One key challenge in optimization is the selection of a suitable set of benchmark problems. A common goal is to find functions which are representative of a class of real-world optimization problems in order to ensure findings on the benchmarks will translate to relevant problem domains. While some problem characteristics are well-covered by popular benchmarking suites, others are often overlooked. One example of such a problem characteristic is permutation invariance, where the search space consists of a set of symmetrical search regions. This type of problem occurs e.g. when a set of solutions has to be found, but the ordering within this set does not matter. The data clustering problem, often seen in machine learning contexts, is a clear example of such an optimization landscape, and has thus been proposed as a base from which optimization benchmarks can be created. In addition to the symmetry aspect, these clustering problems also contain potential regions of neutrality, which can provide an additional challenge to optimization algorithms.

Despite extensive efforts in developing machine learning-based Network Intrusion Detection Systems, inconsistencies in the choice of pre-processing of training data exists across the literature. This study surveys both seminal and recent works, identifying that while the same benchmark datasets and model classes are frequently used, the data pre-processing methods vary significantly, often without in-depth discussion and justification, limiting performance and hindering fair comparison. This work aims to address this problem, by exploring the possibility of a standardised pre-processing for Network Intrusion Detection Systems benchmark datasets We perform a detailed experimental evaluation on the most common machine learning models and four prevalent Network Intrusion Detection Systems benchmark datasets, considering a total of 72 combinations of pre-processing methods for both numerical and categorical fields in the network flow data. Based on our experimental evaluation, we propose a uniform (or ’standardised’) data pre-processing approach, which consistently performs well across the considered datasets and machine learning models. We demonstrate that our proposed method can achieve up to a 11% increase in detection accuracy over commonly applied pre-processing techniques for a shallow neural network (for the purposes of this work, we deem networks with fewer than 3 hidden layers as shallow), versus less-optimal pre-processing approaches. To facilitate future research, we make our code publicly available 1 .

Network Intrusion Detection Systems (NIDSs) are an increasingly important tool for the prevention and mitigation of cyber attacks. Over the past years, a lot of research efforts have aimed at leveraging the increasingly powerful models of Machine Learning (ML) for this purpose. A number of labelled synthetic datasets have been generated and made publicly available by researchers, and they have become the benchmarks via which new ML-based NIDS classifiers are being evaluated. Recently published results show excellent classification performance with these datasets, increasingly approaching 100 percent performance across key evaluation metrics such as Accuracy, F1 score, AUC, etc. Unfortunately, we have not yet seen these excellent academic research results translated into practical NIDS systems with such near-perfect performance. This motivated our research presented in this paper, where we analyse the statistical properties of the benign traffic in three of the more recent and relevant NIDS datasets, (CIC_IDS, UNSW_NB15, TON_IOT), by converting them into a common flow format. As a comparison, we consider two datasets obtained from real-world production networks, one from a university network and one from a medium size Internet Service Provider (ISP). Our results show that the two real-world datasets are quite similar among themselves in regards to most of the considered statistical features. Equally, the three synthetic datasets are also relatively similar within their group. However, and most importantly, our results show a distinct difference of most of the considered statistical features between the three synthetic datasets and the two real-world datasets. Since ML relies on the basic assumption of training and test datasets being sampled from the same distribution, this raises the question of how well the performance results of ML-classifiers trained on the considered synthetic datasets can translate and generalise to real-world networks. We believe this is an interesting and relevant question which provides motivation for further research in this space.

Modularity has been widely studied as a mechanism to improve the capabilities of neural networks through various techniques such as hand-crafted modular architectures and automatic approaches. While these methods have sometimes shown improvements towards generalisation ability, robustness, and efficiency, the mechanisms that enable modularity to give performance advantages are unclear. In this paper, we investigate this issue and find that the amount of network modularity for optimal performance is likely entangled in complex relationships between many other features of the network and problem environment. Therefore, direct optimisation or arbitrary designation of a suitable amount of modularity in neural networks may not be beneficial. We used a classic neuroevolutionary algorithm which enables rich, automatic optimisation and exploration of neural network architectures and weights with varying levels of modularity. The structural modularity and performance of networks generated by the NeuroEvolution of Augmenting Topologies algorithm was assessed on three reinforcement learning tasks, with and without an additional modularity objective. The results of the quality-diversity optimisation algorithm, MAP-Elites, suggest intricate conditional relationships between modularity, performance, and other predefined network features.

The eight papers in this special section focus on applications of evolutionary computation to games to demonstrate several ways in which evolution can push boundaries and explore new areas of what is possible in the realm of games research, with a focus on game-playing, automatic agent parameter tuning, automatic game testing, and procedural content generation.

Interest in reinforcement learning (RL) has recently surged due to the application of deep learning techniques, but these connectionist approaches are opaque compared with symbolic systems. Learning Classifier Systems (LCSs) are evolutionary machine learning systems that can be categorised as eXplainable AI (XAI) due to their rule-based nature. Michigan LCSs are commonly used in RL domains as the alternative Pittsburgh systems (e.g. SAMUEL) suffer from complex algorithmic design and high computational requirements; however they can produce more compact/interpretable solutions than Michigan systems. We aim to develop two novel Pittsburgh LCSs to address RL domains: PPL-DL and PPL-ST. The former acts as a "zeroth-level" system, and the latter revisits SAMUEL's core Monte Carlo learning mechanism for estimating rule strength. We compare our two Pittsburgh systems to the Michigan system XCS across deterministic and stochastic FrozenLake environments. Results show that PPL-ST performs on-par or better than PPL-DL and outperforms XCS in the presence of high levels of environmental uncertainty. Rulesets evolved by PPL-ST can achieve higher performance than those evolved by XCS, but in a more parsimonious and therefore more interpretable fashion, albeit with higher computational cost. This indicates that PPL-ST is an LCS well-suited to producing explainable policies in RL domains.

This paper presents a new Network Intrusion Detection System (NIDS) based on Graph Neural Networks (GNNs). GNNs are a relatively new sub-field of deep neural networks, which can leverage the inherent structure of graph-based data. Training and evaluation data for NIDSs are typically represented as flow records, which can naturally be represented in a graph format. In this paper, we propose E-GraphSAGE, a GNN approach that allows capturing both the edge features of a graph as well as the topological information for network intrusion detection in IoT networks. To the best of our knowledge, our proposal is the first successful, practical, and extensively evaluated approach of applying GNNs on the problem of network intrusion detection for IoT using flow-based data. Our extensive experimental evaluation on four recent NIDS benchmark datasets shows that our approach outperforms the state-of-the-art in terms of key classification metrics, which demonstrates the potential of GNNs in network intrusion detection, and provides motivation for further research.

We propose a new research framework by which the nascent discipline of human-AI teaming can be explored within experimental environments in preparation for transferal to real-world contexts. We examine the existing literature and unanswered research questions through the lens of an Agile approach to construct our proposed framework. Our framework aims to provide a structure for understanding the macro features of this research landscape, supporting holistic research into the acceptability of human-AI teaming to human team members and the affordances of AI team members. The framework has the potential to enhance decision-making and performance of hybrid human-AI teams. Further, our framework proposes the application of Agile methodology for research management and knowledge discovery. We propose a transferability pathway for hybrid teaming to be initially tested in a safe environment, such as a real-time strategy video game, with elements of lessons learned that can be transferred to real-world situations.

Analysing and computing with Gaussian processes arising from infinitely wide neural networks has recently seen a resurgence in popularity. Despite this, many explicit covariance functions of networks with activation functions used in modern networks remain unknown. Furthermore, while the kernels of deep networks can be computed iteratively, theoretical understanding of deep kernels is lacking, particularly with respect to fixed-point dynamics. Firstly, we derive the covariance functions of multi-layer perceptrons (MLPs) with exponential linear units (ELU) and Gaussian error linear units (GELU) and evaluate the performance of the limiting Gaussian processes on some benchmarks. Secondly, and more generally, we analyse the fixed-point dynamics of iterated kernels corresponding to a broad range of activation functions. We find that unlike some previously studied neural network kernels, these new kernels exhibit non-trivial fixed-point dynamics which are mirrored in finite-width neural networks. The fixed point behaviour present in some networks explains a mechanism for implicit regularisation in overparameterised deep models. Our results relate to both the static iid parameter conjugate kernel and the dynamic neural tangent kernel constructions

In many countries around the world (including Australia), the prescribing of opioid analgesic drugs is an increasing trend associated with significant increases in drug-related patient harm such as abuse, overdose, and death. In Australia, the Medicines Regulation and Quality Unit within Queensland Health maintains a database recording opioid analgesic drug prescriptions dispensed across the State (population 4.703 million). In this work, we propose the use of network visualisation and analysis as a tool for improved understanding of these data. Prescribing data for Fentanyl patches, a strong opioid with high potential for misuse and subsequent harm, across Queensland, Australia from 2011 to 2018 is analysed as an example of using network analysis, where prescribing patterns are viewed as a dynamic, bipartite graph of the interactions between patients and prescribers over time. The technique provides a global view of a large state-wide prescribing dataset, including the distribution of subgraph structures present. Local analysis is also carried out to demonstrate the clinical utility of the technique, including the dynamics of the graph structure over time. A variety of network statistics that measure network structural and dynamic properties are presented to reveal the characteristics and trends of drug seeking and prescribing behaviours. This approach has been recognised by healthcare professionals at Queensland Health as leading to new and useful insights on the relationship between patients and prescribers and supporting their advisory role to reduce patient harm from inappropriate use of prescription drugs.

In this paper we present a method for automatically deriving a Reversible Jump Markov chain Monte Carlo sampler from probabilistic programs that specify the target and proposal distributions. The main challenge in automatically deriving such an inference procedure, in comparison to deriving a generic Metropolis-Hastings sampler, is in calculating the Jacobian adjustment to the proposal acceptance ratio. To achieve this, our approach relies on the interaction of several different components, including automatic differentiation, transformation inversion, and optimised code generation. We also present Stochaskell, a new probabilistic programming language embedded in Haskell, which provides an implementation of our method.

Data-driven optimization problems such as clustering provide a real-world representative source of instances for benchmarking continuous metaheuristic optimization algorithms. Although many metaheuristics and other algorithms have been applied to clustering problems, relatively little research has attempted to explore the structure of the problem space and/or fitness landscape for these problems. In contrast, problem-specific analysis and insights for several classes of combinatorial problems have been developed. This paper investigates the structure of the fitness landscapes of clustering problems, focusing on the fundamental parameters that define problem instances (the dimensionality, number and distribution of the data points and the number of clusters). The paper also provides a general method for active, targeted generation of problem instances based on real-world datasets. The results provide a number of new insights into this family of continuous optimization problems as well as methods and guidelines intended to facilitate better experimental evaluation and comparison of continuous optimization algorithms.

In the analysis of machine learning models, it is often convenient to assume that the parameters are IID. This assumption is not satisfied when the parameters are updated through training processes such as Stochastic Gradient Descent. A relaxation of the IID condition is a probabilistic symmetry known as exchangeability. We show the sense in which the weights in MLPs are exchangeable. This yields the result that in certain instances, the layer-wise kernel of fully-connected layers remains approximately constant during training. Our results shed light on such kernel properties throughout training while limiting the use of unrealistic assumptions.

Exploratory Landscape Analysis provides sample-based methods to calculate features of black-box optimization problems in a quantitative and measurable way. Many problem features have been proposed in the literature in an attempt to provide insights into the structure of problem landscapes and to use in selecting an effective algorithm for a given optimization problem. While there has been some success, evaluating the utility of problem features in practice presents some significant challenges. Machine learning models have been employed as part of the evaluation process, but they may require additional information about the problems as well as having their own hyper-parameters, biases and experimental variability. As a result, extra layers of uncertainty and complexity are added into the experimental evaluation process, making it difficult to clearly assess the effect of the problem features. In this article, we propose a novel method for the evaluation of problem features which can be applied directly to individual or groups of features and does not require additional machine learning techniques or confounding experimental factors. The method is based on the feature's ability to detect a prior ranking of similarity in a set of problems. Analysis of Variance (ANOVA) significance tests are used to determine if the feature has successfully distinguished the successive problems in the set. Based on ANOVA test results, a percentage score is assigned to each feature for different landscape characteristics. Experimental results for twelve different features on four problem transformations demonstrate the method and provide quantitative evidence about the ability of different problem features to detect specific properties of problem landscapes.

Metaheuristic algorithms usually aim to solve nonconvex optimization problems in black-box and high-dimensional scenarios. Characterizing and understanding the properties of nonconvex problems is therefore important for effectively analyzing metaheuristic algorithms and their development, improvement and selection for problem solving. This paper establishes a novel analysis framework called nonconvex ratio analysis , which can characterize nonconvex continuous functions by measuring the degree of nonconvexity of a problem. This analysis uses two quantitative measures: the nonconvex ratio for global characterization and the local nonconvex ratio for detailed characterization. Midpoint convexity and Monte Carlo integral are important methods for constructing the measures. Furthermore, as a practical feature, we suggest a rapid characterization measure that uses the local nonconvex ratio and can characterize certain black-box high-dimensional functions using a much smaller sample. Throughout this paper, the effectiveness of the proposed measures is confirmed by numerical experiments using the COCO function set.

An interesting approach to analyzing neural networks that has received renewed attention is to examine the equivalent kernel of the neural network. This is based on the fact that a fully connected feedforward network with one hidden layer, a certain weight distribution, an activation function, and an infinite number of neurons can be viewed as a mapping into a Hilbert space. We derive the equivalent kernels of MLPs with ReLU or Leaky ReLU activations for all rotationally-invariant weight distributions, generalizing a previous result that required Gaussian weight distributions. Additionally, the Central Limit Theorem is used to show that for certain activation functions, kernels corresponding to layers with weight distributions having $0$ mean and finite absolute third moment are asymptotically universal, and are well approximated by the kernel corresponding to layers with spherical Gaussian weights. In deep networks, as depth increases the equivalent kernel approaches a pathological fixed point, which can be used to argue why training randomly initialized networks can be difficult. Our results also have implications for weight initialization.

BACKGROUND: Erroneous patient birthdates are common in health databases. Detection of these errors usually involves manual verification, which can be resource intensive and impractical. By identifying a frequent manifestation of birthdate errors, this paper presents a principled and statistically driven procedure to identify erroneous patient birthdates. RESULTS: Generalized additive models (GAM) enabled explicit incorporation of known demographic trends and birth patterns. With false positive rates controlled, the method identified birthdate contamination with high accuracy. In the health data set used, of the 58 actual incorrect birthdates manually identified by the domain expert, the GAM-based method identified 51, with 8 false positives (resulting in a positive predictive value of 86.0% (51/59) and a false negative rate of 12.0% (7/58)). These results outperformed linear time-series models. CONCLUSIONS: The GAM-based method is an effective approach to identify systemic birthdate errors, a common data quality issue in both clinical and administrative databases, with high accuracy.

In metaheuristic optimization, understanding the relationship between problems and algorithms is important but nontrivial. There has been a growing interest in the literature on techniques for analyzing problems and algorithm performance; however, the validity of the assumptions and implementation choices behind many techniques is often not closely examined. In this paper, we review some interesting theoretical properties regarding sampling techniques and distance metrics in continuous spaces. In particular, we examine the effect of using Euclidean distance in conjunction with uniform random sampling on the behavior of the Dispersion metric. We show that the current methodology employed for the estimation of dispersion has important flaws, and we propose and evaluate modifications to improve the methodology. The modifications are simple and do not add significant complexity or computational effort to the methodology.

Estimation of Distribution Algorithms (EDAs) focus on explicitly modelling dependencies between solution variables. A Gaussian distribution over continuous variables is commonly used, with several different covariance matrix structures ranging from diagonal i.e. Univariate Marginal Distribution Algorithm (UMDAc) to full i.e. Estimation of Multivariate Normal density Algorithm (EMNA). A diagonal covariance model is simple but is unable to directly represent covariances between problem variables. On the other hand, a full covariance model requires estimation of (more) parameters from the selected population. In practice, numerical issues can arise with this estimation problem. In addition, the performance of the model has been shown to be sometimes undesirable. In this paper, a modified Gaussian-based continuous EDA is proposed, called sEDA, that provides a mechanism to control the amount of covariance parameters estimated within the Gaussian model. To achieve this, a simple variable screening technique from experimental design is adapted and combined with an idea inspired by the Pareto-front in multi-objective optimization. Compared to EMNAglobal, the algorithm provides improved numerical stability and can use a smaller selected population. Experimental results are presented to evaluate and compare the performance of the algorithm to UMDAcand EMNAglobal.

In this paper we extend a previously proposed randomized landscape generator in combination with a comparative experimental methodology to study the behavior of continuous metaheuristic optimization algorithms. In particular, we generate two-dimensional landscapes with parameterized, linear ridge structure, and perform pairwise comparisons of algorithms to gain insight into what kind of problems are easy and difficult for one algorithm instance relative to another. We apply this methodology to investigate the specific issue of explicit dependency modeling in simple continuous estimation of distribution algorithms. Experimental results reveal specific examples of landscapes (with certain identifiable features) where dependency modeling is useful, harmful, or has little impact on mean algorithm performance. Heat maps are used to compare algorithm performance over a large number of landscape instances and algorithm trials. Finally, we perform a meta-search in the landscape parameter space to find landscapes which maximize the performance between algorithms. The results are related to some previous intuition about the behavior of these algorithms, but at the same time lead to new insights into the relationship between dependency modeling in EDAs and the structure of the problem landscape. The landscape generator and overall methodology are quite general and extendable and can be used to examine specific features of other algorithms.

Reinforcement learning (RL) is a popular machine learning technique that has many successes in learning how to play classic style games. Applying RL to first person shooter (FPS) games is an interesting area of research as it has the potential to create diverse behaviors without the need to implicitly code them. This paper investigates the tabular Sarsa (λ) RL algorithm applied to a purpose built FPS game. The first part of the research investigates using RL to learn bot controllers for the tasks of navigation, item collection, and combat individually. Results showed that the RL algorithm was able to learn a satisfactory strategy for navigation control, but not to the quality of the industry standard pathfinding algorithm. The combat controller performed well against a rule-based bot, indicating promising preliminary results for using RL in FPS games. The second part of the research used pretrained RL controllers and then combined them by a number of different methods to create a more generalized bot artificial intelligence (AI). The experimental results indicated that RL can be used in a generalized way to control a combination of tasks in FPS bots such as navigation, item collection, and combat.

A major challenge in the development of peptide-based vaccines is finding the right immunogenic element, with efficient and long-lasting immunization effects, from large potential targets encoded by pathogen genomes. Computer models are convenient tools for scanning pathogen genomes to preselect candidate immunogenic peptides for experimental validation. Current methods predict many false positives resulting from a low prevalence of true positives. We develop a test reject method based on the prediction uncertainty estimates determined by Gaussian process regression. This method filters false positives among predicted epitopes from a pathogen genome. The performance of stand-alone Gaussian process regression is compared to other state-of-the-art methods using cross validation on 11 benchmark data sets. The results show that the Gaussian process method has the same accuracy as the top performing algorithms. The combination of Gaussian process regression with the proposed test reject method is used to detect true epitopes from the Vaccinia virus genome. The test rejection increases the prediction accuracy by reducing the number of false positives without sacrificing the method's sensitivity. We show that the Gaussian process in combination with test rejection is an effective method for prediction of T-cell epitopes in large and diverse pathogen genomes, where false positives are of concern.

This paper presents some new analytical results on the continuous Univariate Marginal Distribution Algorithm (UMDAC), which is a well known Estimation of Distribution Algorithm based on Gaussian distributions. As the extension of the current theoretical work built on the assumption of infinite populations, the convergence behavior of UMDAC with finite populations is formally analyzed. We show both analytically and experimentally that, on flat landscapes, the Gaussian model in UMDAC tends to collapse with high probability, which is an important fact that is not well understood before.

Success rate is a commonly adopted performance criterion for evaluating Evolutionary Algorithms due to their inherent randomness. However, the classical large-sample binomial test based on normal distributions is only valid with a relatively large number of trials, which may not be feasible when experimental studies are very time consuming or expensive. In this paper, we give an alternative statistical test, which is suitable for situations where results from only a small number of trials are available.

Optimal Active Learning refers to a framework where the learner actively selects data points to be added to its training set in a statistically optimal way. Under the assumption of log-loss, optimal active learning can be implemented in a relatively simple and efficient manner for regression problems using Gaussian Processes. However (to date), there has been little attempt to study the experimental behavior and performance of this technique. In this paper, we present a detailed empirical evaluation of optimal active learning using Gaussian Processes across a set of seven regression problems from the DELVE repository. In particular, we examine the evaluation of optimal active learning compared to making random queries and the impact of experimental factors such as the size and construction of the different sub-datasets used as part of training and testing the models. It is shown that the multiple sources of variability can be quite significant and suggests that more care needs to be taken in the evaluation of active learning algorithms.

Metaheuristics such as Estimation of Distribution Algorithms and the Cross-Entropy method use probabilistic modelling and inference to generate candidate solutions in optimization problems. The model fitting task in this class of algorithms has largely been carried out to date based on maximum likelihood. An alternative approach that is prevalent in statistics and machine learning is to use Bayesian inference. In this paper, we provide a framework for the application of Bayesian inference techniques in probabilistic model-based optimization. Based on this framework, a simple continuous Bayesian Estimation of Distribution Algorithm is described. We evaluate and compare this algorithm experimentally with its maximum likelihood equivalent, UMDAGc.

We applied support vector machines to the prediction of the subcellular localization of transmembrane proteins, and compared the performance of different sequence kernels on this task. More specifically we measured prediction accuracy, computation time, number of kernel evaluations and number of support vectors for the spectrum, the full spectrum, the wildcard, the mismatch, the local-alignment and the residue-coupling kernel. The local-alignment achieved the highest prediction accuracy, with a Matthews correlation coefficient of 0.51, closely followed by the mismatch kernel. However, the local-alignment kernel was also the most time consuming kernel and seven times slower than the mismatch kernel. The spectrum kernel was the fastest kernel but linked to the highest number of support vectors and kernel evaluations. The residue-coupling kernel showed the lowest number of support vectors and kernel evaluations. No correlation between the number of support vectors and prediction accuracy could be observed. A localization predictor (TMPLoc) has been made available at http://pprowler.itee.uq.edu.au/TMPLoc

With the increased interest in multi-user systems with distributed decision making tasks, such as network centric warfare and free-flight air traffic control, the concept of shared situation awareness (SSA) has become more important. SSA relates to the awareness that different operators have of the system state and the information needs of their team-mates, in order to coordinate their actions effectively. However, there are many different definitions of SSA in the literature and no generally agreed methodology for measuring it. Investigation to date has primarily been conducted through human-in-the-loop (HIL) experiments, which typically are costly and return only subjective results. Agent based modelling has been proposed as a partial solution, using software agents to take the place of humans in experiments. This paper describes the application of agent based modelling to a multi-user system: namely, the SCUDhunt game. The paper examines different ways of sharing information between operators, and their effect on team performance in cases with less than perfect information. Agent based modelling enables performance to be tested on a wide range of scenarios quickly and repeatedly and yields insights that would be difficult to achieve through HIL experiments alone.

This paper presents some initial attempts to mathematically model the dynamics of a continuous Estimation of Distribution Algorithm (EDA) based on a Gaussian distribution and truncation selection. Case studies are conducted on both unimodal and multimodal problems to highlight the effectiveness of the proposed technique and explore some important properties of the EDA. With some general assumptions, we show that, for 1D unimodal problems and with the (µ, λ) scheme: (1). The behaviour of the EDA is dependent only on the general shape of the test function, rather than its specific form; (2). When initialized far from the global optimum, the EDA has a tendency to converge prematurely; (3). Given a certain selection pressure, there is a unique value for the proposed amplification parameter that could help the EDA achieve desirable performance; for 1D multimodal problems: (1). The EDA could get stuck with the (µ, λ) scheme; (2). The EDA will never get stuck with the (µ +λ) scheme.

The research literature on metaheuristic and evolutionary computation has proposed a large number of algorithms for the solution of challenging real-world optimization problems. It is often not possible to study theoretically the performance of these algorithms unless significant assumptions are made on either the algorithm itself or the problems to which it is applied, or both. As a consequence, metaheuristics are typically evaluated empirically using a set of test problems. Unfortunately, relatively little attention has been given to the development of methodologies and tools for the large-scale empirical evaluation and/or comparison of metaheuristics. In this paper, we propose a landscape (test-problem) generator that can be used to generate optimization problem instances for continuous, bound-constrained optimization problems. The landscape generator is parameterized by a small number of parameters, and the values of these parameters have a direct and intuitive interpretation in terms of the geometric features of the landscapes that they produce. An experimental space is defined over algorithms and problems, via a tuple of parameters for any specified algorithm and problem class (here determined by the landscape generator). An experiment is then clearly specified as a point in this space, in a way that is analogous to other areas of experimental algorithmics, and more generally in experimental design. Experimental results are presented, demonstrating the use of the landscape generator. In particular, we analyze some simple, continuous estimation of distribution algorithms, and gain new insights into the behavior of these algorithms using the landscape generator.

Choosing the best parameter setting is a well-known important and challenging task in evolutionary algorithms (EAs). As one of the earliest parameter tuning techniques, the meta-EA approach regards each parameter as a variable and the performance of algorithm as the fitness value and conducts searching on this landscape using various genetic operators. However, there are some inherent issues in this method. For example, some algorithm parameters are generally not searchable because it is difficult to define any sensible distance metric on them. In this paper, a novel approach is proposed by combining the meta-EA approach with a method called racing, which is based on the statistical analysis of algorithm performance with different parameter settings. A series of experiments are conducted to show the reliability and efficiency of this hybrid approach in tuning genetic algorithms (GAs) on two benchmark problems.

A comprehensive set of experiments was conducted with a continuous EDA on 25 test problems provided in the real-parameter optimization special session. It is expected that the results presented here could be used to gain some deeper understanding of the performance of the EDA as well as facilitate the comparison across different algorithms.

The development of Estimation of Distribution Algorithms (EDAs) has largely been driven by using more and more complex statistical models to approximate the structure of search space. However, there are still problems that are difficult for EDAs even with models capable of capturing high order dependences. In this paper, we show that diversity maintenance plays an important role in the performance of EDAs. A continuous EDA based on the Cholesky decomposition is tested on some well-known difficult benchmark problems to demonstrate how different diversity maintenance approaches could be applied to substantially improve its performance.

This paper introduces the design of superconductive magnet configurations in Magnetic Resonance Imaging (MRI) systems as a challenging real-world problem for Evolutionary Algorithms (EAs). Analysis of the problem structure is conducted using a general statistical method, which could be easily applied to other problems. The results suggest that the problem is highly multimodal and likely to present a significant challenge for many algorithms. Through a series of preliminary experiments, a continuous Estimation of Distribution Algorithm (EDA) is shown to be able to generate promising designs with a small computational effort. The importance of utilizing problem-specific knowledge and the ability of an algorithm to capture dependencies in solving complex real-world problems is also highlighted.

Evolutionary algorithms perform optimization using a population of sample solution points. An interesting development has been to view population-based optimization as the process of evolving an explicit, probabilistic model of the search space. This paper investigates a formal basis for continuous, population-based optimization in terms of a stochastic gradient descent on the Kullback-Leibler divergence between the model probability density and the objective function, represented as an unknown density of assumed form. This leads to an update rule that is related and compared with previous theoretical work, a continuous version of the population-based incremental learning algorithm, and the generalized mean shift clustering framework. Experimental results are presented that demonstrate the dynamics of the new algorithm on a set of simple test problems.

Pac-Man is a well-known, real-time computer game that provides an interesting platform for research. We describe an initial approach to developing an artificial agent that replaces the human to play a simplified version of Pac-Man. The agent is specified as a simple finite state machine and ruleset. with parameters that control the probability of movement by the agent given the constraints of the maze at some instant of time. In contrast to previous approaches, the agent represents a dynamic strategy for playing Pac-Man, rather than a pre-programmed maze-solving method. The agent adaptively "learns" through the application of population-based incremental learning (PBIL) to adjust the agents' parameters. Experimental results are presented that give insight into some of the complexities of the game, as well as highlighting the limitations and difficulties of the representation of the agent.

As an alternative to traditional evolutionary algorithms (EAs), population-based incremental learning (PBIL) maintains a probabilistic model of the best individual(s). Originally, PBIL was applied in binary search spaces. Recently, some work has been done to extend it to continuous spaces. In this paper, we review two such extensions of PBIL. An improved version of the PBIL based on Gaussian model is proposed that combines two main features: a new updating rule that takes into account all the individuals and their fitness values and a self-adaptive learning rate parameter. Furthermore, a new continuous PBIL employing a histogram probabilistic model is proposed. Some experiments results are presented that highlight the features of the new algorithms.

In this paper, we address some issue related to evaluating and testing evolutionary algorithms. A landscape generator based on Gaussian functions is proposed for generating a variety of continuous landscapes as fitness functions. Through some initial experiments, we illustrate the usefulness of this landscape generator in testing evolutionary algorithms.

This paper is concerned with the use of scientific visualization methods for the analysis of feedforward neural networks (NNs). Inevitably, the kinds of data associated with the design and implementation of neural networks are of very high dimensionality, presenting a major challenge for visualization. A method is described using the well-known statistical technique of principal component analysis (PCA). This is found to be an effective and useful method of visualizing the learning trajectories of many learning algorithms such as backpropagation and can also be used to provide insight into the learning process and the nature of the error surface.