Simultaneous ant colony optimisation algorithms for learning linguistic fuzzy rules

M. Galea and Q. Shen. Simultaneous ant colony optimisation algorithms for learning linguistic fuzzy rules. A. Abraham, C. Grosan and V. Ramos (Eds.), Swarm Intelligence in Data Mining, pages 75-99.

A new interatomic potential for nanoscale silica

Flikkema, Edwin; Bromley, S.T., (2003) 'A new interatomic potential for nanoscale silica', Chemical Physics Letters 378(5-6) pp.622-629 RAE2008

Modele fizyczne planetoid pojedynczych i podwójnych otrzymane na podstawie przeglądów fotometrycznych: przygotowanie do naukowego wykorzystania danych z kosmicznej misji Gaia

Wydział Fizyki

A Neural Network Architecture for Autonomous Learning, Recognition, and Prediction in a Nonstationary World

In a constantly changing world, humans are adapted to alternate routinely between attending to familiar objects and testing hypotheses about novel ones. We can rapidly learn to recognize and narne novel objects without unselectively disrupting our memories of familiar ones. We can notice fine details that differentiate nearly identical objects and generalize across broad classes of dissimilar objects. This chapter describes a class of self-organizing neural network architectures--called ARTMAP-- that are capable of fast, yet stable, on-line recognition learning, hypothesis testing, and naming in response to an arbitrary stream of input patterns (Carpenter, Grossberg, Markuzon, Reynolds, and Rosen, 1992; Carpenter, Grossberg, and Reynolds, 1991). The intrinsic stability of ARTMAP allows the system to learn incrementally for an unlimited period of time. System stability properties can be traced to the structure of its learned memories, which encode clusters of attended features into its recognition categories, rather than slow averages of category inputs. The level of detail in the learned attentional focus is determined moment-by-moment, depending on predictive success: an error due to over-generalization automatically focuses attention on additional input details enough of which are learned in a new recognition category so that the predictive error will not be repeated. An ARTMAP system creates an evolving map between a variable number of learned categories that compress one feature space (e.g., visual features) to learned categories of another feature space (e.g., auditory features). Input vectors can be either binary or analog. Computational properties of the networks enable them to perform significantly better in benchmark studies than alternative machine learning, genetic algorithm, or neural network models. Some of the critical problems that challenge and constrain any such autonomous learning system will next be illustrated. Design principles that work together to solve these problems are then outlined. These principles are realized in the ARTMAP architecture, which is specified as an algorithm. Finally, ARTMAP dynamics are illustrated by means of a series of benchmark simulations.

Fuzzy ARTMAP: A Neural Network Architecture for Incremental Supervised Learning of Analog Multidimensional Maps

A new neural network architecture is introduced for incremental supervised learning of recognition categories and multidimensional maps in response to arbitrary sequences of analog or binary input vectors. The architecture, called Fuzzy ARTMAP, achieves a synthesis of fuzzy logic and Adaptive Resonance Theory (ART) neural networks by exploiting a close formal similarity between the computations of fuzzy subsethood and ART category choice, resonance, and learning. Fuzzy ARTMAP also realizes a new Minimax Learning Rule that conjointly minimizes predictive error and maximizes code compression, or generalization. This is achieved by a match tracking process that increases the ART vigilance parameter by the minimum amount needed to correct a predictive error. As a result, the system automatically learns a minimal number of recognition categories, or "hidden units", to met accuracy criteria. Category proliferation is prevented by normalizing input vectors at a preprocessing stage. A normalization procedure called complement coding leads to a symmetric theory in which the MIN operator (Λ) and the MAX operator (v) of fuzzy logic play complementary roles. Complement coding uses on-cells and off-cells to represent the input pattern, and preserves individual feature amplitudes while normalizing the total on-cell/off-cell vector. Learning is stable because all adaptive weights can only decrease in time. Decreasing weights correspond to increasing sizes of category "boxes". Smaller vigilance values lead to larger category boxes. Improved prediction is achieved by training the system several times using different orderings of the input set. This voting strategy can also be used to assign probability estimates to competing predictions given small, noisy, or incomplete training sets. Four classes of simulations illustrate Fuzzy ARTMAP performance as compared to benchmark back propagation and genetic algorithm systems. These simulations include (i) finding points inside vs. outside a circle; (ii) learning to tell two spirals apart; (iii) incremental approximation of a piecewise continuous function; and (iv) a letter recognition database. The Fuzzy ARTMAP system is also compared to Salzberg's NGE system and to Simpson's FMMC system.

Real-Time and Data-Driven Operation Optimization and Knowledge Discovery for an Enterprise Information System

An enterprise information system (EIS) is an integrated data-applications platform characterized by diverse, heterogeneous, and distributed data sources. For many enterprises, a number of business processes still depend heavily on static rule-based methods and extensive human expertise. Enterprises are faced with the need for optimizing operation scheduling, improving resource utilization, discovering useful knowledge, and making data-driven decisions.

This thesis research is focused on real-time optimization and knowledge discovery that addresses workflow optimization, resource allocation, as well as data-driven predictions of process-execution times, order fulfillment, and enterprise service-level performance. In contrast to prior work on data analytics techniques for enterprise performance optimization, the emphasis here is on realizing scalable and real-time enterprise intelligence based on a combination of heterogeneous system simulation, combinatorial optimization, machine-learning algorithms, and statistical methods.

On-demand digital-print service is a representative enterprise requiring a powerful EIS.We use real-life data from Reischling Press, Inc. (RPI), a digit-print-service provider (PSP), to evaluate our optimization algorithms.

In order to handle the increase in volume and diversity of demands, we first present a high-performance, scalable, and real-time production scheduling algorithm for production automation based on an incremental genetic algorithm (IGA). The objective of this algorithm is to optimize the order dispatching sequence and balance resource utilization. Compared to prior work, this solution is scalable for a high volume of orders and it provides fast scheduling solutions for orders that require complex fulfillment procedures. Experimental results highlight its potential benefit in reducing production inefficiencies and enhancing the productivity of an enterprise.

We next discuss analysis and prediction of different attributes involved in hierarchical components of an enterprise. We start from a study of the fundamental processes related to real-time prediction. Our process-execution time and process status prediction models integrate statistical methods with machine-learning algorithms. In addition to improved prediction accuracy compared to stand-alone machine-learning algorithms, it also performs a probabilistic estimation of the predicted status. An order generally consists of multiple series and parallel processes. We next introduce an order-fulfillment prediction model that combines advantages of multiple classification models by incorporating flexible decision-integration mechanisms. Experimental results show that adopting due dates recommended by the model can significantly reduce enterprise late-delivery ratio. Finally, we investigate service-level attributes that reflect the overall performance of an enterprise. We analyze and decompose time-series data into different components according to their hierarchical periodic nature, perform correlation analysis,

and develop univariate prediction models for each component as well as multivariate models for correlated components. Predictions for the original time series are aggregated from the predictions of its components. In addition to a significant increase in mid-term prediction accuracy, this distributed modeling strategy also improves short-term time-series prediction accuracy.

In summary, this thesis research has led to a set of characterization, optimization, and prediction tools for an EIS to derive insightful knowledge from data and use them as guidance for production management. It is expected to provide solutions for enterprises to increase reconfigurability, accomplish more automated procedures, and obtain data-driven recommendations or effective decisions.

Modeling the Effects of Habitat Fragmentation on the Movements of the Mongolian Gazelle in the Eastern Mongolian Steppe

The Mongolian gazelle, Procapra gutturosa, resides in the immense and dynamic ecosystem of the Eastern Mongolian Steppe. The Mongolian Steppe ecosystem dynamics, including vegetation availability, change rapidly and dramatically due to unpredictable precipitation patterns. The Mongolian gazelle has adapted to this unpredictable vegetation availability by making long range nomadic movements. However, predicting these movements is challenging and requires a complex model. An accurate model of gazelle movements is needed, as rampant habitat fragmentation due to human development projects - which inhibit gazelles from obtaining essential resources - increasingly threaten this nomadic species. We created a novel model using an Individual-based Neural Network Genetic Algorithm (ING) to predict how habitat fragmentation affects animal movement, using the Mongolian Steppe as a model ecosystem. We used Global Positioning System (GPS) collar data from real gazelles to “train” our model to emulate characteristic patterns of Mongolian gazelle movement behavior. These patterns are: preferred vegetation resources (NDVI), displacement over certain time lags, and proximity to human areas. With this trained model, we then explored how potential scenarios of habitat fragmentation may affect gazelle movement. This model can be used to predict how fragmentation of the Mongolian Steppe may affect the Mongolian gazelle. In addition, this model is novel in that it can be applied to other ecological scenarios, since we designed it in modules that are easily interchanged.

The FUELGEN alternative: an evolutionary approach. The architecture

This paper introduces a few architectural concepts from FUELGEN, that generates a "cloud" of reload patterns, like the generator in the FUELCON expert system, but unlike that generator, is based on a genetic algorithm. There are indications FUELGEN may outperform FUELCON and other tools as reported in the literature, in well-researched case studies, but careful comparisons have to be carried out. This paper complements the information in two other recent papers on FUELGEN. Moreover, a sequel project is outlined.

FUELGEN: effective evolutionary design of refuellings for pressurized water reactors

The paper describes the design of an efficient and robust genetic algorithm for the nuclear fuel loading problem (i.e., refuellings: the in-core fuel management problem) - a complex combinatorial, multimodal optimisation., Evolutionary computation as performed by FUELGEN replaces heuristic search of the kind performed by the FUELCON expert system (CAI 12/4), to solve the same problem. In contrast to the traditional genetic algorithm which makes strong requirements on the representation used and its parameter setting in order to be efficient, the results of recent research results on new, robust genetic algorithms show that representations unsuitable for the traditional genetic algorithm can still be used to good effect with little parameter adjustment. The representation presented here is a simple symbolic one with no linkage attributes, making the genetic algorithm particularly easy to apply to fuel loading problems with differing core structures and assembly inventories. A nonlinear fitness function has been constructed to direct the search efficiently in the presence of the many local optima that result from the constraint on solutions.

Software Release Planning: An Evolutionary and Iterative Approach

Abstract To achieve higher flexibility and to better satisfy actual customer requirements, there is an increasing tendency to develop and deliver software in an incremental fashion. In adopting this process, requirements are delivered in releases and so a decision has to be made on which requirements should be delivered in which release. Three main considerations that need to be taken account of are the technical precedences inherent in the requirements, the typically conflicting priorities as determined by the representative stakeholders, as well as the balance between required and available effort. The technical precedence constraints relate to situations where one requirement cannot be implemented until another is completed or where one requirement is implemented in the same increment as another one. Stakeholder preferences may be based on the perceived value or urgency of delivered requirements to the different stakeholders involved. The technical priorities and individual stakeholder priorities may be in conflict and difficult to reconcile. This paper provides (i) a method for optimally allocating requirements to increments; (ii) a means of assessing and optimizing the degree to which the ordering conflicts with stakeholder priorities within technical precedence constraints; (iii) a means of balancing required and available resources for all increments; and (iv) an overall method called EVOLVE aimed at the continuous planning of incremental software development. The optimization method used is iterative and essentially based on a genetic algorithm. A set of the most promising candidate solutions is generated to support the final decision. The paper evaluates the proposed approach using a sample project.

A novel approach to dynamic modelling of polymer extrusion for improved process control

Polymer extrusion is a complex process and the availability of good dynamic models is key for improved system operation. Previous modelling attempts have failed adequately to capture the non-linearities of the process or prove too complex for control applications. This work presents a novel approach to the problem by the modelling of extrusion viscosity and pressure, adopting a grey box modelling technique that combines mechanistic knowledge with empirical data using a genetic algorithm approach. The models are shown to outperform those of a much higher order generated by a conventional black box technique while providing insight into the underlying processes at work within the extruder.

The VLT-FLAMES survey of massive stars: mass loss and rotation of early-type stars in the SMC

We have studied the optical spectra of a sample of 31 O- and early B-type stars in the Small Magellanic Cloud, 21 of which are associated with the young massive cluster NGC 346. Stellar parameters are determined using an automated fitting method (Mokiem et al. 2005, A&A, 441, 711), which combines the stellar atmosphere code FASTWIND (Puls et al. 2005, A&A, 435, 669) with the genetic algorithm based optimisation routine PIKAIA (Charbonneau 1995, ApJS, 101, 309). Comparison with predictions of stellar evolution that account for stellar rotation does not result in a unique age, though most stars are best represented by an age of 1-3 Myr. The automated method allows for a detailed determination of the projected rotational velocities. The present day v(r) sin i distribution of the 21 dwarf stars in our sample is consistent with an underlying rotational velocity (v(r)) distribution that can be characterised by a mean velocity of about 160-190 km s(-1) and an effective half width of 100-150 km s(-1). The vr distribution must include a small percentage of slowly rotating stars. If predictions of the time evolution of the equatorial velocity for massive stars within the environment of the SMC are correct (Maeder & Meynet 2001, A&A, 373, 555), the young age of the cluster implies that this underlying distribution is representative for the initial rotational velocity distribution. The location in the Hertzsprung-Russell diagram of the stars showing helium enrichment is in qualitative agreement with evolutionary tracks accounting for rotation, but not for those ignoring vr. The mass loss rates of the SMC objects having luminosities of log L-star/L-circle dot greater than or similar to 5.4 are in excellent agreement with predictions by Vink et al. (2001, A&A, 369, 574). However, for lower luminosity stars the winds are too weak to determine. M accurately from the optical spectrum. Three targets were classified as Vz stars, two of which are located close to the theoretical zero-age main sequence. Three lower luminosity targets that were not classified as Vz stars are also found to lie near the ZAMS. We argue that this is related to a temperature effect inhibiting cooler from displaying the spectral features required for the Vz luminosity class.

Learning soft computing control strategies in a modular neural network architecture

Modelling and control of nonlinear dynamical systems is a challenging problem since the dynamics of such systems change over their parameter space. Conventional methodologies for designing nonlinear control laws, such as gain scheduling, are effective because the designer partitions the overall complex control into a number of simpler sub-tasks. This paper describes a new genetic algorithm based method for the design of a modular neural network (MNN) control architecture that learns such partitions of an overall complex control task. Here a chromosome represents both the structure and parameters of an individual neural network in the MNN controller and a hierarchical fuzzy approach is used to select the chromosomes required to accomplish a given control task. This new strategy is applied to the end-point tracking of a single-link flexible manipulator modelled from experimental data. Results show that the MNN controller is simple to design and produces superior performance compared to a single neural network (SNN) controller which is theoretically capable of achieving the desired trajectory. (C) 2003 Elsevier Ltd. All rights reserved.

A Hybrid Evolutionary Approach to the Nurse Rostering Problem

Nurse rostering is a difficult search problem with many constraints. In the literature, a number of approaches have been investigated including penalty function methods to tackle these constraints within genetic algorithm frameworks. In this paper, we investigate an extension of a previously proposed stochastic ranking method, which has demonstrated superior performance to other constraint handling techniques when tested against a set of constrained optimisation benchmark problems. An initial experiment on nurse rostering problems demonstrates that the stochastic ranking method is better in finding feasible solutions but fails to obtain good results with regard to the objective function. To improve the performance of the algorithm, we hybridise it with a recently proposed simulated annealing hyper-heuristic within a local search and genetic algorithm framework. The hybrid algorithm shows significant improvement over both the genetic algorithm with stochastic ranking and the simulated annealing hyper-heuristic alone. The hybrid algorithm also considerably outperforms the methods in the literature which have the previously best known results.

The VLT-FLAMES survey of massive stars: wind properties and evolution of hot massive stars in the Large Magellanic Cloud

We have studied the optical spectra of a sample of 28 O- and early B-type stars in the Large Magellanic Cloud, 22 of which are associated with the young star forming region N11. Our observations sample the central associations of LH9 and LH10, and the surrounding regions. Stellar parameters are determined using an automated fitting method ( Mokiem et al. 2005), which combines the stellar atmosphere code fastwind ( Puls et al. 2005) with the genetic algorithm based optimisation routine PIKAIA ( Charbonneau 1995). We derive an age of 7.0 +/- 1.0 and 3.0 +/- 1.0 Myr for LH9 and LH10, respectively. The age difference and relative distance of the associations are consistent with a sequential star formation scenario in which stellar activity in LH9 triggered the formation of LH10. Our sample contains four stars of spectral type O2. From helium and hydrogen line fitting we find the hottest three of these stars to be similar to 49- 54 kK ( compared to similar to 45- 46 kK for O3 stars). Detailed determination of the helium mass fraction reveals that the masses of helium enriched dwarfs and giants derived in our spectroscopic analysis are systematically lower than those implied by non-rotating evolutionary tracks. We interpret this as evidence for efficient rotationally enhanced mixing leading to the surfacing of primary helium and to an increase of the stellar luminosity. This result is consistent with findings for SMC stars by Mokiem et al. ( 2006). For bright giants and supergiants no such mass discrepancy is found; these stars therefore appear to follow tracks of modestly or non-rotating objects. The set of programme stars was sufficiently large to establish the mass loss rates of OB stars in this Z similar to 1/2 Z(circle dot) environment sufficiently accurate to allow for a quantitative comparison with similar objects in the Galaxy and the SMC. The mass loss properties are found to be intermediate to massive stars in the Galaxy and SMC. Comparing the derived modified wind momenta D-mom as a function of luminosity with predictions for LMC metallicities by Vink et al. ( 2001) yields good agreement in the entire luminosity range that was investigated, i.e. 5.0