a three-dimensional network consisting of resistive and capacitive elements distributed across several interconnected layers. A systematic methodology for deriving a descriptor representation of the network with random allocation of the resistors and capacitors is proposed. Using a transformation of the descriptor representation into standard state-space form, amplitude and phase admittance responses of three-dimensional random RC networks are obtained. Such networks display an emergent behavior with a characteristic Jonscher-like response over a wide range of frequencies. A model approximation study of these networks is performed to infer the admittance response using integral and fractional order models. It was found that a fractional order model with only seven parameters can accurately describe the responses of networks composed of more than 70 nodes and 200 branches with 100 resistors and 100 capacitors. The proposed analysis can be used to model charge migration in amorphous materials, which may be associated to specific macroscopic or microscopic scale fractal geometrical structures in composites displaying a viscoelastic electromechanical response, as well as to model the collective responses of processes governed by random events described using statistical mechanics.

Primary evolving networks and the comparative analysis of robust and fragile structures

In this paper we consider the structure of dynamically evolving networks modelling information and activity moving across a large set of vertices. We adopt the communicability concept that generalizes that of centrality which is defined for static networks. We define the primary network structure within the whole as comprising of the most influential vertices (both as senders and receivers of dynamically sequenced activity). We present a methodology based on successive vertex knockouts, up to a very small fraction of the whole primary network,that can characterize the nature of the primary network as being either relatively robust and lattice-like (with redundancies built in) or relatively fragile and tree-like (with sensitivities and few redundancies). We apply these ideas to the analysis of evolving networks derived from fMRI scans of resting human brains. We show that the estimation of performance parameters via the structure tests of the corresponding primary networks is subject to less variability than that observed across a very large population of such scans. Hence the differences within the population are significant.