He received an MSc degree in Electronics and Telecommunication (Specialization: Computer Control Systems) from the Silesian Technical University, Poland in 1994 and a PhD in Computer Science from the Nottingham Trent University, UK in 1998.
In January 2003, when he joined Bournemouth University, he was given an opportunity to form, build and lead a Computational Intelligence Research Group (CIRG) which had quickly grown and in Sep 2007 became a part of the Smart Technology Research Centre (STRC) which he also founded and led. Building on continued growth and success, in June 2014, he had initiated the creation of an interdisciplinary Data Science Institute at BU which he had also been asked to lead.
Immediately before moving to Sydney in September 2017, he held the positions of a Chair in Computational Intelligence (since 2005) and a Head of Data Science Institute (since 2014) at the Faculty of Science and Technology, Bournemouth University, UK.
Over the last 20 years, Prof. Gabrys has been working at various universities and research and development departments of commercial institutions based in four different countries, which allowed him to establish a wide network of links and collaborations with various commercial organizations and research groups within UK and abroad.
Prof. Gabrys has supervised a number of PhD research students and has been involved in various funded research projects. He has published over 140 research papers, reviewed for various journals, edited books and special issues of journals, chaired conferences, workshops and special sessions and been on programme committees of a large number of international conferences with the Data Science, Computational Intelligence, Machine Learning and Data Mining themes. He is frequently invited to give keynote and plenary talks at international conferences and lectures at internationally leading research centres and commercial research labs.
He is also the Chair, Academic Affairs of KES International, a Co-Editor in Chief of the International Journal of Knowledge Based & Intelligent Engineering Systems, and, among others, was a co-chair of a very successful KES'2006 conference.
Prof. Gabrys was a co-chairman of the Nature-inspired Data Technology focus group within the EU Coordination Action project on Nature-inspired Smart Information Systems (NiSIS). He was also a corresponding person for a Key Node in the European Network on Intelligent Technologies for Smart Adaptive Systems (EUNITE) and a co-chairman of the Research Theory & Development Group on Integration of Methods.
Can supervise: YES
Since accepting his first independent academic position Bogdan's research activities have concentrated on the areas of data science, complex adaptive systems, computational intelligence, machine learning, predictive analytics and their diverse applications. In particular, he has pursued the development of various statistical, machine learning, nature inspired and hybrid intelligent techniques especially targeting data and information fusion, learning and adaptive methods, multiple classifier and prediction systems, processing and modelling of uncertainty in pattern recognition, diagnostic analysis and decision support systems.
Between 2010 and 2014, Prof. Gabrys co-ordinated a large EU funded INFER project which brought together and integrated a number of previous research stream towards a unified framework for automated building and optimisation of predictive models as well as their continous adaptation in changing environments. This continues to be one of his major research pursuits.
I have been involved in teaching in higher education since 1993. I have experience in academic administration, teaching and curriculum development having acted in the past as a member of degree/framework validation panels, degree year leader, module leader and coordinator for undergraduate and Master’s courses and being a member of various academic groups for the undergraduate and postgraduate courses at different universities. I have been involved with the Institute for Learning and Teaching (ILT) in Higher Education, UK since its establishment in 2000 and my teaching experience and track record in education was recognised by the award of the Membership in ILTHE in the same year. Further evidence of excellence and sustained contribution to the higher education in UK was reinforced by the award of the Fellowship in the Higher Education Academy (successor of ILTHE) in 2007.
Apart from teaching in UK I have held various Visiting Professor positions and delivered courses in University of Burgos, Spain; Wroclaw University of Technology, Poland; and the European Centre for Soft Computing/University of Oviedo, Spain.
Since coming to Bournemouth University in 2003 I had been involved in teaching various Computational Intelligence related units on a number of courses at the School of Design Engineering and Computing including: MSc in Applied Artificial Intelligence, BSc (Hons) Computing: Artificial Intelligence, BSc (Hons) Psychology and Computing or MSc in Smart Systems and Technology.
I was the initiator and coordinator of the development of a cross-faculty MSc in Applied Data Analytics at BU which we developed in partnership with SAS Institute and which started in Autumn 2013. This course capitalised on the substantial expertise within the Data Science Institute at BU in the areas of Predictive Analytics, Data Mining, Computational Intelligence, Machine Learning and Big Data to educate a new generation of Data Scientists.
Król, D, Fay, D & Gabryś, B 2015, Propagation Phenomena in Real World Networks, Springer.
This book covers leading research on a wide spectrum of propagation phenomenon and the techniques currently used in its modelling, prediction, analysis and control.
Howlett, RJ & Gabrys, B 2013, InnovationKT-2012 Preface.
Gabrys, B 2006, Knowledge-Based Intelligent Information and Engineering Systems 10th International Conference, KES 2006, Bournemouth, UK, October 9-11 2006, Proceedings, Springer.
The three volume set LNAI 4251, LNAI 4252, and LNAI 4253 constitutes the refereed proceedings of the 10th International Conference on Knowledge-Based Intelligent Information and Engineering Systems, KES 2006, held in Bournemouth, UK, in ...
Gabrys, B, Leiviskä, K & Strackeljan, J 2006, Do Smart Adaptive Systems Exist? Best Practice for Selection and Combination of Intelligent Methods, Springer.
Over three years of various EUNITE activities focusing on issues of adaptation and intelligent behaviour of computing and engineering systems has also led to posing the question which is the title of this book: Do smart adaptive systems ...
Howlett, B, Gabrys, B & Jain, L 2006, Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics: Preface.
© 2019 Elsevier B.V. General fuzzy min-max (GFMM) neural network is a generalization of fuzzy neural networks formed by hyperbox fuzzy sets for classification and clustering problems. Two principle algorithms are deployed to train this type of neural network, i.e., incremental learning and agglomerative learning. This paper presents a comprehensive empirical study of performance influencing factors, advantages, and drawbacks of the general fuzzy min-max neural network on pattern classification problems. The subjects of this study include (1) the impact of maximum hyperbox size, (2) the influence of the similarity threshold and measures on the agglomerative learning algorithm, (3) the effect of data presentation order, (4) comparative performance evaluation of the GFMM with other types of fuzzy min-max neural networks and prevalent machine learning algorithms. The experimental results on benchmark datasets widely used in machine learning showed overall strong and weak points of the GFMM classifier. These outcomes also informed potential research directions for this class of machine learning algorithms in the future.
Salvador, MM, Budka, M & Gabrys, B 2019, 'Automatic Composition and Optimization of Multicomponent Predictive Systems With an Extended Auto-WEKA', IEEE Transactions on Automation Science and Engineering, vol. 16, no. 2, pp. 1-14.View/Download from: UTS OPUS or Publisher's site
IEEE Composition and parameterization of multicomponent predictive systems (MCPSs) consisting of chains of data transformation steps are a challenging task. Auto-WEKA is a tool to automate the combined algorithm selection and hyperparameter (CASH) optimization problem. In this paper, we extend the CASH problem and Auto-WEKA to support the MCPS, including preprocessing steps for both classification and regression tasks. We define the optimization problem in which the search space consists of suitably parameterized Petri nets forming the sought MCPS solutions. In the experimental analysis, we focus on examining the impact of considerably extending the search space (from approximately 22,000 to 812 billion possible combinations of methods and categorical hyperparameters). In a range of extensive experiments, three different optimization strategies are used to automatically compose MCPSs for 21 publicly available data sets. The diversity of the composed MCPSs found is an indication that fully and automatically exploiting different combinations of data cleaning and preprocessing techniques is possible and highly beneficial for different predictive models. We also present the results on seven data sets from real chemical production processes. Our findings can have a major impact on the development of high-quality predictive models as well as their maintenance and scalability aspects needed in modern applications and deployment scenarios.
Khuat, TT, Chen, F & Gabrys, B 2019, 'An Effective Multi-Resolution Hierarchical Granular Representation based Classifier using General Fuzzy Min-Max Neural Network', IEEE Transactions on Fuzzy Systems.View/Download from: UTS OPUS or Publisher's site
IEEE Motivated by the practical demands for simplification of data towards being consistent with human thinking and problem solving as well as tolerance of uncertainty, information granules are becoming important entities in data processing at different levels of data abstraction. This paper proposes a method to construct classifiers from multi-resolution hierarchical granular representations (MRHGRC) using hyperbox fuzzy sets. The proposed approach forms a series of granular inferences hierarchically through many levels of abstraction. An attractive characteristic of our classifier is that it can maintain a high accuracy in comparison to other fuzzy min-max models at a low degree of granularity based on reusing the knowledge learned from lower levels of abstraction. In addition, our approach can reduce the data size significantly as well as handle the uncertainty and incompleteness associated with data in real-world applications. The construction process of the classifier consists of two phases. The first phase is to formulate the model at the greatest level of granularity, while the later stage aims to reduce the complexity of the constructed model and deduce it from data at higher abstraction levels. Experimental analyses conducted comprehensively on both synthetic and real datasets indicated the efficiency of our method in terms of training time and predictive performance in comparison to other types of fuzzy min-max neural networks and common machine learning algorithms.
© 2018 The Author(s). There are many algorithms that can be used for the time-series forecasting problem, ranging from simple (e.g. Moving Average) to sophisticated Machine Learning approaches (e.g. Neural Networks). Most of these algorithms require a number of user-defined parameters to be specified, leading to exponential explosion of the space of potential solutionS. since the trial-and-error approach to finding a good algorithm for solving a given problem is typically intractable, reSearchers and practitioners need to resort to a more intelligent Search strategy, with one option being to constraint the Search space using past experience - an approach known as Meta-learning. Although potentially attractive, Meta-learning comes with its own challengeS. Gathering a sufficient number of Meta-examples, which in turn requires collecting and processing multiple datasets from each problem domain under consideration is perhaps the most prominent issue. In this paper, we are investigating the situations in which the use of additional data can improve performance of a Meta-learning System, with focus on cross-domain transfer of Meta-knowledge. A similarity-based cluster analysis of Meta-features has also been performed in an attempt to discover homogeneous groups of time-series with respect to Meta-learning performance. Although the experiments revealed limited room for improvement over the overall best base-learner, the Meta-learning approach turned out to be a safe choice, minimizing the risk of selecting the least appropriate base-learner.
Kendrick, L, Musial, K & Gabrys, B 2018, 'Change point detection in social networksCritical review with experiments', Computer Science Review, vol. 29, pp. 1-13.View/Download from: UTS OPUS or Publisher's site
© 2018 Elsevier Inc. Change point detection in social networks is an important element in developing the understanding of dynamic systems. This complex and growing area of research has no clear guidelines on what methods to use or in which circumstances. This paper critically discusses several possible network metrics to be used for a change point detection problem and conducts an experimental, comparative analysis using the Enron and MIT networks. Bayesian change point detection analysis is conducted on different global graph metrics (Size, Density, Average Clustering Coefficient, Average Shortest Path) as well as metrics derived from the Hierarchical and Block models (Entropy, Edge Probability, No. of Communities, Hierarchy Level Membership). The results produced the posterior probability of a change point at weekly time intervals that were analysed against ground truth change points using precision and recall measures. Results suggest that computationally heavy generative models offer only slightly better results compared to some of the global graph metrics. The simplest metrics used in the experiments, i.e. nodes and links numbers, are the recommended choice for detecting overall structural changes.
Qin, M, Lei, K, Gabrys, B & Musial-Gabrys, K 2018, 'Adaptive community detection incorporating topology and content in social networks✰', Knowledge-Based Systems, vol. 161, pp. 342-356.View/Download from: UTS OPUS or Publisher's site
© 2018 In social network analysis, community detection is a basic step to understand the structure and function of networks. Some conventional community detection methods may have limited performance because they merely focus on the networks’ topological structure. Besides topology, content information is another significant aspect of social networks. Although some state-of-the-art methods started to combine these two aspects of information for the sake of the improvement of community partitioning, they often assume that topology and content carry similar information. In fact, for some examples of social networks, the hidden characteristics of content may unexpectedly mismatch with topology. To better cope with such situations, we introduce a novel community detection method under the framework of non-negative matrix factorization (NMF). Our proposed method integrates topology as well as content of networks and has an adaptive parameter (with two variations) to effectively control the contribution of content with respect to the identified mismatch degree. Based on the disjoint community partition result, we also introduce an additional overlapping community discovery algorithm, so that our new method can meet the application requirements of both disjoint and overlapping community detection. The case study using real social networks shows that our new method can simultaneously obtain the community structures and their corresponding semantic description, which is helpful to understand the semantics of communities. Related performance evaluations on both artificial and real networks further indicate that our method outperforms some state-of-the-art methods while exhibiting more robust behavior when the mismatch between topology and content is observed.
© 2016 Elsevier Inc. This overview presents the current state-of-the-art of self-adaptive technologies within virtual reality (VR) training. Virtual reality training and assessment is increasingly used for five key areas: medical, industrial & commercial training, serious games, rehabilitation and remote training such as Massive Open Online Courses (MOOCs). Adaptation can be applied to five core technologies of VR including haptic devices, stereo graphics, adaptive content, assessment and autonomous agents. Automation of VR training can contribute to automation of actual procedures including remote and robotic assisted surgery which reduces injury and improves accuracy of the procedure. Automated haptic interaction can enable tele-presence and virtual artefact tactile interaction from either remote or simulated environments. Automation, machine learning and data driven features play an important role in providing trainee-specific individual adaptive training content. Data from trainee assessment can form an input to autonomous systems for customised training and automated difficulty levels to match individual requirements. Self-adaptive technology has been developed previously within individual technologies of VR training. One of the conclusions of this research is that while it does not exist, an enhanced portable framework is needed and it would be beneficial to combine automation of core technologies, producing a reusable automation framework for VR training.
Wang, W, Jiao, P, He, D, Jin, D, Pan, L & Gabrys, B 2016, 'Autonomous overlapping community detection in temporal networks: A dynamic Bayesian nonnegative matrix factorization approach', KNOWLEDGE-BASED SYSTEMS, vol. 110, pp. 121-134.View/Download from: Publisher's site
Community detection in complex networks is a fundamental data analysis task in various domains, and how to effectively find overlapping communities in real applications is still a challenge. In this work, we propose a new unified model and method for finding the best overlapping communities on the basis of the associated node and link partitions derived from the same framework. Specifically, we first describe a unified model that accommodates node and link communities (partitions) together, and then present a nonnegative matrix factorization method to learn the parameters of the model. Thereafter, we infer the overlapping communities based on the derived node and link communities, i.e., determine each overlapped community between the corresponding node and link community with a greedy optimization of a local community function conductance. Finally, we introduce a model selection method based on consensus clustering to determine the number of communities. We have evaluated our method on both synthetic and real-world networks with ground-truths, and compared it with seven state-of-the-art methods. The experimental results demonstrate the superior performance of our method over the competing ones in detecting overlapping communities for all analysed data sets. Improved performance is particularly pronounced in cases of more complicated networked community structures.
Arsene, CTC & Gabrys, B 2014, 'Mixed simulation-state estimation of water distribution systems based on a least squares loop flows state estimator', APPLIED MATHEMATICAL MODELLING, vol. 38, no. 2, pp. 599-619.View/Download from: Publisher's site
Apeh, E & Gabrys, B 2013, 'Detecting and Visualizing the Change in Classification of Customer Profiles based on Transactional Data', Evolving Systems, vol. 4, no. 1, pp. 27-42.View/Download from: Publisher's site
Customer transactions tend to change over time with changing customer behaviour patterns. Classifier models, however, are often designed to perform prediction on data which is assumed to be static. These classifier models thus deteriorate in performance over time when predicting in the context of evolving data. Robust adaptive classification models are therefore needed to detect and adjust to the kind of changes that are common in transactional data. This paper presents an investigation into using change mining to monitor the adaptive classification of customers based on their transactions through moving time windows. The classification performance of two-class decision tree ensembles built using the data binning process based on the number of items purchased was monitored over varying 3, 6, 9 and 12 months time windows. The changing class values of the customer profiles were analysed and described. Results from our experiments show that the proposed approach can be used for learning and adapting to changing customer profiles. © 2012 Springer-Verlag.
Arsene, CTC & Gabrys, B 2013, 'Probabilistic finite element predictions of the human lower limb model in total knee replacement', MEDICAL ENGINEERING & PHYSICS, vol. 35, no. 8, pp. 1116-1132.View/Download from: Publisher's site
Budka, M & Gabrys, B 2013, 'Density-Preserving Sampling: Robust and Efficient Alternative to Cross-Validation for Error Estimation', IEEE TRANSACTIONS ON NEURAL NETWORKS AND LEARNING SYSTEMS, vol. 24, no. 1, pp. 22-34.View/Download from: Publisher's site
Kadlec, P & Gabrys, B 2013, 'Erratum to Architecture for development of adaptive on-line prediction models(Memetic Comp., (2009), 1, (241-269), 10.1007/s12293-009-0017-8)', Memetic Computing, vol. 5, no. 1, p. 79.View/Download from: Publisher's site
Lemke, C, Riedel, S & Gabrys, B 2013, 'Evolving forecast combination structures for airline revenue management', Journal of Revenue and Pricing Management, vol. 12, no. 3, pp. 221-234.View/Download from: Publisher's site
Forecasting is at the heart of every revenue management system, providing necessary input to capacity control, pricing and overbooking functionalities. For airlines, the key to efficient capacity control is determining the time of when to restrict bookings in a lower-fare class to leave space for later booking high-fare customers. This work presents findings of a collaboration project between Bournemouth University and Lufthansa Systems AG, a company providing revenue management software for airline carriers. The main aim is to increase net booking forecast accuracy by modifying one of its components, the cancellation forecast. Complementing an available set of three traditional individual algorithms, an additional method is presented and added to the method pool. Furthermore, diversification of model parameters and level of learning is discussed to increase the number of individual forecasts even further. Finally, the evolution of forecast combination structures is investigated and shown to be beneficial on an airline data set. © 2013 Macmillan Publishers Ltd.
Stahl, F, Gabrys, B, Gaber, MM & Berendsen, M 2013, 'An overview of interactive visual data mining techniques for knowledge discovery', WILEY INTERDISCIPLINARY REVIEWS-DATA MINING AND KNOWLEDGE DISCOVERY, vol. 3, no. 4, pp. 239-256.View/Download from: Publisher's site
Arsene, CTC, Gabrys, B & Al-Dabass, D 2012, 'Decision support system for water distribution systems based on neural networks and graphs theory for leakage detection', EXPERT SYSTEMS WITH APPLICATIONS, vol. 39, no. 18, pp. 13214-13224.View/Download from: Publisher's site
Eastwood, M & Gabrys, B 2012, 'Generalised bottom-up pruning: A model level combination of decision trees', EXPERT SYSTEMS WITH APPLICATIONS, vol. 39, no. 10, pp. 9150-9158.View/Download from: Publisher's site
Tsakonas, A & Gabrys, B 2012, 'GRADIENT: Grammar-driven genetic programming framework for building multi-component, hierarchical predictive systems', EXPERT SYSTEMS WITH APPLICATIONS, vol. 39, no. 18, pp. 13253-13266.View/Download from: Publisher's site
Zliobaite, I, Bifet, A, Gaber, M, Gabrys, B, Gama, J, Minku, L & Musial, K 2012, 'Next challenges for adaptive learning systems', ACM SIGKDD Explorations Newsletter, vol. 14, no. 1, pp. 48-48.View/Download from: Publisher's site
Budka, M & Gabrys, B 2011, 'Electrostatic field framework for supervised and semi-supervised learning from incomplete data', NATURAL COMPUTING, vol. 10, no. 2, pp. 921-945.View/Download from: Publisher's site
Ruta, D, Gabrys, B & Lemke, C 2011, 'A Generic Multilevel Architecture for Time Series Prediction', IEEE TRANSACTIONS ON KNOWLEDGE AND DATA ENGINEERING, vol. 23, no. 3, pp. 350-359.View/Download from: Publisher's site
Budka, M, Gabrys, B & Musial, K 2011, 'On Accuracy of PDF Divergence Estimators and Their Applicability to Representative Data Sampling', ENTROPY, vol. 13, no. 7, pp. 1229-1266.View/Download from: Publisher's site
Budka, M & Gabrys, B 2009, 'Electrostatic field classifier for deficient data', Advances in Intelligent and Soft Computing, vol. 57, pp. 311-318.
© Springer-Verlag Berlin Heidelberg 2009. This paper investigates the suitability of recently developed models based on the physical field phenomena for classification of incomplete datasets. An original approach to exploiting incomplete training data with missing features and labels, involving extensive use of electrostatic charge analogy has been proposed. Classification of incomplete patterns has been investigated using a local dimensionality reduction technique, which aims at exploiting all available information rather than trying to estimate the missing values. The performance of all proposed methods has been tested on a number of benchmark datasets for a wide range of missing data scenarios and compared to the performance of some standard techniques.
This work presents an architecture for the development of on-line prediction models. The architecture defines unified modular environment based on three concepts from machine learning, these are: (i) ensemble methods, (ii) local learning, and (iii) meta learning. The three concepts are organised in a three layer hierarchy within the architecture. For the actual prediction making any data-driven predictive method such as artificial neural network, support vector machines, etc. can be implemented and plugged in. In addition to the predictive methods, data pre-processing methods can also be implemented as plug-ins. Models developed according to the architecture can be trained and operated in different modes. With regard to the training, the architecture supports the building of initial models based on a batch of training data, but if this data is not available the models can also be trained in incremental mode. In a scenario where correct target values are (occasionally) available during the run-time, the architecture supports life-long learning by providing several adaptation mechanisms across the three hierarchical levels. In order to demonstrate its practicality, we show how the issues of current soft sensor development and maintenance can be effectively dealt with by using the architecture as a construction plan for the development of adaptive soft sensing algorithms. © Springer-Verlag 2009.
Despite recent successes and advancements in artificial intelligence and machine learning, this domain remains under continuous challenge and guidance from phenomena and processes observed in natural world. Humans remain unsurpassed in their efficiency of dealing and learning from uncertain information coming in a variety of forms, whereas more and more robust learning and optimisation algorithms have their analytical engine built on the basis of some nature-inspired phenomena. Excellence of neural networks and kernel-based learning methods, an emergence of particle-, swarms-, and social behaviour-based optimisation methods are just few of many facts indicating a trend towards greater exploitation of nature inspired models and systems. This work intends to demonstrate how a simple concept of a physical field can be adopted to build a complete framework for supervised and unsupervised learning methodology. An inspiration for artificial learning has been found in the mechanics of physical fields found on both micro and macro scales. Exploiting the analogies between data and charged particles subjected to gravity, electrostatic and gas particle fields, a family of new algorithms has been developed and applied to classification, clustering and data condensation while properties of the field were further used in a unique visualisation of classification and classifier fusion models. The paper covers extensive pictorial examples and visual interpretations of the presented techniques along with some comparative testing over well-known real and artificial datasets. © Springer Science+Business Media B.V. 2007.
Eastwood, M & Gabrys, B 2007, 'The dynamics of negative correlation learning', JOURNAL OF VLSI SIGNAL PROCESSING SYSTEMS FOR SIGNAL IMAGE AND VIDEO TECHNOLOGY, vol. 49, no. 2, pp. 251-263.View/Download from: Publisher's site
Liu, H, Howlett, RJ & Gabrys, B 2007, 'Special issue: Extended papers selected from KES-2006', International Journal of Knowledge-Based and Intelligent Engineering Systems, vol. 11, no. 4, pp. 199-200.View/Download from: Publisher's site
Riedel, S, Gabrys, B & Riedel, S 2007, 'Combination of multi level forecasts', JOURNAL OF VLSI SIGNAL PROCESSING SYSTEMS FOR SIGNAL IMAGE AND VIDEO TECHNOLOGY, vol. 49, no. 2, pp. 265-280.View/Download from: Publisher's site
Individual classification models are recently challenged by combined pattern recognition systems, which often show better performance. In such systems the optimal set of classifiers is first selected and then combined by a specific fusion method. For a small number of classifiers optimal ensembles can be found exhaustively, but the burden of exponential complexity of such search limits its practical applicability for larger systems. As a result, simpler search algorithms and/or selection criteria are needed to reduce the complexity. This work provides a revision of the classifier selection methodology and evaluates the practical applicability of diversity measures in the context of combining classifiers by majority voting. A number of search algorithms are proposed and adjusted to work properly with a number of selection criteria including majority voting error and various diversity measures. Extensive experiments carried out with 15 classifiers on 27 datasets indicate inappropriateness of diversity measures used as selection criteria in favour of the direct combiner error based search. Furthermore, the results prompted a novel design of multiple classifier systems in which selection and fusion are recurrently applied to a population of best combinations of classifiers rather than the individual best. The improvement of the generalisation performance of such system is demonstrated experimentally. © 2004 Elsevier B.V. All rights reserved.
Gabrys, B 2002, 'Neuro-fuzzy approach to processing inputs with missing values in pattern recognition problems', INTERNATIONAL JOURNAL OF APPROXIMATE REASONING, vol. 30, no. 3, pp. 149-179.View/Download from: Publisher's site
Ruta, D & Gabrys, B 2002, 'A theoretical analysis of the limits of Majority Voting errors for Multiple Classifier Systems', PATTERN ANALYSIS AND APPLICATIONS, vol. 5, no. 4, pp. 333-350.View/Download from: Publisher's site
Gabrys, B & Bargiela, A 2000, 'General fuzzy min-max neural network for clustering and classification', IEEE TRANSACTIONS ON NEURAL NETWORKS, vol. 11, no. 3, pp. 769-783.View/Download from: Publisher's site
© 2019, Springer Nature Switzerland AG. In the last few years, we have witnessed a resurgence of interest in neural networks. The state-of-the-art deep neural network architectures are however challenging to design from scratch and requiring computationally costly empirical evaluations. Hence, there has been a lot of research effort dedicated to effective utilisation and adaptation of previously proposed architectures either by using transfer learning or by modifying the original architecture. The ultimate goal of designing a network architecture is to achieve the best possible accuracy for a given task or group of related tasks. Although there have been some efforts to automate network architecture design process, most of the existing solutions are still very computationally intensive. This work presents a framework to automatically find a good set of hyper-parameters resulting in reasonably good accuracy, which at the same time is less computationally expensive than the existing approaches. The idea presented here is to frame the hyper-parameter selection and tuning within the reinforcement learning regime. Thus, the parameters of a meta-learner, RNN, and hyper-parameters of the target network are tuned simultaneously. Our meta-learner is being updated using policy network and simultaneously generates a tuple of hyper-parameters which are utilized by another network. The network is trained on a given task for a number of steps and produces validation accuracy whose delta is used as reward. The reward along with the state of the network, comprising statistics of network’s final layer outcome and training loss, are fed back to the meta-learner which in turn generates a tuned tuple of hyper-parameters for the next time-step. Therefore, the effectiveness of a recommended tuple can be tested very quickly rather than waiting for the network to converge. This approach produces accuracy close to the state-of-the-art approach and is found to be comparatively less comput...
© 2019, Springer Nature Switzerland AG. This paper proposes an efficient domain adaption approach using deep learning along with transfer and meta-level learning. The objective is to identify how many blocks (i.e. groups of consecutive layers) of a pre-trained image classification network need to be fine-tuned based on the characteristics of the new task. In order to investigate it, a number of experiments have been conducted using different pre-trained networks and image datasets. The networks were fine-tuned, starting from the blocks containing the output layers and progressively moving towards the input layer, on various tasks with characteristics different from the original task. The amount of fine-tuning of a pre-trained network (i.e. the number of top layers requiring adaptation) is usually dependent on the complexity, size, and domain similarity of the original and new tasks. Considering these characteristics, a question arises of how many blocks of the network need to be fine-tuned to get maximum possible accuracy? Which of a number of available pre-trained networks require fine-tuning of the minimum number of blocks to achieve this accuracy? The experiments, that involve three network architectures each divided into 10 blocks on average and five datasets, empirically confirm the intuition that there exists a relationship between the similarity of the original and new tasks and the depth of network needed to fine-tune in order to achieve accuracy comparable with that of a model trained from scratch. Further analysis shows that the fine-tuning of the final top blocks of the network, which represent the high-level features, is sufficient in most of the cases. Moreover, we have empirically verified that less similar tasks require fine-tuning of deeper portions of the network, which however is still better than training a network from scratch.
Akbar, MS & Gabrys, B 2018, 'Data analytics enhanced data visualization and interrogation with parallel coordinates plots', 26th International Conference on Systems Engineering, ICSEng 2018 - Proceedings, International Conference on Systems Engineering, IEEE, Sydney, Australia, Australia.View/Download from: UTS OPUS or Publisher's site
© 2018 IEEE. Parallel coordinates plots (PCPs) suffer from curse of dimensionality when used with larger multidimensional datasets. Curse of dimentionality results in clutter which hides important visual data trends among coordinates. A number of solutions to address this problem have been proposed including filtering, aggregation, and dimension reordering. These solutions, however, have their own limitations with regard to exploring relationships and trends among the coordinates in PCPs. Correlation based coordinates reordering techniques are among the most popular and have been widely used in PCPs to reduce clutter, though based on the conducted experiments, this research has identified some of their limitations. To achieve better visualization with reduced clutter, we have proposed and evaluated dimensions reordering approach based on minimization of the number of crossing pairs. In the last step, k-means clustering is combined with reordered coordinates to highlight key trends and patterns. The conducted comparative analysis have shown that minimum crossings pairs approach performed much better than other applied techniques for coordinates reordering, and when combined with k-means clustering, resulted in better visualization with significantly reduced clutter.
Khuat, TT & Gabrys, B 2019, 'Accelerated Training Algorithms of General Fuzzy Min-Max Neural Network Using GPU for Very High Dimensional Data', Neural Information Processing (LNCS), International Conference on Neural Information Processing, Springer International Publishing, Sydney, Australia, pp. 583-595.View/Download from: UTS OPUS or Publisher's site
One of the issues of training a general fuzzy min-max neural network (GFMM) on very high dimensional data is a long training time even if the number of samples is relatively low. This is a quite common problem shared by many prototype-based methods requiring frequently repeated distance or similarity calculations. This paper proposes the method of accelerating the learning algorithms of the GFMM by, first, reformulating and representing them in a format allowing for their parallel execution and subsequently leveraging the computational power of the graphics processing unit (GPU). The original implementation of GFMM is modified by matrix computations to be executed on the GPU for the very high-dimensional datasets. The empirical results on two very high-dimensional datasets indicated that the training and testing processes performed on Nvidia Quadro P5000 GPU were from 10 to 35 times faster compared to those running serially on the Xeon CPU while retaining the same classification accuracy.
Jin, D, Liu, Z, He, D, Gabrys, B & Musial, K 2018, 'Robust detection of communities with multi-semantics in large attributed networks', Knowledge Science, Engineering and Management 11th International Conference, KSEM 2018 Changchun, China, August 17–19, 2018 Proceedings (LNAI 11061), International Conference on Knowledge Science, Engineering and Management, Springer, Changchun, China, pp. 362-376.View/Download from: UTS OPUS or Publisher's site
© 2018, Springer Nature Switzerland AG. In this paper, we are interested in how to explore and utilize the relationship between network communities and semantic topics in order to find the strong explanatory communities robustly. First, the relationship between communities and topics displays different situations. For example, from the viewpoint of semantic mapping, their relationship can be one-to-one, one-to-many or many-to-one. But from the standpoint of underlying community structures, the relationship can be consistent, partially consistent or completely inconsistent. Second, it will be helpful to not only find communities more precise but also reveal the communities’ semantics that shows the relationship between communities and topics. To better describe this relationship, we introduce the transition probability which is an important concept in Markov chain into a well-designed nonnegative matrix factorization framework. This new transition probability matrix with a suitable prior which plays the role of depicting the relationship between communities and topics can perform well in this task. To illustrate the effectiveness of the proposed new approach, we conduct some experiments on both synthetic and real networks. The results show that our new method is superior to baselines in accuracy. We finally conduct a case study analysis to validate the new method’s strong interpretability to detected communities.
Al-Jubouri, B & Gabrys, B 2017, 'Diversity and Locality in Multi-Component, Multi-Layer Predictive Systems: A Mutual Information Based Approach', ADVANCED DATA MINING AND APPLICATIONS, ADMA 2017, International Conference on Advanced Data Mining and Applications (ADMA), SPRINGER INTERNATIONAL PUBLISHING AG, Singapore, SINGAPORE, pp. 313-325.View/Download from: Publisher's site
Salvador, MM, Budka, M & Gabrys, B 2017, 'Modelling multi-component predictive systems as petri nets', 15th International Industrial Simulation Conference 2017, ISC 2017, pp. 17-23.
© 2017 Institute of Electrical and Electronics Engineers Inc.. All rights reserved. Building reliable data-driven predictive systems requires a considerable amount of human effort, especially in the data preparation and cleaning phase. In many application domains, multiple preprocessing steps need to be applied in sequence, constituting a 'workflow' and facilitating reproducibility. The concatenation of such workflow with a predictive model forms a Multi-Component Predictive System (MCPS). Automatic MCPS composition can speed up this process by taking the human out of the loop, at the cost of model transparency (i.e. not being comprehensible by human experts). In this paper, we adopt and suitably re-define the Well-handled with Regular Iterations Work Flow (WRI-WF) Petri nets to represent MCPSs. The use of such WRI-WF nets helps to increase the transparency of MCPSs required in industrial applications and make it possible to automatically verify the composed workflows. We also present our experience and results of applying this representation to model soft sensors in chemical production plants.
Gao, F, Musial, K & Gabrys, B 2017, 'A community bridge boosting social network link prediction model', Proceedings of the 2017 IEEE/ACM International Conference on Advances in Social Networks Analysis and Mining, ASONAM 2017, pp. 683-689.View/Download from: Publisher's site
© 2017 Association for Computing Machinery. Link prediction in social networks is a very challenging research problem. The majority of existing approaches are based on the assumption that a given network evolves following a single phenomenon, e.g.”rich get richer” or”friend of my friend is my friend”. However, dynamics of network dynamic changes over time and different parts of the network evolve in different manner. Because of that, we hypothesise that the prediction accuracy can be improved by providing different treatment to different nodes and links. Building on that assumption, we propose a Community Bridge Boosting Prediction Model (CBBPM) that treats certain bridge nodes differently depending on their structural position. For such bridge nodes their similarity score obtained using traditional link-based prediction methods is boosted. By doing so the importance of these nodes is increased and at the same time ensuring that the CBBPM can be used with any existing link prediction method. Our experimental results show that such bridge node similarity boosting mechanism can improve the accuracy of traditional link prediction methods.
Qin, M, Jin, D, He, D, Gabrys, B & Musial, K 2017, 'Adaptive community detection incorporating topology and content in social networks', Proceedings of the 2017 IEEE/ACM International Conference on Advances in Social Networks Analysis and Mining, ASONAM 2017, pp. 675-682.View/Download from: Publisher's site
© 2017 Association for Computing Machinery. In social network analysis, community detection is a basic step to understand the structure, function and semantics of networks. Some conventional community detection methods may have limited performance because they merely focus on topological structure of networks. In addition to topology, content information is another significant aspect of social networks. Some state-of-the-art methods started to combine these two aspects of information, but they often assume that topology and content share the same characteristics. However, for some examples of social networks, content may mismatch with topological structure. In order to better cope with such situations, we introduce a novel community detection method under the framework of nonnegative matrix factorization (NMF). Our proposed method integrates topology and content of networks, and introduces a novel adaptive parameter for controlling the contribution of content with respect to the identified mismatch degree between the topological and content information. The case study using real social networks show that our new method can simultaneously obtain community partition and the corresponding semantic descriptions. Experiments on both artificial networks and real social networks further indicate that our method outperforms some state-of-the-art methods while exhibiting more robust behaviour when the mismatch topological and content information is observed.
Al-Jubouri, B & Gabrys, B 2016, 'Local learning for multi-layer, multi-component predictive system', KNOWLEDGE-BASED AND INTELLIGENT INFORMATION & ENGINEERING SYSTEMS: PROCEEDINGS OF THE 20TH INTERNATIONAL CONFERENCE KES-2016, 20th International Conference on Knowledge - Based and Intelligent Information and Engineering Systems (KES), ELSEVIER SCIENCE BV, York, ENGLAND, pp. 723-732.View/Download from: Publisher's site
Bakirov, R, Gabrys, B & Fay, D 2016, 'Augmenting Adaptation with Retrospective Model Correction for Non-Stationary Regression Problems', 2016 INTERNATIONAL JOINT CONFERENCE ON NEURAL NETWORKS (IJCNN), International Joint Conference on Neural Networks (IJCNN), IEEE, Vancouver, CANADA, pp. 771-779.
He, H, Maple, C, Watson, T, Tiwari, A, Mehnen, J, Jin, Y & Gabrys, B 2016, 'The Security Challenges in the IoT enabled Cyber-Physical Systems and Opportunities for Evolutionary Computing & Other Computational Intelligence', 2016 IEEE CONGRESS ON EVOLUTIONARY COMPUTATION (CEC), IEEE Congress on Evolutionary Computation (CEC) held as part of IEEE World Congress on Computational Intelligence (IEEE WCCI), IEEE, Vancouver, CANADA, pp. 1015-1021.
He, H, Tiwari, A, Mehnen, J, Watson, T, Maple, C, Jin, Y & Gabrys, B 2016, 'Incremental Information Gain Analysis of Input Attribute Impact on RBF-Kernel SVM Spam Detection', 2016 IEEE CONGRESS ON EVOLUTIONARY COMPUTATION (CEC), IEEE Congress on Evolutionary Computation (CEC) held as part of IEEE World Congress on Computational Intelligence (IEEE WCCI), IEEE, Vancouver, CANADA, pp. 1022-1029.
Salvador, MM, Budka, M & Gabrys, B 2016, 'Effects of change propagation resulting from adaptive preprocessing in multicomponent predictive systems', KNOWLEDGE-BASED AND INTELLIGENT INFORMATION & ENGINEERING SYSTEMS: PROCEEDINGS OF THE 20TH INTERNATIONAL CONFERENCE KES-2016, 20th International Conference on Knowledge - Based and Intelligent Information and Engineering Systems (KES), ELSEVIER SCIENCE BV, York, ENGLAND, pp. 713-722.View/Download from: Publisher's site
Salvador, MM, Budka, M & Gabrys, B 2016, 'Towards Automatic Composition of Multicomponent Predictive Systems', HYBRID ARTIFICIAL INTELLIGENT SYSTEMS, 11th International Conference on Hybrid Artificial Intelligence Systems (HAIS), SPRINGER-VERLAG BERLIN, Seville, SPAIN, pp. 27-39.View/Download from: Publisher's site
Vaughan, N & Gabrys, B 2016, 'Comparing and combining time series trajectories using Dynamic Time Warping', KNOWLEDGE-BASED AND INTELLIGENT INFORMATION & ENGINEERING SYSTEMS: PROCEEDINGS OF THE 20TH INTERNATIONAL CONFERENCE KES-2016, 20th International Conference on Knowledge - Based and Intelligent Information and Engineering Systems (KES), ELSEVIER SCIENCE BV, York, ENGLAND, pp. 474-483.View/Download from: Publisher's site
Bakirov, R, Gabrys, B & Fay, D 2015, 'On sequences of different adaptive mechanisms in non-stationary regression problems', Proceedings of the International Joint Conference on Neural Networks.View/Download from: Publisher's site
© 2015 IEEE. Existing adaptive predictive methods often use multiple adaptive mechanisms as part of their coping strategy in non-stationary environments. These mechanisms are usually deployed in a prescribed order which does not change. In this work we investigate and provide a comparative analysis of the effects of using a flexible order of adaptive mechanisms' deployment resulting in varying adaptation sequences. As a vehicle for this comparison, we use an adaptive ensemble method for regression in batch learning mode which employs several adaptive mechanisms to react to the changes in data. Using real world data from the process industry we demonstrate that such flexible deployment of available adaptive methods embedded in a cross-validatory framework can benefit the predictive accuracy over time.
Al-Jubouri, B & Gabrys, B 2014, 'Multicriteria Approaches for Predictive Model Generation: A Comparative Experimental Study', 2014 IEEE SYMPOSIUM ON COMPUTATIONAL INTELLIGENCE IN MULTI-CRITERIA DECISION-MAKING (MCDM), 2014 IEEE Symposium on Computational Intelligence in Multi-Criteria Decision-Making (MCDM), IEEE, Orlando, FL, pp. 64-71.
Budka, M, Eastwood, M, Gabrys, B, Kadlec, P, Salvador, MM, Schwan, S, Tsakonas, A & Zliobaite, I 2014, 'From Sensor Readings to Predictions: On the Process of Developing Practical Soft Sensors', ADVANCES IN INTELLIGENT DATA ANALYSIS XIII, 13th International Symposium on Intelligent Data Analysis (IDA), SPRINGER INT PUBLISHING AG, Fac Club, Leuven, BELGIUM, pp. 49-60.
Gabrys, B 2013, 'Robust Adaptive Predictive Modeling and Data Deluge (Extended Abstract)', MAN-MACHINE INTERACTIONS 3, 3rd International Conference on Man-Machine Interactions (ICMMI), SPRINGER INT PUBLISHING AG, Brenna, POLAND, pp. 39-41.
Le, M, Nauck, D, Gabrys, B & Martin, T 2014, 'Sequential Clustering for Event Sequences and Its Impact on Next Process Step Prediction', INFORMATION PROCESSING AND MANAGEMENT OF UNCERTAINTY IN KNOWLEDGE-BASED SYSTEMS, PT I, 15th International Conference on Information Processing and Management of Uncertainty in Knowledge-based Systems (IPMU), SPRINGER-VERLAG BERLIN, Montpellier, FRANCE, pp. 168-178.
Nowak, P, Czeczot, J, Klopot, T, Szymura, M & Gabrys, B 2014, 'Linearizing controller for higher-degree nonlinear processes with compensation for modeling inaccuracies: Practical validation and future developments', ICINCO 2014 - Proceedings of the 11th International Conference on Informatics in Control, Automation and Robotics, pp. 691-698.
This work shows the results of the practical implementation of the linearizing controller for the example laboratory pneumatic process of the third relative degree. Controller design is based on the Lie algebra framework but in contrast to the previous attempts, the on-line model update method is suggested to ensure offset-free control. The paper details the proposed concept and reports the experiences from the practical implementation of the suggested controller. The superiority of the proposed approach over the conventional PI controller is demonstrated by experimental results. Based on the experiences and the validation results, the possibilities of the potential application of the data-driven soft sensors for further improvement of the control performance are discussed.
Salvador, MM, Gabrys, B & Zliobaite, I 2014, 'Online detection of shutdown periods in chemical plants: A case study', KNOWLEDGE-BASED AND INTELLIGENT INFORMATION & ENGINEERING SYSTEMS 18TH ANNUAL CONFERENCE, KES-2014, 18th Annual International Conference on Knowledge-Based and Intelligent Information and Engineering Systems (KES), ELSEVIER SCIENCE BV, Pomeranian Sci & Technol, Gdynia, POLAND, pp. 580-588.View/Download from: Publisher's site
Tsakonas, A & Gabrys, B 2014, 'Application of base learners as conditional input for fuzzy rule-based combined system', Studies in Computational Intelligence, pp. 19-32.View/Download from: Publisher's site
© Springer International Publishing Switzerland 2014. The aim of this work is to examine the possibility of using the output of base learners as antecedents for fuzzy rule-based hybrid ensembles. We select a flexible, grammar-driven framework for generating ensembles that combines multilayer perceptrons and support vector machines by means of genetic programming. We assess the proposed model in three real-world regression problems and we test it against multi-level, hierarchical ensembles. Our first results show that for a given large size of the base learners pool, the outputs of some of them can be useful in the antecedent parts to produce accurate ensembles, while at the same time other more accurate members of the same pool contribute in the consequent part.
Bakirov, R & Gabrys, B 2013, 'Investigation of Expert Addition Criteria for Dynamically Changing Online Ensemble Classifiers with Multiple Adaptive Mechanisms', ARTIFICIAL INTELLIGENCE APPLICATIONS AND INNOVATIONS, AIAI 2013, 9th IFIP WG 12.5 International Conference on Artificial Intelligence Applications and Innovations (AIAI), SPRINGER-VERLAG BERLIN, Paphos, CYPRUS, pp. 646-656.
Le, M, Nauck, D & Gabrys, B 2013, 'Sequential approaches for predicting business process outcome and process failure warning', CEUR Workshop Proceedings, pp. 1-15.
Large service companies like telecommunication businesses run complex customer service processes in order to provide communication services to their customers. The flawless execution of these processes is essential since customer service is an important differentiator for these companies. They must also be able to predict if processes will complete successfully or run into exceptions in order to intervene at the right time, pre-empt problems and maintain customer service. Business process data is sequential in nature and can be very diverse. Thus, there is a need for an efficient sequential forecasting methodology that can cope with the diversity of the business data. In response to these requirements, in this paper we propose an approach which is a combination of KNN (K nearest neighbour) and sequence alignment for predicting process outcome. The proposed approach exploits temporal categorical features of the extracted data to predict the process outcomes using sequence alignment technique, and also addresses the diversity aspect of the data by considering subsets of similar process sequences, based on KNNs. We have shown, via a set of experiments, that our model offers better results when compared with original KNNs and random guess-based methods. We also introduce a rule based technique based on GOSPADE, which detects the repetitions of individual tasks and investigates the relationship between them and the process failure. The results are demonstrated in a comprehensive performance study on real business process data sets.
Musial, K, Gabrys, B & Buczko, M 2013, 'What kind of network are you? - Using local and global characteristics in network categorisation tasks', 2013 IEEE/ACM INTERNATIONAL CONFERENCE ON ADVANCES IN SOCIAL NETWORKS ANALYSIS AND MINING (ASONAM), IEEE/ACM International Conference on Advances in Social Networks Analysis and Mining (ASONAM), IEEE, Niagara Falls, CANADA, pp. 1366-1373.
Apeh, E, Žliobaite, I, Pechenizkiy, M & Gabrys, B 2012, 'Predicting multi-class customer profiles based on transactions: A case study in food sales', Res. and Dev. in Intelligent Syst. XXIX: Incorporating Applications and Innovations in Intel. Sys. XX - AI 2012, 32nd SGAI Int. Conf. on Innovative Techniques and Applications of Artificial Intel., pp. 213-218.View/Download from: Publisher's site
Predicting the class of customer profiles is a key task in marketing, which enables businesses to approach the customers in a right way to satisfy the customer's evolving needs. However, due to costs, privacy and/or data protection, only the business' owned transactional data is typically available for constructing customer profiles. We present a new approach that is designed to efficiently and accurately handle the multi-class classification of customer profiles built using sparse and skewed transactional data. Our approach first bins the customer profiles on the basis of the number of items transacted. The discovered bins are then partitioned and prototypes within each of the discovered bins selected to build the multi-class classifier models. The results obtained from using four multi-class classifiers on real-world transactional data consistently show the critical numbers of items at which the predictive performance of customer profiles can be substantially improved. © Springer-Verlag London 2012.
Le, M, Gabrys, B & Nauck, D 2012, 'A hybrid model for business process event prediction', Res. and Dev. in Intelligent Syst. XXIX: Incorporating Applications and Innovations in Intel. Sys. XX - AI 2012, 32nd SGAI Int. Conf. on Innovative Techniques and Applications of Artificial Intel., pp. 179-192.View/Download from: Publisher's site
Process event prediction is the prediction of various properties of the remaining path of a process sequence or workflow. The prediction is based on the data extracted from a combination of historical (closed) and/or live (open) workflows (jobs or process instances). In real-world applications, the problem is compounded by the fact that the number of unique workflows (process prototypes) can be enormous, their occurrences can be limited, and a real process may deviate from the designed process when executed in real environment and under realistic constraints. It is necessary for an efficient predictor to be able to cope with the diverse characteristics of the data.We also have to ensure that useful process data is collected to build the appropriate predictive model. In this paper we propose an extension of Markov models for predicting the next step in a process instance.We have shown, via a set of experiments, that our model offers better results when compared to methods based on random guess, Markov models and Hidden Markov models. The data for our experiments comes from a real live process in a major telecommunication company. © Springer-Verlag London 2012.
Tsakonas, A & Gabrys, B 2012, 'Fuzzy base predictor outputs as conditional selectors for evolved combined prediction system', IJCCI 2012 - Proceedings of the 4th International Joint Conference on Computational Intelligence, pp. 34-41.
In this paper, we attempt to incorporate trained base learners outputs as inputs to the antecedent parts in fuzzy rule-based construction of hybrid ensembles. To accomplish this we adopt a versatile framework for the production of ensemble systems that uses a grammar driven genetic programming to evolve combinations of multilayer perceptrons and support vector machines. We evaluate the proposed architecture using three realworld regression tasks and compare it with multi-level, hierarchical ensembles. The conducted preliminary experiments showed very interesting results indicating that given a large pool of base predictors to choose from, the outputs of some of them, when applied to fuzzy sets, can be used as selectors for building accurate ensembles from other more accurate and complementary members of the same base predictor pool.
Apeh, E & Gabrys, B 2011, 'Change mining of customer profiles based on transactional data', Proceedings - IEEE International Conference on Data Mining, ICDM, pp. 560-567.View/Download from: Publisher's site
Customer transactions tend to change overtime with changing customer behaviour patterns. Classifier models, however, are often designed to perform prediction on data which is assumed to be static. These classifier models thus deteriorate in performance overtime when predicting in the context of evolving data. Robust adaptive classification models are therefore needed to detect and adjust to the kind of changes that are common in transactional data. This paper presents an investigation into using change mining to monitor the adaptive classification of customers based on their transactions through a moving time window. Results from our experiments show that our approach can be used for learning and adapting to changing customer profiles. © 2011 IEEE.
Apeh, ET, Gabrys, B & Schierz, A 2011, 'Customer profile classification using transactional data', Proceedings of the 2011 3rd World Congress on Nature and Biologically Inspired Computing, NaBIC 2011, pp. 37-43.View/Download from: Publisher's site
Customer profiles are by definition made up of factual and transactional data. It is often the case that due to reasons such as high cost of data acquisition and/or protection, only the transactional data are available for data mining operations. Transactional data, however, tend to be highly sparse and skewed due to a large proportion of customers engaging in very few transactions. This can result in a bias in the prediction accuracy of classifiers built using them towards the larger proportion of customers with fewer transactions. This paper investigates an approach for accurately and confidently grouping and classifying customers in bins on the basis of the number of their transactions. The experiments we conducted on a highly sparse and skewed real-world transactional data show that our proposed approach can be used to identify a critical point at which customer profiles can be more confidently distinguished. © 2011 IEEE.
Eastwood, M & Gabrys, B 2011, 'Model level combination of tree ensemble hyperboxes via GFMM', Proceedings - 2011 8th International Conference on Fuzzy Systems and Knowledge Discovery, FSKD 2011, pp. 443-447.View/Download from: Publisher's site
An ensemble of decision trees defines an overlapping set of hyperboxes. These hyperboxes in turn define a disjoint set of hyperboxes each with an associated vector of individual decisions. These vectors can be used to robustly label the hyperboxes by class, or to define soft labels. We sample from these hyperboxes and use them to build a single classifier within the General Fuzzy Min-Max (GFMM) framework that gains information from many different resamplings of the data through the ensemble from which it is built. This method is found to build robust GFMM models, with improved performance on most datasets compared to the basic GFMM. © 2011 IEEE.
Khan, L, Pechenizkiy, M, Zliobaite, I, Agrawal, C, Bifet, A, Delany, SJ, Dries, A, Fan, W, Gabrys, B, Gama, J, Gao, J, Gopalkrishnan, V, Holmes, G, Katakis, I, Kuncheva, L, Van Leeuwen, M, Masud, M, Menasalvas, E, Minku, L, Pfahringer, B, Polikar, R, Rodrigues, PP, Tsoumakas, G & Tsymbal, A 2011, 'Preface', Proceedings - IEEE International Conference on Data Mining, ICDM.View/Download from: Publisher's site
Tsakonas, A & Gabrys, B 2011, 'Evolving Takagi-Sugeno-Kang fuzzy systems using multi population grammar-guided genetic programming', ECTA 2011 FCTA 2011 - Proceedings of the International Conference on Evolutionary Computation Theory and Applications and International Conference on Fuzzy Computation Theory and Applications, pp. 278-281.
This work proposes a novel approach for the automatic generation and tuning of complete Takagi-Sugeno-Kang fuzzy rule based systems. The examined system aims to explore the effects of a reduced search space for a genetic programming framework by means of grammar guidance that describes candidate structures of fuzzy rule based systems. The presented approach applies context-free grammars to generate individuals and evolve solutions through the search process of the algorithm. A multi-population approach is adopted for the genetic programming system, in order to increase the depth of the search process. Two candidate grammars are examined in one regression problem and one system identification task. Preliminary results are included and discussion proposes further research directions.
Ahmad, M, Alexandrou, I, Al-Nuaimy, W, Amavasai, BP, An, YY, Ariwa, E, Arteche, J, Audrino, F, Ayesh, A, Baber, C, Bailey, C, Balkan, N, Barria, J, Bartosova, J, Benkrid, K, Bleijs, H, Bluck, M, Bose, I, Bouzas, PR, Braiden, PM, Brdys, M, Burriesci, G, Cannataro, M, Carvalho, A, Chang, CC, Chen, D, Chen, GG, Chen, YS, Chiclana, F, Cooke, A, Das, DB, Davis, DN, Dayoub, I, Deb, S, Demetriou, IC, Devai, F, Dilmaghani, RS, Dini, D, Drikakis, D, Durkan, C, Durodola, J, Etebar, K, Fenn, P, Figueiredo, A, Florou, G, Freear, S, Gabrys, B, Galbraith, GH, Gaskell, PH, Gaura, E, Ge, ZQ, Ghafouri-Shiraz, H, Ghavami, M, Giannopoulos, K, Gonzalez, REP, Gracia, AM, Grecos, C, Guan, L, Gulpinar, N, Guo, R, Guo, Y, Hardalupas, Y, He, L, Herrero, JR, Hicks, BJ, Hines, EL, Hodgson, S, Horsfall, A, Hosein, P, Hu, F, Hu, H, Ijomah, W, Ming, J, James, A, Jancovic, P, Jhumka, A, Kamareddine, F, Kannan, R, Karsligil, ME, Katircioglu, ST, Khalid, A, Kokossis, A, Kontis, K, Kulekci, MO, Laukaitis, A, Leeson, M, Limbachiya, MC, Li, L, Li, L, Lin, P, Ling, WK, Macias Lopez, EM, Lovas, T, Luglio, M, Mainardi, S, Mahanti, PK, Marinos, I, Maropoulos, P, Mativenga, P & Mavrommatis, G 2010, 'WCE 2010 - World Congress on Engineering 2010: Preface', WCE 2010 - World Congress on Engineering 2010.
Budka, M & Gabrys, B 2010, 'Correntropy-based density-preserving data sampling as an alternative to standard cross-validation', Proceedings of the International Joint Conference on Neural Networks.View/Download from: Publisher's site
Estimation of the generalization ability of a predictive model is an important issue, as it indicates expected performance on previously unseen data and is also used for model selection. Currently used generalization error estimation procedures like cross-validation (CV) or bootstrap are stochastic and thus require multiple repetitions in order to produce reliable results, which can be computationally expensive if not prohibitive. The correntropy-based Density Preserving Sampling procedure (DPS) proposed in this paper eliminates the need for repeating the error estimation procedure by dividing the available data into subsets, which are guaranteed to be representative of the input dataset. This allows to produce low variance error estimates with accuracy comparable to 10 times repeated cross-validation at a fraction of computations required by CV, which has been investigated using a set of publicly available benchmark datasets and standard classifiers. © 2010 IEEE.
Budka, M & Gabrys, B 2010, 'Ridge regression ensemble for toxicity prediction', ICCS 2010 - INTERNATIONAL CONFERENCE ON COMPUTATIONAL SCIENCE, PROCEEDINGS, International Conference on Computational Science (ICCS), ELSEVIER SCIENCE BV, Univ Amsterdam, Amsterdam, NETHERLANDS, pp. 193-201.View/Download from: Publisher's site
Kadlec, P & Gabrys, B 2010, 'Adaptive on-line prediction soft sensing without historical data', Proceedings of the International Joint Conference on Neural Networks.View/Download from: Publisher's site
Current soft sensing algorithms assume the availability of a large amount of training data. The collection of the historical data often takes a lot of time and can be expensive. At the same time not being able to provide sufficient amount of training data can result in sacrificing the performance of the soft sensor. This can be problematic in situations, where a soft sensor is urgently required and, at the same time, there is not enough training data available. This situation can occur, for example, when a new plant is taken into operation or, more critically, when there is a significant change in some parameters (e.g. operating point or the input materials) in a running plant. To deal with such a situation, we propose an algorithm, called Recursive Soft Sensing Algorithm (ReSSA), which delivers predictions without any explicit training phase. The proposed algorithm is based on the recursive functionality of the RPLS technique, which is embedded into local learning framework. More than that, during the run-time of the algorithm, it is not necessary to store any past data as the algorithm requires only the latest data point for its operation and recursive adaptation. In order to demonstrate the performance of the proposed method, it is applied to the prediction of a catalyst activity in a multi-tube reactor. © 2010 IEEE.
Korsunsky, AM, Hunter, A, Hukins, DWL, Gelman, L, Hogger, CJ, Ceglarek, DJ, Payne, S, Ao, SI, Ahmad, M, Alexandrou, I, Al-Nuaimy, W, Amavasai, BP, An, YY, Ariwa, E, Arteche, J, Audrino, F, Ayesh, A, Baber, C, Bailey, C, Balkan, N, Barria, J, Bartosova, J, Benkrid, K, Bleijs, H, Bluck, M, Bose, I, Bouzas, PR, Braiden, PM, Brdys, M, Burriesci, G, Cannataro, M, Carvalho, A, Chang, CC, Chen, D, Chen, GG, Chen, YS, Chiclana, F, Cooke, A, Das, DB, Davis, DN, Dayoub, I, Raman, SDCV, Demetriou, IC, Devai, F, Dilmaghani, RS, Dini, D, Drikakis, D, Durkan, C, Durodola, J, Etebar, K, Fenn, P, Figueiredo, A, Florou, G, Freear, S, Gabrys, B, Galbraith, GH, Gaskell, PH, Gaura, E, Ge, ZQ, Ghafouri-Shiraz, H, Ghavami, M, Giannopoulos, K, Pruneda Gonzalez, RE, Gracia, AM, Grecos, C, Guan, L, Gulpinar, N, Guo, R, Guo, Y, Hardalupas, Y, He, L, Herrero, JR, Hicks, BJ, Hines, EL, Hodgson, S, Horsfall, A, Hosein, P, Hu, F, Hu, O, Ijomah, W, Ming, J, James, A, Jancovic, P, Jhumka, A, Kamareddine, F, Kannan, R, Karsligil, ME, Katircioglu, ST, Khalid, A, Kokossis, A, Kontis, K, Kulekci, MO, Laukaitis, A, Leeson, M, Limbachiya, MC, Li, L, Li, L, Lin, P, Ling, WK & Macias Lopez, EM 2010, 'WCE 2010 - World Congress on Engineering 2010: Preface', WCE 2010 - World Congress on Engineering 2010.
Lemke, C & Gabrys, B 2010, 'Meta-learning for time series forecasting in the NN GC1 competition', 2010 IEEE World Congress on Computational Intelligence, WCCI 2010.View/Download from: Publisher's site
There are no algorithms that generally perform better or worse than random when looking at all possible data sets according to the no-free-lunch theorem. A specific forecasting method will hence naturally have different performances in different empirical studies. This makes it impossible to draw general conclusions, however, there will of course be specific problems for which one algorithm performs better than another in practice. Meta-learning exploits this fact by linking characteristics of the data set to the performances of methods, adapting the selection or combination of base methods to a specific problem. This contribution describes an approach using meta-learning for time series forecasting in the NN GC1 competition. In order to generate bigger and more reliable meta-data set, data of the past NN3 and NN5 competitions have been included. A pool of individual forecasting and combination models are combined using a ranking algorithm with weights being determined by past performance on similar series. © 2010 IEEE.
Eastwood, M & Gabrys, B 2009, 'A Non-sequential Representation of Sequential Data for Churn Prediction', KNOWLEDGE-BASED AND INTELLIGENT INFORMATION AND ENGINEERING SYSTEMS, PT I, PROCEEDINGS, 13th International Conference on Knowledge-Based Intelligent Information and Engineering Systems, SPRINGER-VERLAG BERLIN, Univ Chile, Fac Phys Sci & Math, Santiago, CHILE, pp. 209-218.
Gabrys, B 2009, 'Learning with Missing or Incomplete Data', IMAGE ANALYSIS AND PROCESSING - ICIAP 2009, PROCEEDINGS, 15th International Conference on Image Analysis and Processing (ICIAP 2009), SPRINGER-VERLAG BERLIN, Vietri sul Mare, ITALY, pp. 1-4.
Kadlec, P & Gabrys, B 2009, 'Evolving on-line prediction model dealing with industrial data sets', 2009 IEEE WORKSHOP ON EVOLVING AND SELF-DEVELOPING INTELLIGENT SYSTEMS, IEEE Workshop on Evolving and Self-Developing Intelligent Systems, IEEE, Nashville, TN, pp. 24-31.View/Download from: Publisher's site
Kadlec, P & Gabrys, B 2009, 'Soft sensor based on adaptive local learning', Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics), pp. 1172-1179.View/Download from: Publisher's site
When it comes to application of computational learning techniques in practical scenarios, like for example adaptive inferential control, it is often difficult to apply the state-of-the-art techniques in a straight forward manner and usually some effort has to be dedicated to tuning either the data, in a form of data pre-processing, or the modelling techniques, in form of optimal parameter search or modification of the training algorithm. In this work we present a robust approach to on-line predictive modelling which is focusing on dealing with challenges like noisy data, data outliers and in particular drifting data which are often present in industrial data sets. The approach is based on the local learning approach, where models of limited complexity focus on partitions of the input space and on an ensemble building technique which combines the predictions of the particular local models into the final predicted value. Furthermore, the technique provides the means for on-line adaptation and can thus be deployed in a dynamic environment which is demonstrated in this work in terms of an application of the presented approach to a raw industrial data set exhibiting drifting data, outliers, missing values and measurement noise. © 2009 Springer Berlin Heidelberg.
In this work we present a summary of the review on data-driven soft sensors published in  together with a proposal of how to deal with the identified issues and challenges. We discuss the most common approaches for the development of soft sensors followed by a critical analysis of the main issues in the current soft sensor development. Currently, these are the time which has to be spent on the model development including data pre-processing and model building together with the effort which needs to be spent on periodical performance assessment and re-training of the model. Based on the identified problems we propose a solution based on a model development architecture which can accommodate different data preprocessing techniques, predictive modelling methods as well as approaches for model adaptation. The architecture is based on a structure which unifies several concepts from machine learning such as ensemble methods, local learning, meta learning and concept drift handling. Using the above mechanisms it provides means for automated data pre-processing, model validation, selection and adaptation which can be used to significantly simplify the soft sensor building and maintenance process. Copyright © 2007 International Federation of Automatic Control.
Lemke, C, Riedel, S & Gabrys, B 2009, 'Dynamic combination of forecasts generated by diversification procedures applied to forecasting of airline cancellations', IEEE/IAFE Conference on Computational Intelligence for Financial Engineering, Proceedings (CIFEr), pp. 85-91.View/Download from: Publisher's site
The combination of forecasts is a well established procedure for improving forecast performance and decreasing the risk of selecting an inferior model out of an existing pool of models. Work in this area mainly focuses on combining several functionally different models, but some publications also deal with combining forecasts with the same functional approach. In the latter case, individual forecasts are generated by diversifying one or more model parameters or, if dealing with hierarchical data, by using forecasts from different levels. This work looks at multi-dimensional data from airline industry, with the aim of improving the forecast of cancellation rates for bookings. Three different methods are employed for the generation of individual forecasts. Forecast combinations are usually implemented in a more or less static structure, either including all available forecasts or trimming a fixed percentage of the worst performing models. For a big number of individual forecasts, this procedure can become inefficient. In this paper, a dynamic approach of pooling and trimming is applied to the generated forecasts for airline cancellation data. © 2009 IEEE.
Musial, K, Juszczyszyn, K, Gabrys, B & Kazienko, P 2008, 'Patterns of Interactions in Complex Social Networks Based on Coloured Motifs Analysis', ADVANCES IN NEURO-INFORMATION PROCESSING, PT II, 15th International Conference on Neuro-Information Processing, SPRINGER-VERLAG BERLIN, Auckland, NEW ZEALAND, pp. 607-+.
Juszczyszyn, K, Kazienko, P, Musiai, K & Gabrys, B 2008, 'Temporal changes in connection patterns of an email-based social network', Proceedings - 2008 IEEE/WIC/ACM International Conference on Web Intelligence and Intelligent Agent Technology - Workshops, WI-IAT Workshops 2008, pp. 9-12.View/Download from: Publisher's site
Network motifs are small subgraphs that reflect local network topology and were shown to be useful for creating profiles that reveal several properties of the network. Analysis of three-node motifs (triads) was used in this paper to track the temporal changes in the structure of large social network derived from email communication between the employees of Wroclaw University of Technology. © 2008 IEEE.
Kadlec, P & Gabrys, B 2008, 'Adaptive Local Learning Soft Sensor for Inferential Control Support', 2008 INTERNATIONAL CONFERENCE ON COMPUTATIONAL INTELLIGENCE FOR MODELLING CONTROL & AUTOMATION, VOLS 1 AND 2, International Conference on Computational Intelligence for Modelling, Control and Automation, IEEE, Vienna, AUSTRIA, pp. 243-248.View/Download from: Publisher's site
Kadlec, P & Gabrys, B 2008, 'Gating Artificial Neural Network Based Soft Sensor', NEW CHALLENGES IN APPLIED INTELLIGENCE TECHNOLOGIES, 21st International Conference on Industrial, Engineering and Other Applications of Applied Intelligent Systems, SPRINGER, Wroclaw, POLAND, pp. 193-202.
Kadlec, P & Gabrys, B 2008, 'Learnt Topology Gating Artificial Neural Networks', 2008 IEEE INTERNATIONAL JOINT CONFERENCE ON NEURAL NETWORKS, VOLS 1-8, International Joint Conference on Neural Networks, IEEE, Hong Kong, PEOPLES R CHINA, pp. 2604-2611.View/Download from: Publisher's site
Lemke, C & Gabrys, B 2008, 'Do we need experts for time series forecasting?', ESANN 2008 Proceedings, 16th European Symposium on Artificial Neural Networks - Advances in Computational Intelligence and Learning, pp. 253-258.
This study examines a selection of off-the-shelf forecasting and forecast combination algorithms with a focus on assessing their practical relevance by drawing conclusions for non-expert users. Some of the methods have only recently been introduced and have not been part in comparative empirical evaluations before. Considering the advances of forecasting techniques, this analysis addresses the question whether we need human expertise for forecasting or whether the investigated methods provide comparable performance.
Incremental learning (IL) plays a key role in many real-world applications where data arrives over time. It is mainly concerned with learning models in an everchanging environment. In this paper, we review some of the incremental learning algorithms and evaluate them within the same experimental settings in order to provide as objective comparative study as possible. These algorithms include fuzzy ARTMAP, nearest generalized exemplar, growing neural gas, generalized fuzzy min-max neural network, and IL based on function decomposition (ILFD). © 2007 IEEE.
Macas, M, Gabrys, B, Ruta, D & Lhotska, L 2007, 'Particle swarm optimisation of multiple classifier systems', COMPUTATIONAL AND AMBIENT INTELLIGENCE, 9th International Work-Conference on Artificial Neural Networks, SPRINGER-VERLAG BERLIN, San Sebastian, SPAIN, pp. 333-+.
Riedel, S & Gabrys, B 2007, 'Dynamic pooling for the combination of forecasts generated using multi level learning', 2007 IEEE INTERNATIONAL JOINT CONFERENCE ON NEURAL NETWORKS, VOLS 1-6, International Joint Conference on Neural Networks, IEEE, Orlando, FL, pp. 454-+.View/Download from: Publisher's site
Ruta, D & Gabrys, B 2007, 'Neural network ensembles for time series prediction', 2007 IEEE INTERNATIONAL JOINT CONFERENCE ON NEURAL NETWORKS, VOLS 1-6, International Joint Conference on Neural Networks, IEEE, Orlando, FL, pp. 1204-1209.View/Download from: Publisher's site
Ruta, D & Gabrys, B 2007, 'Reducing spatial data complexity for classification models', COMPUTATIONAL METHODS IN SCIENCE AND ENGINEERING VOL 1, International Conference on Computational Methods in Science and Engineering, AMER INST PHYSICS, Corfu, GREECE, pp. 603-613.
Sahel, Z, Bouchachia, A, Gabrys, B & Rogers, P 2007, 'Adaptive mechanisms for classification problems with drifting data', KNOWLEDGE-BASED INTELLIGENT INFORMATION AND ENGINEERING SYSTEMS: KES 2007 - WIRN 2007, PT II, PROCEEDINGS, 11th International Conference on Knowledge-Based Intelligent Informational and Engineering Systems/17th Italian Workshop on Neural Networks, SPRINGER-VERLAG BERLIN, Vietri sul Mare, ITALY, pp. 419-426.
Apeh, ET & Gabrys, B 2006, 'Clustering for data matching', KNOWLEDGE-BASED INTELLIGENT INFORMATION AND ENGINEERING SYSTEMS, PT 1, PROCEEDINGS, 10th International Conference on Knowledge-Based and Intelligent Information and Engineering Systems, SPRINGER-VERLAG BERLIN, Bournemouth, ENGLAND, pp. 1216-1225.
Corchado, E, Baruque, B & Gabrys, B 2006, 'Maximum likelihood topology preserving ensembles', INTELLIGENT DATA ENGINEERING AND AUTOMATED LEARNING - IDEAL 2006, PROCEEDINGS, 7th International Conference on Intelligent Data Engineering and Automated Learning (IDEAL 2006), SPRINGER-VERLAG BERLIN, Univ Burgos, Burgos, SPAIN, pp. 1434-1442.
Gabrys, B, Baruque, B & Corchado, E 2006, 'Outlier resistant PCA ensembles', KNOWLEDGE-BASED INTELLIGENT INFORMATION AND ENGINEERING SYSTEMS, PT 3, PROCEEDINGS, 10th International Conference on Knowledge-Based and Intelligent Information and Engineering Systems, SPRINGER-VERLAG BERLIN, Bournemouth, ENGLAND, pp. 432-440.
Riedel, S & Gabrys, B 2004, 'Hierarchical Multilevel Approaches of Forecast Combination', OPERATIONS RESEARCH PROCEEDINGS 2004, Annual International Conference of the German-Operations-Research-Society, SPRINGER-VERLAG BERLIN, Tilburg, NETHERLANDS, pp. 479-486.
Gabrys, B 2001, 'Learning hybrid neuro-fuzzy classifier models from data: to combine or not to combine?', FUZZY SETS AND SYSTEMS, International Workshop on Hybrid Methods for Adaptive Systems held at EUNITE 2001, ELSEVIER SCIENCE BV, Puerto de la Cruz, SPAIN, pp. 39-56.View/Download from: Publisher's site
Gabrys, B & Petrakieva, L 1970, 'Combining labelled and unlabelled data in the design of pattern classification systems', INTERNATIONAL JOURNAL OF APPROXIMATE REASONING, Workshop on Hybrid Methods for Adaptive Systems, ELSEVIER SCIENCE INC, Albufeira, PORTUGAL, pp. 251-273.View/Download from: Publisher's site
Petrakieva, L & Gabrys, B 2002, 'Selective sampling for combined learning from labelled and unlabelled data', APPLICATIONS AND SCIENCE IN SOFT COMPUTING, 4th International Conference on Recent Advances in Soft Computing, SPRINGER-VERLAG BERLIN, NOTTINGHAM TRENT UNIV, NOTTINGHAM, ENGLAND, pp. 139-146.
Ruta, D & Gabrys, B 2002, 'Physical field models for pattern classification', SOFT COMPUTING, 1st International Conference on Computing in an Imperfect World (SOFT-WARE 2002), SPRINGER-VERLAG, BELFAST, NORTH IRELAND, pp. 126-141.View/Download from: Publisher's site
Gabrys, B 2000, 'Agglomerative learning algorithms for general fuzzy min-max neural network', JOURNAL OF VLSI SIGNAL PROCESSING SYSTEMS FOR SIGNAL IMAGE AND VIDEO TECHNOLOGY, 10th IEEE Workshop on Neural Networks for Signal Processing (NNSP 2000), KLUWER ACADEMIC PUBL, UNIV SYDNEY, SYDNEY, AUSTRALIA, pp. 67-82.View/Download from: Publisher's site
Gabrys, B 2002, 'Combining neuro-fuzzy classifiers for improved generalisation and reliability', PROCEEDING OF THE 2002 INTERNATIONAL JOINT CONFERENCE ON NEURAL NETWORKS, VOLS 1-3, International Joint Conference on Neural Networks (IJCNN 02), IEEE, HONOLULU, HI, pp. 2410-2415.View/Download from: Publisher's site
Ruta, D & Gabrys, B 2002, 'New measure of classifier dependency in multiple classifier systems', MULTIPLE CLASSIFIER SYSTEMS, 3rd International Workshop on Multiple Classifier Systems, SPRINGER-VERLAG BERLIN, CAGLIARI, ITALY, pp. 127-136.
Ruta, D & Gabrys, B 2002, 'Static field approach for pattern classification', SOFT-WARE 2002: COMPUTING IN AN IMPERFECT WORLD, 1st International Conference on Computing in an Imperfect World (SOFT-WARE 2002), SPRINGER-VERLAG BERLIN, BELFAST, NORTH IRELAND, pp. 232-246.
Ruta, D & Gabrys, B 2001, 'Application of the evolutionary algorithms for classifier selection in multiple classifier systems with majority voting', Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics), pp. 399-408.
© Springer-Verlag Berlin Heidelberg 2001. In many pattern recognition tasks, an approach based on combining classifiers has shown a significant potential gain in comparison to the performance of an individual best classifier. This improvement turned out to be subject to a sufficient level of diversity exhibited among classifiers, which in general can be assumed as a selective property of classifier subsets. Given a large number of classifiers, an intelligent classifier selection process becomes a crucial issue of multiple classifier system design. In this paper, we have investigated three evolutionary optimization methods for the classifier selection task. Based on our previous studies of various diversity measures and their correlation with majority voting error we have adopted majority voting performance computed for the validation set directly as a fitness function guiding the search. To prevent from training data overfitting we extracted a population of best unique classifier combinations, and used them for second stage majority voting. In this work we intend to show empirically, that using efficient evolutionary-based selection leads to the results comparable to absolutely best, found exhaustively. Moreover, as we showed for selected datasets, introducing a second stage combining by majority voting has the potential for both, further improvement of the recognition rate and increase of the reliability of combined outputs.
Gabrys, B 2000, 'Agglomerative learning for general fuzzy min-max neural network', NEURAL NETWORKS FOR SIGNAL PROCESSING X, VOLS 1 AND 2, PROCEEDINGS, 10th IEEE Workshop on Neural Networks for Signal Processing (NNSP 2000), IEEE, UNIV SYDNEY, SYDNEY, AUSTRALIA, pp. 692-701.
Gabrys, B 2000, 'Pattern classification for incomplete data', KES'2000: FOURTH INTERNATIONAL CONFERENCE ON KNOWLEDGE-BASED INTELLIGENT ENGINEERING SYSTEMS & ALLIED TECHNOLOGIES, VOLS 1 AND 2, PROCEEDINGS, 4th International Conference on Knowledge-Based Intelligent Engineering Systems and Allied Technologies, IEEE, UNIV BRIGHTON, BRIGHTON, ENGLAND, pp. 454-457.
Gabrys, B & Bargiela, A 1998, 'Simulation of water distribution systems', ESS'98 - SIMULATION TECHNOLOGY: SCIENCE AND ART, 10th European Simulation Symposium (ESS 98), SOC COMPUTER SIMULATION, NOTTINGHAM TRENT UNIV, NOTTINGHAM, ENGLAND, pp. 273-277.
There have been a large number of partners, both companies and individuals, with whom I have had an undeniable pleasure to work and collaborate both formally and informally on a number of projects over the years.
My research group has always had a strategy of establishing long term collaborative research relationships with major international companies including SAS Institute (business analytics software and services, UK/US), British Telecommunications plc (telecommunications industry, UK), Lufthansa Systems (airline industry, Germany) or Evonik Industries (process industry, Germany) who are some of the previous industrial co-sponsors of a number of our research projects.
The group has also collaborated with many other academic partners and companies which within such large European projects as EUNITE and NiSIS included the co-ordinator of these projects and another of our collaborators, ELITE Foundation from Germany.
Within EU INFER project where, among others, we worked on a development of a Software Platform for Robust and Adaptive Predictive Systems we had been very closely collaborating with the Research & Engineering Centre (REC), a software company from Poland.
Another aspect of our activities were the Knowledge Transfer projects with Small and Medium Enterprises (SMEs) the example of which are the industrial collaborations within DTI and industry funded Knowledge Transfer Partnership programmes with QGate Software Ltd, Next Control Systems Ltd or We Are Base Ltd.
One of my personal and my group’s undertakings in terms of promoting research in Knowledge-based and Intelligent Systems, facilitating fruitful interaction between researchers and establishing research links between our group and various centres worldwide has been our joint leadership, together with Prof. Robert Howlett and Prof Lakhmi Jain, of KES International Research Organisation. KES International is a non-profit organisation which has a large number of members worldwide, and with its organisation of very successful series of conferences regularly attracting over 500 participants (e.g. KES’2006 conference which was hosted in Bournemouth or the KES’2016 conference held in York, both jointly chaired by Prof Gabrys and Prof. Howlett), associated KES journal published by IOS Press and a vibrant international research community have been a source of many collaborative activities.
An example of an exciting academic collaboration within UK was our membership in the EPSRC funded e-Research South consortium (led by the University of Oxford and with other partners including STFC, UKOLN, Universities of Southampton, Reading, Brunel and King's College London).
The list here, which by no means is complete, is just an acknowledgement of the people with whom I have had pleasure to be involved in various research, business and academic activities.
I like the ones who have and continue to argue with me about some obscure detail on regular basis almost as much as those who are prepared to sponsor our curiosity called research without too much of an argument.
So if you are interested in collaborating with us either because you are curious about the things we do or it could greatly benefit your company or business we are always happy to talk and constantly looking for new inspirations and challenges.