Robust control of nonlinear stochastic systems by modelling conditional distributions of control signals

We introduce a novel inversion-based neuro-controller for solving control problems involving uncertain nonlinear systems that could also compensate for multi-valued systems. The approach uses recent developments in neural networks, especially in the context of modelling statistical distributions, which are applied to forward and inverse plant models. Provided that certain conditions are met, an estimate of the intrinsic uncertainty for the outputs of neural networks can be obtained using the statistical properties of networks. More generally, multicomponent distributions can be modelled by the mixture density network. In this work a novel robust inverse control approach is obtained based on importance sampling from these distributions. This importance sampling provides a structured and principled approach to constrain the complexity of the search space for the ideal control law. The performance of the new algorithm is illustrated through simulations with example systems.

Polarisation mode dispersion correlations with the coarse-step method

Having a fixed differential-group delay (DGD) term b′ in the coarse-step method results in a repetitive pattern in the autocorrelation function (ACF). We solve this problem by inserting a varying DGD term at each integration step. Furthermore we compute the range of values needed for b′ and simulate the phenomenon of polarisation mode dispersion for different statistical distributions of b′. We examine systematically the modified coarse-step method compared to the analytical model, through our simulation results. © 2006 Elsevier B.V. All rights reserved.

Exploiting uncertainty in nonlinear stochastic control problem

This work introduces a novel inversion-based neurocontroller for solving control problems involving uncertain nonlinear systems which could also compensate for multi-valued systems. The approach uses recent developments in neural networks, especially in the context of modelling statistical distributions, which are applied to forward and inverse plant models. Provided that certain conditions are met, an estimate of the intrinsic uncertainty for the outputs of neural networks can be obtained using the statistical properties of networks. More generally, multicomponent distributions can be modelled by the mixture density network. Based on importance sampling from these distributions a novel robust inverse control approach is obtained. This importance sampling provides a structured and principled approach to constrain the complexity of the search space for the ideal control law. The developed methodology circumvents the dynamic programming problem by using the predicted neural network uncertainty to localise the possible control solutions to consider. Convergence of the output error for the proposed control method is verified by using a Lyapunov function. Several simulation examples are provided to demonstrate the efficiency of the developed control method. The manner in which such a method is extended to nonlinear multi-variable systems with different delays between the input-output pairs is considered and demonstrated through simulation examples.

A mixture density network approach to modelling and exploiting uncertainty in nonlinear control problems

In nonlinear and stochastic control problems, learning an efficient feed-forward controller is not amenable to conventional neurocontrol methods. For these approaches, estimating and then incorporating uncertainty in the controller and feed-forward models can produce more robust control results. Here, we introduce a novel inversion-based neurocontroller for solving control problems involving uncertain nonlinear systems which could also compensate for multi-valued systems. The approach uses recent developments in neural networks, especially in the context of modelling statistical distributions, which are applied to forward and inverse plant models. Provided that certain conditions are met, an estimate of the intrinsic uncertainty for the outputs of neural networks can be obtained using the statistical properties of networks. More generally, multicomponent distributions can be modelled by the mixture density network. Based on importance sampling from these distributions a novel robust inverse control approach is obtained. This importance sampling provides a structured and principled approach to constrain the complexity of the search space for the ideal control law. The developed methodology circumvents the dynamic programming problem by using the predicted neural network uncertainty to localise the possible control solutions to consider. A nonlinear multi-variable system with different delays between the input-output pairs is used to demonstrate the successful application of the developed control algorithm. The proposed method is suitable for redundant control systems and allows us to model strongly non-Gaussian distributions of control signal as well as processes with hysteresis. © 2004 Elsevier Ltd. All rights reserved.

Bayesian invariant measurements of generalisation for discrete distributions

Neural network learning rules can be viewed as statistical estimators. They should be studied in Bayesian framework even if they are not Bayesian estimators. Generalisation should be measured by the divergence between the true distribution and the estimated distribution. Information divergences are invariant measurements of the divergence between two distributions. The posterior average information divergence is used to measure the generalisation ability of a network. The optimal estimators for multinomial distributions with Dirichlet priors are studied in detail. This confirms that the definition is compatible with intuition. The results also show that many commonly used methods can be put under this unified framework, by assume special priors and special divergences.

The dynamics of matrix momentum

We analyse the matrix momentum algorithm, which provides an efficient approximation to on-line Newton's method, by extending a recent statistical mechanics framework to include second order algorithms. We study the efficacy of this method when the Hessian is available and also consider a practical implementation which uses a single example estimate of the Hessian. The method is shown to provide excellent asymptotic performance, although the single example implementation is sensitive to the choice of training parameters. We conjecture that matrix momentum could provide efficient matrix inversion for other second order algorithms.

Natural gradient matrix momentum

Natural gradient learning is an efficient and principled method for improving on-line learning. In practical applications there will be an increased cost required in estimating and inverting the Fisher information matrix. We propose to use the matrix momentum algorithm in order to carry out efficient inversion and study the efficacy of a single step estimation of the Fisher information matrix. We analyse the proposed algorithm in a two-layer network, using a statistical mechanics framework which allows us to describe analytically the learning dynamics, and compare performance with true natural gradient learning and standard gradient descent.

Statistical mechanics of low-density parity check error-correcting codes over Galois fields

A variation of low-density parity check (LDPC) error-correcting codes defined over Galois fields (GF(q)) is investigated using statistical physics. A code of this type is characterised by a sparse random parity check matrix composed of C non-zero elements per column. We examine the dependence of the code performance on the value of q, for finite and infinite C values, both in terms of the thermodynamical transition point and the practical decoding phase characterised by the existence of a unique (ferromagnetic) solution. We find different q-dependence in the cases of C = 2 and C ≥ 3; the analytical solutions are in agreement with simulation results, providing a quantitative measure to the improvement in performance obtained using non-binary alphabets.

The theory of on-line learning: a statistical physics approach

In this paper we review recent theoretical approaches for analysing the dynamics of on-line learning in multilayer neural networks using methods adopted from statistical physics. The analysis is based on monitoring a set of macroscopic variables from which the generalisation error can be calculated. A closed set of dynamical equations for the macroscopic variables is derived analytically and solved numerically. The theoretical framework is then employed for defining optimal learning parameters and for analysing the incorporation of second order information into the learning process using natural gradient descent and matrix-momentum based methods. We will also briefly explain an extension of the original framework for analysing the case where training examples are sampled with repetition.

Low density parity check codes: a statistical physics perspective

The modem digital communication systems are made transmission reliable by employing error correction technique for the redundancies. Codes in the low-density parity-check work along the principles of Hamming code, and the parity-check matrix is very sparse, and multiple errors can be corrected. The sparseness of the matrix allows for the decoding process to be carried out by probability propagation methods similar to those employed in Turbo codes. The relation between spin systems in statistical physics and digital error correcting codes is based on the existence of a simple isomorphism between the additive Boolean group and the multiplicative binary group. Shannon proved general results on the natural limits of compression and error-correction by setting up the framework known as information theory. Error-correction codes are based on mapping the original space of words onto a higher dimensional space in such a way that the typical distance between encoded words increases.

Velocity distributions in a plate heat exchanger channel

The development of a Laser Doppler Anemometer technique to measure the velocity distribution in a commercial plate heat exchanger is described. Detailed velocity profiles are presented and a preliminary investigation is reported on flow behaviour through a single cell in the channel matrix. The objective of the study was to extend previous investigations of plate heat exchanger flow patterns in the laminar range with the eventual aim of establishing the effect of flow patterns on heat transfer performance, thus leading to improved plate heat exchanger design and design methods. Accurate point velocities were obtained by Laser Anemometry in a perspex replica of the metal channel. Oil was used as a circulating liquid with a refractive index matched to that of the perspex so that the laser beams were not distorted. Cell-by-cell velocity measurements over a range of Reynolds number up to ten showed significant liquid mal-distribution. Local cell velocities were found to be as high as twenty seven times average velocity, contrary to the previously held belief of four times. The degree of mal-distribution varied across the channel as well as in the vertical direction, and depended on the upward or downward direction of flow. At Reynolds numbers less than one, flow zig-zagged from one side of the channel to the other in wave form, but increases in Reynolds number improved liquid distribution. A detailed examination of selected cells showed velocity variations in different directions, together with variation within individual cells. Experimental results are also reported on the flow split when passing through a single cell in a section of a channel . These observations were used to explain mal-distribution in the perspex channel itself.

Matrix-free mass spectrometric imaging using laser desorption ionisation Fourier transform ion cyclotron resonance mass spectrometry

Mass spectrometry imaging (MSI) is a powerful tool in metabolomics and proteomics for the spatial localization and identification of pharmaceuticals, metabolites, lipids, peptides and proteins in biological tissues. However, sample preparation remains a crucial variable in obtaining the most accurate distributions. Common washing steps used to remove salts, and solvent-based matrix application, allow analyte spreading to occur. Solvent-free matrix applications can reduce this risk, but increase the possibility of ionisation bias due to matrix adhesion to tissue sections. We report here the use of matrix-free MSI using laser desorption ionisation performed on a 12 T Fourier transform ion cyclotron resonance (FTICR) mass spectrometer. We used unprocessed tissue with no post-processing following thaw-mounting on matrix-assisted laser desorption ionisation (MALDI) indium-tin oxide (ITO) target plates. The identification and distribution of a range of phospholipids in mouse brain and kidney sections are presented and compared with previously published MALDI time-of-flight (TOF) MSI distributions.

Typical kernel size and number of sparse random matrices over Galois fields: a statistical physics approach

Using methods of statistical physics, we study the average number and kernel size of general sparse random matrices over GF(q), with a given connectivity profile, in the thermodynamical limit of large matrices. We introduce a mapping of GF(q) matrices onto spin systems using the representation of the cyclic group of order q as the q-th complex roots of unity. This representation facilitates the derivation of the average kernel size of random matrices using the replica approach, under the replica symmetric ansatz, resulting in saddle point equations for general connectivity distributions. Numerical solutions are then obtained for particular cases by population dynamics. Similar techniques also allow us to obtain an expression for the exact and average number of random matrices for any general connectivity profile. We present numerical results for particular distributions.

Sparsely spread CDMA - A statistical mechanics-based analysis

Sparse code division multiple access (CDMA), a variation on the standard CDMA method in which the spreading (signature) matrix contains only a relatively small number of nonzero elements, is presented and analysed using methods of statistical physics. The analysis provides results on the performance of maximum likelihood decoding for sparse spreading codes in the large system limit. We present results for both cases of regular and irregular spreading matrices for the binary additive white Gaussian noise channel (BIAWGN) with a comparison to the canonical (dense) random spreading code. © 2007 IOP Publishing Ltd.

The degradation of gel-spun poly(beta-hydroxybutyrate) fibrous matrix

Poly(β-hydroxybutyrate), (PHB), is a biologically produced, biodegradable thennoplastic with commercial potential. In this work the qualitative and quantitative investigations of the structure and degradation of a previously unstudied, novel, fibrous form of PHB, were completed. This gel-spun PHB fibrous matrix, PHB(FM), which has a similar appearance to cotton wool, possesses a relatively complex structure which combines a large volume with a low mass and has potential for use as a wound scaffolding device. As a result of the intrinsic problems presented by this novel structure, a new experimental procedure was developed to analyze the degradation of the PHB to its monomer hydroxybutyric acid, (HBA). This procedure was used in an accelerated degradation model which accurately monitored the degradation of the undegraded and degraded fractions of a fibrous matrix and the degradation of its PHB component. The in vitro degradation mechanism was also monitored using phase contrast and scanning electron microscopy, differential scanning calorimetry, fibre diameter distributions and Fourier infra-red photoacoustic spectroscopy. The accelerated degradation model was used to predict the degradation of the samples in the physiological model and this provided a clearer picture as to the samples potential biodegradation as medical implantation devices. The degradation of the matrices was characterized by an initial penetration of the degradative medium and weakening of the fibre integrity due to cleavage of the ester linkages, this then led to the physical collapse of the fibres which increased the surface area to volume ratio of the sample and facilitated its degradation. Degradation in the later stages was reduced due to the experimental kinetics, compaction and degradation resistant material, most probably the highly crystalline regions of the PHB. The in vitro degradation of the PHB(FM) was influenced by blending with various polysaccharides, copolymerizing with poly(~-hydroxyvalerate), (PHV), and changes to the manufacturing process. The degradation was also detennined to be faster than that of conventional melt processed PHB based samples. It was concluded that the material factors such as processing, sample size and shape affected the degradation of PHB based samples with the major factor of sample surface area to volume ratio being of paramount importance in determining the degradation of a sample.