No description available.

Many complex systems in nature are governed by simple local interactions, although a number are also described by global interactions. For example, within the field of hydraulics the Navier-Stokes equations describe free-surface water flow, through means of the global preservation of water volume, momentum and energy. However, solving such partial differential equations (PDEs) is computationally expensive when applied to large 2D flow problems. An alternative which reduces the computational complexity, is to use a local derivative to approximate the PDEs, such as finite difference methods, or Cellular Automata (CA). The high speed processing of such simulations is important to modern scientific investigation especially within urban flood modelling, as urban expansion continues to increase the number of impervious areas that need to be modelled. Large numbers of model runs or large spatial or temporal resolution simulations are required in order to investigate, for example, climate change, early warning systems, and sewer design optimisation. The recent introduction of the Graphics Processor Unit (GPU) as a general purpose computing device (General Purpose Graphical Processor Unit, GPGPU) allows this hardware to be used for the accelerated processing of such locally driven simulations. A novel CA transformation for use with GPUs is proposed here to make maximum use of the GPU hardware. CA models are defined by the local state transition rules, which are used in every cell in parallel, and provide an excellent platform for a comparative study of possible alternative state transition rules. Writing local state transition rules for CA systems is a difficult task for humans due to the number and complexity of possible interactions, and is known as the ‘inverse problem’ for CA. Therefore, the use of Genetic Programming (GP) algorithms for the automatic development of state transition rules from example data is also investigated in this thesis. GP is investigated as it is capable of searching the intractably large areas of possible state transition rules, and producing near optimal solutions. However, such population-based optimisation algorithms are limited by the cost of many repeated evaluations of the fitness function, which in this case requires the comparison of a CA simulation to given target data. Therefore, the use of GPGPU hardware for the accelerated learning of local rules is also developed. Speed-up factors of up to 50 times over serial Central Processing Unit (CPU) processing are achieved on simple CA, up to 5-10 times speedup over the fully parallel CPU for the learning of urban flood modelling rules. Furthermore, it is shown GP can generate rules which perform competitively when compared with human formulated rules. This is achieved with generalisation to unseen terrains using similar input conditions and different spatial/temporal resolutions in this important application domain.

Several studies indicate that the data-driven models have proven to be potentially useful tools in hydrological modeling. Nevertheless, it is a common perception among researchers and practitioners that the usefulness of the system theoretic models is limited to forecast applications, and they cannot be used as a tool for scientific investigations. Also, the system-theoretic models are believed to be less reliable as they characterize the hydrological processes by learning the input-output patterns embedded in the dataset and not based on strong physical understanding of the system. It is imperative that the above concerns needs to be addressed before the data-driven models can gain wider acceptability by researchers and practitioners.<p>In this research different methods and tools that can be adopted to promote transparency in the data-driven models are probed with the objective of extending the usefulness of data-driven models beyond forecast applications as a tools for scientific investigations, by providing additional insights into the underlying input-output patterns based on which the data-driven models arrive at a decision. In this regard, the utility of self-organizing networks (competitive learning and self-organizing maps) in learning the patterns in the input space is evaluated by developing a novel neural network model called the spiking modular neural networks (SMNNs). The performance of the SMNNs is evaluated based on its ability to characterize streamflows and actual evapotranspiration process. Also the utility of self-organizing algorithms, namely genetic programming (GP), is evaluated with regards to its ability to promote transparency in data-driven models. The robustness of the GP to evolve its own model structure with relevant parameters is illustrated by applying GP to characterize the actual-evapotranspiration process. The results from this research indicate that self-organization in learning, both in terms of self-organizing networks and self-organizing algorithms, could be adopted to promote transparency in data-driven models.<p>In pursuit of improving the reliability of the data-driven models, different methods for incorporating uncertainty estimates as part of the data-driven model building exercise is evaluated in this research. The local-scale models are shown to be more reliable than the global-scale models in characterizing the saturated hydraulic conductivity of soils. In addition, in this research, the importance of model structure uncertainty in geophysical modeling is emphasized by developing a framework to account for the model structure uncertainty in geophysical modeling. The contribution of the model structure uncertainty to the predictive uncertainty of the model is shown to be larger than the uncertainty associated with the model parameters. Also it has been demonstrated that increasing the model complexity may lead to a better fit of the function, but at the cost of an increasing level of uncertainty. It is recommended that the effect of model structure uncertainty should be considered for developing reliable hydrological models.

Ultrasonography is widely used in medical diagnosis with the advantages of being low cost, non-invasive and capable of real time imaging. When interpreting ultrasonographic images of mammalian ovaries, the structures of interest are follicles, corpora lutea (CL) and stroma. This thesis presents an approach to perform CL texture analysis, including detection and segmentation, based on the classiers trained by genetic programming (GP). The objective of CL detection is to determine whether there is a CL in the ovarian images, while the goal of segmentation is to localize the CL within the image. Genetic programming (GP) oers a solution through the evolution of computer programs by methods inspired by the mechanisms of natural selection. Herein, we use rotationally invariant local binary patterns (LBP) to encode the local texture features. These are used by the programs which are manipulated by GP to obtain highly t CL classiers. Grayscale standardization was performed on all images in our data set based on the reference grayscale in each image. CL classication programs were evolved by genetic programming and tested on ultrasonographic images of ovaries. On the bovine dataset, our CL detection algorithm is reliable and robust. The detection algorithm correctly determined the presence or absence of a CL in 93.3% of 60 test images. The segmentation algorithm achieved a mean ( standard deviation) sensitivity and specicity of 0.87 (0.14) and 0.91 (0.05), respectively, over the 30 CL images. Our CL segmentation algorithm is an improvement over the only previously published algorithm, since our method is fully automatic and does not require the placement of an initial contour. The success of these algorithms demonstrates that similar algorithms designed for analysis of in vivo human ovaries are likely viable.

Ultrasonography is widely used in medical diagnosis with the advantages of being low cost, non-invasive and capable of real time imaging. When interpreting ultrasonographic images of mammalian ovaries, the structures of interest are follicles, corpora lutea (CL) and stroma. This thesis presents an approach to perform CL texture analysis, including detection and segmentation, based on the classiers trained by genetic programming (GP). The objective of CL detection is to determine whether there is a CL in the ovarian images, while the goal of segmentation is to localize the CL within the image. Genetic programming (GP) oers a solution through the evolution of computer programs by methods inspired by the mechanisms of natural selection. Herein, we use rotationally invariant local binary patterns (LBP) to encode the local texture features. These are used by the programs which are manipulated by GP to obtain highly t CL classiers. Grayscale standardization was performed on all images in our data set based on the reference grayscale in each image. CL classication programs were evolved by genetic programming and tested on ultrasonographic images of ovaries. On the bovine dataset, our CL detection algorithm is reliable and robust. The detection algorithm correctly determined the presence or absence of a CL in 93.3% of 60 test images. The segmentation algorithm achieved a mean ( standard deviation) sensitivity and specicity of 0.87 (0.14) and 0.91 (0.05), respectively, over the 30 CL images. Our CL segmentation algorithm is an improvement over the only previously published algorithm, since our method is fully automatic and does not require the placement of an initial contour. The success of these algorithms demonstrates that similar algorithms designed for analysis of in vivo human ovaries are likely viable.

Several studies indicate that the data-driven models have proven to be potentially useful tools in hydrological modeling. Nevertheless, it is a common perception among researchers and practitioners that the usefulness of the system theoretic models is limited to forecast applications, and they cannot be used as a tool for scientific investigations. Also, the system-theoretic models are believed to be less reliable as they characterize the hydrological processes by learning the input-output patterns embedded in the dataset and not based on strong physical understanding of the system. It is imperative that the above concerns needs to be addressed before the data-driven models can gain wider acceptability by researchers and practitioners.<p>In this research different methods and tools that can be adopted to promote transparency in the data-driven models are probed with the objective of extending the usefulness of data-driven models beyond forecast applications as a tools for scientific investigations, by providing additional insights into the underlying input-output patterns based on which the data-driven models arrive at a decision. In this regard, the utility of self-organizing networks (competitive learning and self-organizing maps) in learning the patterns in the input space is evaluated by developing a novel neural network model called the spiking modular neural networks (SMNNs). The performance of the SMNNs is evaluated based on its ability to characterize streamflows and actual evapotranspiration process. Also the utility of self-organizing algorithms, namely genetic programming (GP), is evaluated with regards to its ability to promote transparency in data-driven models. The robustness of the GP to evolve its own model structure with relevant parameters is illustrated by applying GP to characterize the actual-evapotranspiration process. The results from this research indicate that self-organization in learning, both in terms of self-organizing networks and self-organizing algorithms, could be adopted to promote transparency in data-driven models.<p>In pursuit of improving the reliability of the data-driven models, different methods for incorporating uncertainty estimates as part of the data-driven model building exercise is evaluated in this research. The local-scale models are shown to be more reliable than the global-scale models in characterizing the saturated hydraulic conductivity of soils. In addition, in this research, the importance of model structure uncertainty in geophysical modeling is emphasized by developing a framework to account for the model structure uncertainty in geophysical modeling. The contribution of the model structure uncertainty to the predictive uncertainty of the model is shown to be larger than the uncertainty associated with the model parameters. Also it has been demonstrated that increasing the model complexity may lead to a better fit of the function, but at the cost of an increasing level of uncertainty. It is recommended that the effect of model structure uncertainty should be considered for developing reliable hydrological models.

Το διάνυσμα προτύπων αποτελεί σύνθεση ενός συνόλου χαρακτηριστικών γνωρισμάτων και ταξινομείται σε δύο κατηγορίες: τα αρχικά χαρακτηριστικά γνωρίσματα και ένα σύνολο μη γραμμικών προβολών των αρχικών χαρακτηριστικών γνωρισμάτων. Στην ταξινόμηση, η κατασκευή χαρακτηριστικών γνωρισμάτων είναι ένα βήμα προ-επεξεργασίας στο οποίο ένα ή περισσότερα γνωρίσματα κατασκευάζονται από ένα αρχικό σύνολο. Ο αριθμός και ο τύπος των χαρακτηριστικών γνωρισμάτων είναι κρίσιμα για την ακρίβεια ταξινόμησης και την υπολογιστική πολυπλοκότητα. Καθώς ο αριθμός των χαρακτηριστικών γνωρισμάτων αυξάνει, απαιτούνται πρόσθετα παραδείγματα για να ολοκληρώσουν μια αξιόπιστη διαδικασία κατάρτισης, επιτρέποντας περισσότερες δυνατότητες αξιόπιστης γενίκευσης χωρίς υπέρ-εκπαίδευση αποτελεσμάτων. Σε αυτή τη διπλωματική εργασία εξετάζουμε και αναλύουμε τη χρήση του Γενετικού προγραμματισμού για την προ-επεξεργασία δεδομένων έτσι ώστε να κατασκευάσουμε μη γραμμικά, ιδιαίτερα προφητικά, χαρακτηριστικά γνωρίσματα από τα αρχικά. Για το σκοπό αυτό χρησιμοποιούμε γενετικούς αλγορίθμους και συγκεκριμένα τον C4.5 αλγόριθμο εκμάθησης δέντρων απόφασης, τον G-Net, ένα διανεμημένο εξελικτικό αλγόριθμο ικανό να συμπεράνει τους ταξινομητές από τα προ-συγκεντρωμένα στοιχεία καθώς και τη μέθοδο που βασίζεται στο συνδυασμό της καθιερωμένης τεχνικής της γραμματικής εξέλιξης και των τεχνητών νευρικών δικτύων. Εφαρμόζοντας τον γενετικό προγραμματισμό σε διάφορα σύνολα δεδομένων ταξινόμησης επιτυγχάνουμε μεγαλύτερη ακρίβεια ταξινόμησης. Όλοι οι αλγόριθμοι που χρησιμοποιήθηκαν έδωσαν πολύ καλύτερη απόδοση στα σύνολα δεδομένων όταν συμπεριλήφθηκε μια ενιαία εξελιγμένη μεταβλητή, επιλύοντας προβλήματα που βρίσκονται πολύ συχνά στις εφαρμογές υπολογιστών όπως Ιατρική και ελαττωματική διάγνωση, πρόγνωση, αναγνώριση εικόνας, κατηγοριοποίηση κειμένων, προσαρμοστική σκιαγράφηση χρηστών. / -

This research investigates the ability of genetic programming to build profitable trad- ing strategies for the Foreign Exchange Market (FX) of one major currency pair (EURUSD) using one hour prices from July 1, 2009 to November 30, 2012. We rec- ognize that such environments are likely to be non-stationary and we do not expect that a single training partition, used to train a trading agent, represents all likely future behaviours. The proposed adaptive retraining algorithm – hereafter FXGP – detects poor trading behaviours and trains a new trading agent. This represents a significant departure from current practice which assumes some form of continuous evolution. Extensive benchmarking is performed against the widely used EURUSD currency pair. The non-stationary nature of the task is shown to result in a prefer- ence for exploration over exploitation. Moreover, adopting a behavioural approach to detecting retraining events is more effective than assuming incremental adaptation on a continuous basis. From the application perspective, we demonstrate that use of a validation partition and Stop-Loss (S/L) orders significantly improves the perfor- mance of a trading agent. In addition the task of co-evolving of technical indicators (TI) and the decision trees (DT) for deploying trading agent is explicitly addressed. The results of 27 experiments of 100 simulations each demonstrate that FXGP sig- nificantly outperforms existing approaches and generates profitable solutions with a high probability.

The ultimate goal in robotics is to create machines which are more independent and rely less on humans to guide them in their operation. There are many sub-systems which may be present in such a robot, one of which is path planning — the ability to determine a sequence of positions or configurations between an initial and goal position within a particular obstacle cluttered workspace. Many classical path planning techniques have been developed, but these tend to have drawbacks such as their computational requirements; the suitability of the plans they produce for a particular application; or how well they are able to generalise to unseen problems. In recent years, evolutionary based problem solving techniques have seen a rise in popularity, possibly coinciding with the improvement in the computational power afforded researches by successful developments in hardware. These techniques adopt some of the features of natural evolution and mimic them in a computer. The increase in the number of publications in the areas of Genetic Algorithms (GA) and Genetic Programming (GP) demonstrate the success achieved when applying these techniques to ever more problem areas. This dissertation presents research conducted to determine whether there is a place for Evolutionary Approaches, and specifically GA and GP, in the development of future path planning techniques.

Grammatical Evolution (GE) is a method of utilising a general purpose evolutionary algorithm to ???evolve??? programs written in an arbitrary BNF grammar. This thesis extends GE as follows: GE as an extension of Genetic Programming (GP) A novel method of automatically extracting information from the grammar is introduced. This additional information allows the use of GP style crossover which in turn allows GE to perform identically to a strongly typed GP system as well as a non-typed (or canonical) GP system. Two test problems are presented one which is more easily solved by the GP style crossover and one which favours the tradition GE ???Ripple Crossover???. With this new crossover operator GE can now emulate GP (as well as retaining its own unique features) and can therefore now be seen as an extension of GP. Dynamically Defined Functions An extension to the BNF grammar is presented which allows the use of dynamically defined functions (DDFs). DDFs provide an alternative to the traditional approach of Automatically Defined Functions (ADFs) but have the advantage that the number of functions and their parameters do not need to be specified by the user in advance. In addition DDFs allow the architecture of individuals to change dynamically throughout the course of the run without requiring the introduction of any new form of operator. Experimental results are presented confirming the effectiveness of DDFs. Self-Selecting (or variable) crossover. A self-selecting operator is introduced which allows the system to determine, during the course of the run, which crossover operator to apply; this is tested over several problem domains and (especially where small populations are used) is shown to be effective in aiding the system to overcome local optima. Spatial Co-Evolution in Age Layered Planes (SCALP) A method of combining Hornby???s ALPS metaheuristic and a spatial co-evolution system used by Mitchell is presented; the new SCALP system is tested over three problem domains of increasing difficulty and performs extremely well in each of them.