Certain models from uncertain data: The algebraic case

The problem of deriving possible linear relations from data affected by additive noise has received remarkable attention in recent years particularly regarding the assumptions (‘prejudices’) behind the procedures leading to unique models. Unlike traditional approaches leading to unique solutions (least squares, maximum likelihood, etc.) the Frisch scheme, belonging to the family of Errors-in-Variables (EV) schemes, leads to a whole family of models compatible with a set of noisy data and is considered as mildly affected by prejudices. This paper shows how also under the assumptions of the Frisch scheme it is possible to obtain a unique model from uncertain data and also to derive the actual amount of noise when the noiseless data are linked by a single linear relation. The more general EV case of non-independent additive noises is then considered and it is shown how also under these assumptions it is possible to obtain the unique set of parameters linking the noiseless data and how the family of compatible noise covariance matrices is defined, in this case, by the infinite elements of a linear variety.     

Real-time retrieval of Leaf Area Index from MODIS time series data

Real-time/near real-time inversion of land surface biogeophysical variables from satellite observations is required to monitor rapid land surface changes, and provide the necessary input for numerical weather forecasting models and decision support systems. This paper develops a new inversion method for the real-time estimation of the Leaf Area Index (LAI) of land surfaces from MODIS time series reflectance data (MOD09A1). It consists of a series of procedures, including time series data smoothing, data quality control and real-time estimation of LAI. After the historical LAI time series is smoothed by a multi-step Savitzky-Golay filter to determine the upper LAI envelope, a Seasonal Auto-Regressive Integrated Moving Average (SARIMA) model is used to derive the LAI climatology. Based on the climatology from the SARIMA model to evolve LAI in time, a dynamic model is then constructed and used to provide the short-range forecast of LAI. Predictions from this model are used with Ensemble Kalman Filter (EnKF) techniques to recursively update biophysical variables as new observations arrive. The validation results produced using MODIS surface reflectance data and field-measured LAI data at eight BELMANIP sites show that the real-time inversion method is able to efficiently produce a relatively smooth LAI product. In addition, the accuracy is significantly improved over the MODIS LAI product.     

A parallel evolutionary algorithm to optimize dynamic data types in embedded systems

New multimedia embedded applications are increasingly dynamic, and rely on dynamically-allocated data types (DDTs) to store their data. The optimization of DDTs for each target embedded system is a time-consuming process due to the large searching space of possible DDTs implementations. That implies the minimization of embedded design variables (memory accesses, power consumption and memory usage). Up to know, some very effective heuristic algorithms have been developed in order to solve this problem, but it is unknown how good the selected DDTs are since the problem is NP-complete and cannot be fully explored. In these cases the use of parallel processing can be very useful because it allows not only to explore more solutions spending the same time, but also to implement new algorithms. This paper describes several parallel evolutionary algorithms for DDTs optimization in Embedded Systems, where parallelism improves the solutions found by the corresponding sequential algorithm, which indeed is quite effective compared with other previously proposed procedures. Experimental results show how a novel parallel multi-objective genetic algorithm, which combines NSGA-II and SPEA2, allows designers to reach a larger number of solutions than previous approximations.     

Efficient implementation of image interpolation as an inverse problem

This paper presents three computationally efficient solutions for the image interpolation problem which are developed in a general framework. This framework is based on dealing with the problem as an inverse problem. Based on the observation model, our objective is to obtain a high resolution image which is as close as possible to the original high resolution image subject to certain constraints. In the first solution, a linear minimum mean square error (LMMSE) approach is suggested. The necessary assumptions required to reduce the computational complexity of the LMMSE solution are presented. The sensitivity of the LMMSE solution to these assumptions is studied. In the second solution, the concept of entropy maximization of the required high resolution image a priori is used. The implementation of the suggested maximum entropy solution as a single sparse matrix inversion is presented. Finally, the well-known regularization technique used in iterative nature in image interpolation and image restoration is revisited. An efficient sectioned implementation of regularized image interpolation, which avoids the large number of iterations required in the interactive technique, is presented. In our suggested regularized solution, the computational time is linearly proportional to the dimensions of the image to be interpolated and a single matrix inversion of moderate dimensions is required. This property allows its implementation in interpolating images of any dimensions which is a great problem in iterative techniques. The effect of the choice of the regularization parameter on the suggested regularized image interpolation solution is studied. The performance of all the above-mentioned solutions is compared to traditional polynomial based interpolation techniques such as cubic O-MOMS and to iterative interpolation as well. The suitability of each solution to interpolating different images is also studied.     

Nonlinear inversion of potential-field data using a hybrid-encoding genetic algorithm

Using a genetic algorithm to solve an inverse problem of complex nonlinear geophysical equations is advantageous because it does not require computer gradients of models or “good” initial models. The multi-point search of a genetic algorithm makes it easier to find the globally optimal solution while avoiding falling into a local extremum. As is the case in other optimization approaches, the search efficiency for a genetic algorithm is vital in finding desired solutions successfully in a multi-dimensional model space. A binary-encoding genetic algorithm is hardly ever used to resolve an optimization problem such as a simple geophysical inversion with only three unknowns. The encoding mechanism, genetic operators, and population size of the genetic algorithm greatly affect search processes in the evolution. It is clear that improved operators and proper population size promote the convergence. Nevertheless, not all genetic operations perform perfectly while searching under either a uniform binary or a decimal encoding system. With the binary encoding mechanism, the crossover scheme may produce more new individuals than with the decimal encoding. On the other hand, the mutation scheme in a decimal encoding system will create new genes larger in scope than those in the binary encoding. This paper discusses approaches of exploiting the search potential of genetic operations in the two encoding systems and presents an approach with a hybrid-encoding mechanism, multi-point crossover, and dynamic population size for geophysical inversion. We present a method that is based on the routine in which the mutation operation is conducted in the decimal code and multi-point crossover operation in the binary code. The mix-encoding algorithm is called the hybrid-encoding genetic algorithm (HEGA). HEGA provides better genes with a higher probability by a mutation operator and improves genetic algorithms in resolving complicated geophysical inverse problems. Another significant result is that final solution is determined by the average model derived from multiple trials instead of one computation due to the randomness in a genetic algorithm procedure. These advantages were demonstrated by synthetic and real-world examples of inversion of potential-field data.     

3D porosity prediction from seismic inversion and neural networks

In this work, we address the problem of transforming seismic reflection data into an intrinsic rock property model. Specifically, we present an application of a methodology that allows interpreters to obtain effective porosity 3D maps from post-stack 3D seismic amplitude data, using measured density and sonic well log data as constraints. In this methodology, a 3D acoustic impedance model is calculated from seismic reflection amplitudes by applying an L₁-norm sparse-spike inversion algorithm in the time domain, followed by a recursive inversion performed in the frequency domain. A 3D low-frequency impedance model is estimated by kriging interpolation of impedance values calculated from well log data. This low-frequency model is added to the inversion result which otherwise provides only a relative numerical scale. To convert acoustic impedance into a single reservoir property, a feed-forward Neural Network (NN) is trained, validated and tested using gamma-ray and acoustic impedance values observed at the well log positions as input and effective porosity values as target. The trained NN is then applied for the whole reservoir volume in order to obtain a 3D effective porosity model. While the particular conclusions drawn from the results obtained in this work cannot be generalized, such results suggest that this workflow can be applied successfully as an aid in reservoir characterization, especially when there is a strong non-linear relationship between effective porosity and acoustic impedance.     

Mapping of Sahelian vegetation parameters from ERS scatterometer data with an evolution strategies algorithm

The West African Sahel rainfall regime is known for its spatio-temporal variability at different scales which has a strong impact on vegetation development. This study presents results of the combined use of a simple water balance model, a radiative transfer model and ERS scatterometer data to produce map of vegetation biomass and thus vegetation cover at a spatial resolution of 25 km. The backscattering coefficient measured by spaceborne wind scatterometers over Sahel shows a marked seasonality linked to the drastic changes of both soil and vegetation dielectric properties associated to the alternating dry and wet seasons. For lack of a direct observation, METEOSAT rainfall estimates are used to calculate temporal series of soil moisture with the help of a water balance model. This a priori information is used as input of the radiative transfer model that simulates the interaction between the radar wave and the surface components (soil and vegetation). Then, an inversion algorithm is applied to retrieve vegetation aerial mass from the ERS scatterometer data. Because of the nonlinear feature of the inverse problem to be solved, the inversion is performed using a global stochastic nonlinear inversion method. A good agreement is obtained between the inverse solutions and independent field measurements with mean and standard deviation of −54 and 130 kg of dry matter by hectare (kg DM/ha), respectively. The algorithm is then applied to a 350,000 km² area including the Malian Gourma and Seno region and a Sahelian part of Burkina Faso during two contrasted seasons (1999 and 2000). At the considered resolution, the obtained herbaceous mass maps show a global qualitative consistency (r²=0.71) with NDVI images acquired by the VEGETATION instrument.     

Weighted least squares set estimation from l∞-normbounded noise data

The problem of parameter set estimation from pointwise bounded-error data is considered. The possibilities of employing l₂ -projection procedures to solve the problem are explored, and exact as well as approximate outer-bounding solutions are proposed. In particular, the properties of weighted least squares set estimation in this l_∞ norm bounded-error context and the implementation of a resulting minimum-volume parallelotope-bounding algorithm are discussed     

Discretized learning automata solutions to the capacity assignment problem for prioritized networks

We present a discretized learning automaton (LA) solution to the capacity assignment (CA) problem which focuses on finding the best possible set of capacities for the links that satisfy the traffic requirements in a prioritized network while minimizing the cost. Most approaches consider a single class of packets flowing through the network, but in reality, different classes of packets with different average packet lengths and different priorities are transmitted over the networks. This generalized model is the focus of this paper. Although the problem is inherently NP-hard, a few approximate solutions have been proposed in the literature. Marayuma and Tang (1977) proposed a single algorithm composed of several elementary heuristic procedures. Other solutions tackle the problem by using modern-day artificial intelligence (AI) paradigms such as simulated annealing and genetic algorithms (GAs). In 2000, we introduced a new method, superior to these, that uses continuous LA. In this paper, we present a discretized LA solution to the problem. This solution uses a meta-action philosophy new to the field of LA, and is probably the best available solution to this extremely complex problem.     

The genetic algorithm applied to haplotype data at the LDL receptor locus

Conventional statistical methods based upon single restriction fragment length polymorphisms often prove inadequate in studies of genetic variation. Cladistic analysis has been suggested as an alternative, but requires basic assumptions that usually cannot be met. We wanted to test whether it could be a workable approach to apply the genetic algorithm, an artificial intelligence method, to haplotype data. The genetic algorithm creates in-computer artificial 'individuals', all having 'genes' coding for solutions to a problem. The individuals are allowed to compete and 'mate', individuals with genes coding for better solutions mating more often. Genes coding for good solutions survive through generations of the genetic algorithm. At the end of the run, the best solutions can be extracted. We applied the genetic algorithm to data consisting of cholesterol values and haplotypes made up of seven restriction sites at the LDL receptor locus. The persons included were 114 FH (familial hypercholesterolemia) patients and 61 normals. The genetic algorithm found the restriction sites 1 (Sph1 in intron 6), 2 (StuI in exon 8), and 7 (ApaLI site in the 3' flanking region) were associated with high cholesterol levels. As a validity check we used runs of the genetic algorithm applied to 'artificial patients', i.e. artificially generated haplotypes linked to artificially generated cholesterol values. This demonstrated the genetic algorithm consistently found the appropriate haplotype. We conclude that the genetic algorithm may be a useful tool for studying genetic variation.     

A high-performance application data environment for large-scale scientific computations

Effective high-level data management is becoming an important issue with more and more scientific applications manipulating huge amounts of secondary-storage and tertiary-storage data using parallel processors. A major problem facing the current solutions to this data management problem is that these solutions either require a deep understanding of specific data storage architectures and file layouts to obtain the best performance (as in high-performance storage management systems and parallel file systems), or they sacrifice significant performance in exchange for ease-of-use and portability (as in traditional database management systems). We discuss the design, implementation, and evaluation of a novel application development environment for scientific computations. This environment includes a number of components that make it easy for the programmers to code and run their applications without much programming effort and, at the same time, to harness the available computational and storage power on parallel architectures.     

LAI retrieval from multiangular image classification and inversion of a ray tracing model

A non-conventional approach for the estimation of leaf area index (LAI) and leaf angle distribution (LAD), based on the use of information contained in multiangular images and the inversion of a canopy ray tracing model, is proposed in this work. As an alternative to the use of overall image reflectance data, the image fraction components, i.e. sunlit and shaded leaves and soil, are obtained by supervised classification of ground-based multiangular images acquired using an inexpensive colour infrared camera, the Dycam ADC. These data are used for the inversion of a numerical model of a vegetation canopy in which the latter is described as composed of randomly distributed disks (leaves). The model was developed using the POV-ray ray tracer. Model inversion is carried out using the look-up-table approach. The proposed methodology was tested using an extensive data set gathered on the potato crop during experimental trials carried out at Viterbo (Italy) for 3 years. The results show that LAI was successfully estimated with a RMSE varying from 0.29 to 0.75 in the different years. The potential sources of error in both estimated and measured LAI values are extensively discussed.     

Robustness-based design optimization under data uncertainty

This paper proposes formulations and algorithms for design optimization under both aleatory (i.e., natural or physical variability) and epistemic uncertainty (i.e., imprecise probabilistic information), from the perspective of system robustness. The proposed formulations deal with epistemic uncertainty arising from both sparse and interval data without any assumption about the probability distributions of the random variables. A decoupled approach is proposed in this paper to un-nest the robustness-based design from the analysis of non-design epistemic variables to achieve computational efficiency. The proposed methods are illustrated for the upper stage design problem of a two-stage-to-orbit (TSTO) vehicle, where the information on the random design inputs are only available as sparse point data and/or interval data. As collecting more data reduces uncertainty but increases cost, the effect of sample size on the optimality and robustness of the solution is also studied. A method is developed to determine the optimal sample size for sparse point data that leads to the solutions of the design problem that are least sensitive to variations in the input random variables.     

Optimizing thermal design of data center cabinets with a new multi-objective genetic algorithm

There is an ever increasing need to use optimization methods for thermal design of data centers and the hardware populating them. Airflow simulations of cabinets and data centers are computationally intensive and this problem is exacerbated when the simulation model is integrated with a design optimization method. Generally speaking, thermal design of data center hardware can be posed as a constrained multi-objective optimization problem. A popular approach for solving this kind of problem is to use Multi-Objective Genetic Algorithms (MOGAs). However, the large number of simulation evaluations needed for MOGAs has been preventing their applications to realistic engineering design problems. In this paper, details of a substantially more efficient MOGA are formulated and demonstrated through a thermal analysis simulation model of a data center cabinet. First, a reduced-order model of the cabinet problem is constructed using the Proper Orthogonal Decomposition (POD). The POD model is then used to form the objective and constraint functions of an optimization model. Next, this optimization model is integrated with the new MOGA. The new MOGA uses a “kriging” guided operation in addition to conventional genetic algorithm operations to search the design space for global optimal design solutions. This approach for optimal design is essential to handle complex multi-objective situations, where the optimal solutions may be non-obvious from simple analyses or intuition. It is shown that in optimizing the data center cabinet problem, the new MOGA outperforms a conventional MOGA by estimating the Pareto front using 50% fewer simulation calls, which makes its use very promising for complex thermal design problems. Recommended by: Monem Beitelmal     

Exploring process data

With the growth of computer usage at all levels in the process industries, the volume of available data has also grown enormously, sometimes to levels that render analysis difficult. Most of this data may be characterized as historical in the sense that it was not collected on the basis of experiments designed to test specific statistical hypotheses. Consequently, the resulting datasets are likely to contain unexpected features (e.g. outliers from various sources, unsuspected correlations between variables, etc.). This observation is important for two reasons: first, these data anomalies can completely negate the results obtained by standard analysis procedures, particularly those based on squared error criteria (a large class that includes many SPC and chemometrics techniques). Secondly and sometimes more importantly, an understanding of these data anomalies may lead to extremely valuable insights. For both of these reasons, it is important to approach the analysis of large historical datasets with the initial objective of uncovering and understanding their gross structure and character. This paper presents a brief survey of some simple procedures that have been found to be particularly useful at this preliminary stage of analysis.     

GRAFT, a complete system for data fusion

Data fusion concerns the problem of merging information coming from independent sources. Also known as statistical matching, file grafting or microdata merging, it is a challenging problem for statisticians. The increasing growth of collected data makes combining different sources of information an attractive alternative to single source data. The interest in data fusion derives, in certain cases, from the impossibility of attaining specific information from one source of data and the reduction of the cost entailed by this operation and, in all cases, from taking greater advantage of the available collected information. The GRAFT system is presented. It is a multipurpose data fusion system based on the k-nearest neighbor (k-nn) hot deck imputation method. The system aim is to cope with many data fusion problems and domains. The k-nn is a very demanding algorithm. The solutions envisaged and their cost, which allow this methodology to be used in a wide range of real problems, are presented.     

Phishing detection based Associative Classification data mining

Website phishing is considered one of the crucial security challenges for the online community due to the massive numbers of online transactions performed on a daily basis. Website phishing can be described as mimicking a trusted website to obtain sensitive information from online users such as usernames and passwords. Black lists, white lists and the utilisation of search methods are examples of solutions to minimise the risk of this problem. One intelligent approach based on data mining called Associative Classification (AC) seems a potential solution that may effectively detect phishing websites with high accuracy. According to experimental studies, AC often extracts classifiers containing simple “If-Then” rules with a high degree of predictive accuracy. In this paper, we investigate the problem of website phishing using a developed AC method called Multi-label Classifier based Associative Classification (MCAC) to seek its applicability to the phishing problem. We also want to identify features that distinguish phishing websites from legitimate ones. In addition, we survey intelligent approaches used to handle the phishing problem. Experimental results using real data collected from different sources show that AC particularly MCAC detects phishing websites with higher accuracy than other intelligent algorithms. Further, MCAC generates new hidden knowledge (rules) that other algorithms are unable to find and this has improved its classifiers predictive performance.     

On the sizing of a solar thermal electricity plant for multiple objectives using evolutionary optimization

Design, implementation and operation of solar thermal electricity plants are no more an academic task, rather they have become a necessity. In this paper, we work with power industries to formulate a multi-objective optimization model and attempt to solve the resulting problem using classical as well as evolutionary optimization techniques. On a set of four objectives having complex trade-offs, our proposed procedure first finds a set of trade-off solutions showing the entire range of optimal solutions. Thereafter, the evolutionary optimization procedure is combined with a multiple criterion decision making (MCDM) approach to focus on preferred regions of the trade-off frontier. Obtained solutions are compared with a classical generating method. Eventually, a decision-maker is involved in the process and a single preferred solution is obtained in a systematic manner. Starting with generating a wide spectrum of trade-off solutions to have a global understanding of feasible solutions, then concentrating on specific preferred regions for having a more detailed understanding of preferred solutions, and then zeroing on a single preferred solution with the help of a decision-maker demonstrates the use of multi-objective optimization and decision making methodologies in practice. As a by-product, useful properties among decision variables that are common to the obtained solutions are gathered as vital knowledge for the problem. The procedures used in this paper are ready to be used to other similar real-world problem solving tasks.     

基于改进粒子群优化的无标记数据鲁棒聚类算法

已有的聚类算法大多仅考虑单一的目标,导致对某些形状的数据集性能较弱,对此提出一种基于改进粒子群优化的无标记数据鲁棒聚类算法。优化阶段:首先,采用多目标粒子群优化的经典形式生成聚类解集合;然后,使用K-means算法生成随机分布的初始化种群,并为其分配随机初始化的速度;最终,采用MaxiMin策略确定帕累托最优解。决策阶段：测量帕累托解集与理想解的距离,将距离最短的帕累托解作为最终聚类解。对比实验结果表明,本算法对不同形状的数据集均可获得较优的类簇数量,对目标问题的复杂度具有较好的鲁棒性。