A mixture of experts network structure construction algorithm for modelling and control

This paper introduces a new fast, effective and practical model structure construction algorithm for a mixture of experts network system utilising only process data. The algorithm is based on a novel forward constrained regression procedure. Given a full set of the experts as potential model bases, the structure construction algorithm, formed on the forward constrained regression procedure, selects the most significant model base one by one so as to minimise the overall system approximation error at each iteration, while the gate parameters in the mixture of experts network system are accordingly adjusted so as to satisfy the convex constraints required in the derivation of the forward constrained regression procedure. The procedure continues until a proper system model is constructed that utilises some or all of the experts. A pruning algorithm of the consequent mixture of experts network system is also derived to generate an overall parsimonious construction algorithm. Numerical examples are provided to demonstrate the effectiveness of the new algorithms. The mixture of experts network framework can be applied to a wide variety of applications ranging from multiple model controller synthesis to multi-sensor data fusion.

Application of a hybrid tracking algorithm to motion analysis

An algorithm for tracking multiple feature positions in a dynamic image sequence is presented. This is achieved using a combination of two trajectory-based methods, with the resulting hybrid algorithm exhibiting the advantages of both. An optimizing exchange algorithm is described which enables short feature paths to be tracked without prior knowledge of the motion being studied. The resulting partial trajectories are then used to initialize a fast predictor algorithm which is capable of rapidly tracking multiple feature paths. As this predictor algorithm becomes tuned to the feature positions being tracked, it is shown how the location of occluded or poorly detected features can be predicted. The results of applying this tracking algorithm to data obtained from real-world scenes are then presented.

A variational method to retrieve the extinction profile in liquid clouds using multiple field-of-view lidar

Liquid clouds play a profound role in the global radiation budget but it is difficult to remotely retrieve their vertical profile. Ordinary narrow field-of-view (FOV) lidars receive a strong return from such clouds but the information is limited to the first few optical depths. Wideangle multiple-FOV lidars can isolate radiation scattered multiple times before returning to the instrument, often penetrating much deeper into the cloud than the singly-scattered signal. These returns potentially contain information on the vertical profile of extinction coefficient, but are challenging to interpret due to the lack of a fast radiative transfer model for simulating them. This paper describes a variational algorithm that incorporates a fast forward model based on the time-dependent two-stream approximation, and its adjoint. Application of the algorithm to simulated data from a hypothetical airborne three-FOV lidar with a maximum footprint width of 600m suggests that this approach should be able to retrieve the extinction structure down to an optical depth of around 6, and total opticaldepth up to at least 35, depending on the maximum lidar FOV. The convergence behavior of Gauss-Newton and quasi-Newton optimization schemes are compared. We then present results from an application of the algorithm to observations of stratocumulus by the 8-FOV airborne “THOR” lidar. It is demonstrated how the averaging kernel can be used to diagnose the effective vertical resolution of the retrieved profile, and therefore the depth to which information on the vertical structure can be recovered. This work enables exploitation of returns from spaceborne lidar and radar subject to multiple scattering more rigorously than previously possible.

Trade-offs and responsiveness of the single-layer urban canopy parametrization in WRF: an offline evaluation using the MOSCEM optimization algorithm and field observations

For an increasing number of applications, mesoscale modelling systems now aim to better represent urban areas. The complexity of processes resolved by urban parametrization schemes varies with the application. The concept of fitness-for-purpose is therefore critical for both the choice of parametrizations and the way in which the scheme should be evaluated. A systematic and objective model response analysis procedure (Multiobjective Shuffled Complex Evolution Metropolis (MOSCEM) algorithm) is used to assess the fitness of the single-layer urban canopy parametrization implemented in the Weather Research and Forecasting (WRF) model. The scheme is evaluated regarding its ability to simulate observed surface energy fluxes and the sensitivity to input parameters. Recent amendments are described, focussing on features which improve its applicability to numerical weather prediction, such as a reduced and physically more meaningful list of input parameters. The study shows a high sensitivity of the scheme to parameters characterizing roof properties in contrast to a low response to road-related ones. Problems in partitioning of energy between turbulent sensible and latent heat fluxes are also emphasized. Some initial guidelines to prioritize efforts to obtain urban land-cover class characteristics in WRF are provided. Copyright © 2010 Royal Meteorological Society and Crown Copyright.

FAMOUS, faster: using parallel computing techniques to accelerate the FAMOUS/HadCM3 climate model with a focus on the radiative transfer algorithm

We have optimised the atmospheric radiation algorithm of the FAMOUS climate model on several hardware platforms. The optimisation involved translating the Fortran code to C and restructuring the algorithm around the computation of a single air column. Instead of the existing MPI-based domain decomposition, we used a task queue and a thread pool to schedule the computation of individual columns on the available processors. Finally, four air columns are packed together in a single data structure and computed simultaneously using Single Instruction Multiple Data operations. The modified algorithm runs more than 50 times faster on the CELL’s Synergistic Processing Elements than on its main PowerPC processing element. On Intel-compatible processors, the new radiation code runs 4 times faster. On the tested graphics processor, using OpenCL, we find a speed-up of more than 2.5 times as compared to the original code on the main CPU. Because the radiation code takes more than 60% of the total CPU time, FAMOUS executes more than twice as fast. Our version of the algorithm returns bit-wise identical results, which demonstrates the robustness of our approach. We estimate that this project required around two and a half man-years of work.

Solar wind control of magnetospheric energy content: substorm quenching and multiple onsets

In this paper we report coordinated multispacecraft and ground-based observations of a double substorm onset close to Scandinavia on November 17, 1996. The Wind and the Geotail spacecraft, which were located in the solar wind and the subsolar magnetosheath, respectively, recorded two periods of southward directed interplanetary magnetic field (IMF). These periods were separated by a short northward IMF excursion associated with a solar wind pressure pulse, which compressed the magnetosphere to such a degree that Geotail for a short period was located outside the bow shock. The first period of southward IMF initiated a substorm growth. phase, which was clearly detected by an array of ground-based instrumentation and by Interball in the northern tail lobe. A first substorm onset occurred in close relation to the solar wind pressure pulse impinging on the magnetopause and almost simultaneously with the northward turning of the IMF. However, this substorm did not fully develop. In clear association with the expansion of the magnetosphere at the end of the pressure pulse, the auroral expansion was stopped, and the northern sky cleared. We will present evidence that the change in the solar wind dynamic pressure actively quenched the energy available for any further substorm expansion. Directly after this period, the magnetometer network detected signatures of a renewed substorm growth phase, which was initiated by the second southward turning of the IMF and which finally lead to a second, and this time complete, substorm intensification. We have used our multipoint observations in order to understand the solar wind control of the substorm onset and substorm quenching. The relative timings between the observations on the various satellites and on the ground were used to infer a possible causal relationship between the solar wind pressure variations and consequent substorm development. Furthermore, using a relatively simple algorithm to model the tail lobe field and the total tail flux, we show that there indeed exists a close relationship between the relaxation of a solar wind pressure pulse, the reduction of the tail lobe field, and the quenching of the initial substorm.

Joint multiple dictionary learning for tensor sparse coding

Traditional dictionary learning algorithms are used for finding a sparse representation on high dimensional data by transforming samples into a one-dimensional (1D) vector. This 1D model loses the inherent spatial structure property of data. An alternative solution is to employ Tensor Decomposition for dictionary learning on their original structural form —a tensor— by learning multiple dictionaries along each mode and the corresponding sparse representation in respect to the Kronecker product of these dictionaries. To learn tensor dictionaries along each mode, all the existing methods update each dictionary iteratively in an alternating manner. Because atoms from each mode dictionary jointly make contributions to the sparsity of tensor, existing works ignore atoms correlations between different mode dictionaries by treating each mode dictionary independently. In this paper, we propose a joint multiple dictionary learning method for tensor sparse coding, which explores atom correlations for sparse representation and updates multiple atoms from each mode dictionary simultaneously. In this algorithm, the Frequent-Pattern Tree (FP-tree) mining algorithm is employed to exploit frequent atom patterns in the sparse representation. Inspired by the idea of K-SVD, we develop a new dictionary update method that jointly updates elements in each pattern. Experimental results demonstrate our method outperforms other tensor based dictionary learning algorithms.

A new adaptive multiple modelling approach for non-linear and non-stationary systems

This paper proposes a novel adaptive multiple modelling algorithm for non-linear and non-stationary systems. This simple modelling paradigm comprises K candidate sub-models which are all linear. With data available in an online fashion, the performance of all candidate sub-models are monitored based on the most recent data window, and M best sub-models are selected from the K candidates. The weight coefficients of the selected sub-model are adapted via the recursive least square (RLS) algorithm, while the coefficients of the remaining sub-models are unchanged. These M model predictions are then optimally combined to produce the multi-model output. We propose to minimise the mean square error based on a recent data window, and apply the sum to one constraint to the combination parameters, leading to a closed-form solution, so that maximal computational efficiency can be achieved. In addition, at each time step, the model prediction is chosen from either the resultant multiple model or the best sub-model, whichever is the best. Simulation results are given in comparison with some typical alternatives, including the linear RLS algorithm and a number of online non-linear approaches, in terms of modelling performance and time consumption.

Using non-homogeneous Poisson models with multiple change-points to estimate the number of ozone exceedances in Mexico City

In this paper, we consider some non-homogeneous Poisson models to estimate the probability that an air quality standard is exceeded a given number of times in a time interval of interest. We assume that the number of exceedances occurs according to a non-homogeneous Poisson process (NHPP). This Poisson process has rate function lambda(t), t >= 0, which depends on some parameters that must be estimated. We take into account two cases of rate functions: the Weibull and the Goel-Okumoto. We consider models with and without change-points. When the presence of change-points is assumed, we may have the presence of either one, two or three change-points, depending of the data set. The parameters of the rate functions are estimated using a Gibbs sampling algorithm. Results are applied to ozone data provided by the Mexico City monitoring network. In a first instance, we assume that there are no change-points present. Depending on the adjustment of the model, we assume the presence of either one, two or three change-points. Copyright (C) 2009 John Wiley & Sons, Ltd.

FAST, PARALLEL AND SECURE CRYPTOGRAPHY ALGORITHM USING LORENZ`S ATTRACTOR

A novel cryptography method based on the Lorenz`s attractor chaotic system is presented. The proposed algorithm is secure and fast, making it practical for general use. We introduce the chaotic operation mode, which provides an interaction among the password, message and a chaotic system. It ensures that the algorithm yields a secure codification, even if the nature of the chaotic system is known. The algorithm has been implemented in two versions: one sequential and slow and the other, parallel and fast. Our algorithm assures the integrity of the ciphertext (we know if it has been altered, which is not assured by traditional algorithms) and consequently its authenticity. Numerical experiments are presented, discussed and show the behavior of the method in terms of security and performance. The fast version of the algorithm has a performance comparable to AES, a popular cryptography program used commercially nowadays, but it is more secure, which makes it immediately suitable for general purpose cryptography applications. An internet page has been set up, which enables the readers to test the algorithm and also to try to break into the cipher.

Multiple camera people detection and tracking using support integration

This paper proposes a method to locate and track people by combining evidence from multiple cameras using the homography constraint. The proposed method use foreground pixels from simple background subtraction to compute evidence of the location of people on a reference ground plane. The algorithm computes the amount of support that basically corresponds to the ""foreground mass"" above each pixel. Therefore, pixels that correspond to ground points have more support. The support is normalized to compensate for perspective effects and accumulated on the reference plane for all camera views. The detection of people on the reference plane becomes a search for regions of local maxima in the accumulator. Many false positives are filtered by checking the visibility consistency of the detected candidates against all camera views. The remaining candidates are tracked using Kalman filters and appearance models. Experimental results using challenging data from PETS`06 show good performance of the method in the presence of severe occlusion. Ground truth data also confirms the robustness of the method. (C) 2010 Elsevier B.V. All rights reserved.

A multi-GPU algorithm for large-scale neuronal networks

Large-scale simulations of parts of the brain using detailed neuronal models to improve our understanding of brain functions are becoming a reality with the usage of supercomputers and large clusters. However, the high acquisition and maintenance cost of these computers, including the physical space, air conditioning, and electrical power, limits the number of simulations of this kind that scientists can perform. Modern commodity graphical cards, based on the CUDA platform, contain graphical processing units (GPUs) composed of hundreds of processors that can simultaneously execute thousands of threads and thus constitute a low-cost solution for many high-performance computing applications. In this work, we present a CUDA algorithm that enables the execution, on multiple GPUs, of simulations of large-scale networks composed of biologically realistic Hodgkin-Huxley neurons. The algorithm represents each neuron as a CUDA thread, which solves the set of coupled differential equations that model each neuron. Communication among neurons located in different GPUs is coordinated by the CPU. We obtained speedups of 40 for the simulation of 200k neurons that received random external input and speedups of 9 for a network with 200k neurons and 20M neuronal connections, in a single computer with two graphic boards with two GPUs each, when compared with a modern quad-core CPU. Copyright (C) 2010 John Wiley & Sons, Ltd.

Pharmacogenetics of Warfarin: Development of a Dosing Algorithm for Brazilian Patients

A dosing algorithm including genetic (VKORC1 and CYP2C9 genotypes) and nongenetic factors (age, weight, therapeutic indication, and cotreatment with amiodarone or simvastatin) explained 51% of the variance in stable weekly warfarin doses in 390 patients attending an anticoagulant clinic in a Brazilian public hospital. The VKORC1 3673G>A genotype was the most important predictor of warfarin dose, with a partial R(2) value of 23.9%. Replacing the VKORC1 3673G>A genotype with VKORC1 diplotype did not increase the algorithm`s predictive power. We suggest that three other single-nucleotide polymorphisms (SNPs) (5808T>G, 6853G>C, and 9041G>A) that are in strong linkage disequilibrium (LD) with 3673G>A would be equally good predictors of the warfarin dose requirement. The algorithm`s predictive power was similar across the self-identified ""race/color"" subsets. ""Race/color"" was not associated with stable warfarin dose in the multiple regression model, although the required warfarin dose was significantly lower (P = 0.006) in white (29 +/- 13 mg/week, n = 196) than in black patients (35 +/- 15 mg/week, n = 76).

Solving the Vehicle Routing Problem with Genetic ALgorithm and Simulated Annealing

This Thesis Work will concentrate on a very interesting problem, the Vehicle Routing Problem (VRP). In this problem, customers or cities have to be visited and packages have to be transported to each of them, starting from a basis point on the map. The goal is to solve the transportation problem, to be able to deliver the packages-on time for the customers,-enough package for each Customer,-using the available resources- and – of course - to be so effective as it is possible.Although this problem seems to be very easy to solve with a small number of cities or customers, it is not. In this problem the algorithm have to face with several constraints, for example opening hours, package delivery times, truck capacities, etc. This makes this problem a so called Multi Constraint Optimization Problem (MCOP). What’s more, this problem is intractable with current amount of computational power which is available for most of us. As the number of customers grow, the calculations to be done grows exponential fast, because all constraints have to be solved for each customers and it should not be forgotten that the goal is to find a solution, what is best enough, before the time for the calculation is up. This problem is introduced in the first chapter: form its basics, the Traveling Salesman Problem, using some theoretical and mathematical background it is shown, why is it so hard to optimize this problem, and although it is so hard, and there is no best algorithm known for huge number of customers, why is it a worth to deal with it. Just think about a huge transportation company with ten thousands of trucks, millions of customers: how much money could be saved if we would know the optimal path for all our packages.Although there is no best algorithm is known for this kind of optimization problems, we are trying to give an acceptable solution for it in the second and third chapter, where two algorithms are described: the Genetic Algorithm and the Simulated Annealing. Both of them are based on obtaining the processes of nature and material science. These algorithms will hardly ever be able to find the best solution for the problem, but they are able to give a very good solution in special cases within acceptable calculation time.In these chapters (2nd and 3rd) the Genetic Algorithm and Simulated Annealing is described in details, from their basis in the “real world” through their terminology and finally the basic implementation of them. The work will put a stress on the limits of these algorithms, their advantages and disadvantages, and also the comparison of them to each other.Finally, after all of these theories are shown, a simulation will be executed on an artificial environment of the VRP, with both Simulated Annealing and Genetic Algorithm. They will both solve the same problem in the same environment and are going to be compared to each other. The environment and the implementation are also described here, so as the test results obtained.Finally the possible improvements of these algorithms are discussed, and the work will try to answer the “big” question, “Which algorithm is better?”, if this question even exists.

Optimization Of Groundwater Remediation With New Efficient Parallel Algorithm For Global Optimization

Application of optimization algorithm to PDE modeling groundwater remediation can greatly reduce remediation cost. However, groundwater remediation analysis requires a computational expensive simulation, therefore, effective parallel optimization could potentially greatly reduce computational expense. The optimization algorithm used in this research is Parallel Stochastic radial basis function. This is designed for global optimization of computationally expensive functions with multiple local optima and it does not require derivatives. In each iteration of the algorithm, an RBF is updated based on all the evaluated points in order to approximate expensive function. Then the new RBF surface is used to generate the next set of points, which will be distributed to multiple processors for evaluation. The criteria of selection of next function evaluation points are estimated function value and distance from all the points known. Algorithms created for serial computing are not necessarily efficient in parallel so Parallel Stochastic RBF is different algorithm from its serial ancestor. The application for two Groundwater Superfund Remediation sites, Umatilla Chemical Depot, and Former Blaine Naval Ammunition Depot. In the study, the formulation adopted treats pumping rates as decision variables in order to remove plume of contaminated groundwater. Groundwater flow and contamination transport is simulated with MODFLOW-MT3DMS. For both problems, computation takes a large amount of CPU time, especially for Blaine problem, which requires nearly fifty minutes for a simulation for a single set of decision variables. Thus, efficient algorithm and powerful computing resource are essential in both cases. The results are discussed in terms of parallel computing metrics i.e. speedup and efficiency. We find that with use of up to 24 parallel processors, the results of the parallel Stochastic RBF algorithm are excellent with speed up efficiencies close to or exceeding 100%.