[show abstract][hide abstract] ABSTRACT: Solving multi-objective problems usually results in a set of Perto-optimal solutions, or a Pareto front. Assessing the quality of these solutions, however, and comparing the performance of different multi-objective optimisers is still not very well understood. Current trends ei-ther model the outcome of the optimiser as a probability density function in the objective space, or defines an indicator that quantify the overall performance of the optimiser. Here an approach based on the concept of mutual information is proposed. The approach models the probability density function of the optimisers' output and use that to define an in-dicator, namely the amount of shared information among the compared Pareto fronts. The strength of the new approach is not only in better assessment of performance but also the interpretability of the results it provides.
Intelligent Data Engineering and Automated Learning – IDEAL 2013 Lecture Notes in Computer Science Volume; 10/2013
[show abstract][hide abstract] ABSTRACT: Abstract This paper improves a recently developed multi-objective particle swarm optimizer (D(2)MOPSO) that incorporates dominance with decomposition used in the context of multi-objective optimisation. Decomposition simplifies a multi-objective problem(MOP) by transforming it to a set of aggregation problems, whereas dominance plays a major role in building the leaders' archive. D(2)MOPSO introduces a new archiving technique that facilitates attaining better diversity and coverage in both objective and solution spaces. The improved method is evaluated on standard benchmarks including both constrained and unconstrained test problems, by comparing it with three state-of-the-art multi-objective evolutionary algorithms: MOEA/D, OMOPSO and dMOPSO. The comparison and analysis of the experimental results, supported by statistical tests, indicate that the proposed algorithm is highly competitive, efficient and applicable to a wide range of multi-objective optimisation problems.
[show abstract][hide abstract] ABSTRACT: Objectives
To construct new prostate cancer staging lookup tables based on a dataset collated by the British Association of Urological Surgeons (BAUS) and to validate them and compare their predictive power with Partin tables.
Patients and methods
Complete data on 1701 patients was collated between 1999 and 2008 across 57 UK centres. Lookup tables were created for prediction of pathological stage (PS) using PSA level, biopsy Gleason score (GS) and clinical stage, replicating Partin's original approach.
Tables were generated using logistic regression (LR) and bootstrap resampling methods and were internally validated and externally validated using concordance indices (CI) and area under the receiver operating characteristic curve (AUROC) respectively.
The CI and AUROC analyses indicate that Partin tables performed poorly on UK data in comparison with US data.
The UK prostate cancer tables performed better than Partin tables but the predictive power of all models was relatively poor.
The study shows that the predictive power of Partin tables is reduced when applied to the UK population.
Models generated using LR methodology have fundamental limitations, and we suggest alternative modelling methods such as Bayesian networks.
British Journal of Medical and Surgical Urology 09/2012; 5(5):224–235.
[show abstract][hide abstract] ABSTRACT: Surrogate models of fitness have been presented as a way of reducing the number of fitness evaluations required by evolutionary algorithms. This is of particular interest with expensive fitness functions where the time taken for building the model is outweighed by the savings of using fewer function evaluations. In this article, we show how a Markov network model can be used as a surrogate fitness function for a genetic algorithm in a new algorithm called Markov Fitness Model Genetic Algorithm (MFM-GA). We thoroughly investigate its application to a fitness function for feature selection in Case-Based Reasoning (CBR), using a range of standard benchmarks from the CBR community. This fitness function requires considerable computation time to evaluate and we show that using the surrogate offers a significant decrease in total run-time compared to a GA using the true fitness function. This comes at the cost of a reduction in the global best fitness found. We demonstrate that the quality of the solutions obtained by MFM-GA improves significantly with model rebuilding. Comparisons with a classic GA, a GA using fitness inheritance and a selection of filter selection methods for CBR shows that MFM-GA provides a good trade-off between fitness quality and run-time.
[show abstract][hide abstract] ABSTRACT: Learning a good model structure is important to the efficient solving of problems by estimation of distribution algorithms. In this paper we present the results of a series of experiments, applying a structure learning algorithm for undirected probabilistic graphical models based on statistical dependency tests to three fitness functions with different selection operators, proportions and pressures. The number of spurious interactions found by the algorithm are measured and reported. Truncation selection, and its complement (selecting only low fitness solutions) prove quite robust, resulting in a similar number of spurious dependencies regardless of selection pressure. In contrast, tournament and fitness proportionate selection are strongly affected by the selection proportion and pressure.
[show abstract][hide abstract] ABSTRACT: Prediction of prostate cancer pathological stage is an essential step in a patient's pathway. It determines the treatment that will be applied further. In current practice, urologists use the pathological stage predictions provided in Partin tables to support their decisions. However, Partin tables are based on logistic regression (LR) and built from US data. Our objective is to investigate a range of both predictive methods and of predictive variables for pathological stage prediction and assess them with respect to their predictive quality based on U.K. data.
The latest version of Partin tables was applied to a large scale British dataset in order to measure their performances by mean of concordance index (c-index). The data was collected by the British Association of Urological Surgeons (BAUS) and gathered records from over 1700 patients treated with prostatectomy in 57 centers across UK. The original methodology was replicated using the BAUS dataset and evaluated using concordance index. In addition, a selection of classifiers, including, among others, LR, artificial neural networks and Bayesian networks (BNs) was applied to the same data and compared with each other using the area under the ROC curve (AUC). Subsets of the data were created in order to observe how classifiers perform with the inclusion of extra variables. Finally a local dataset prepared by the Aberdeen Royal Infirmary was used to study the effect on predictive performance of using different variables.
Partin tables have low predictive quality (c-index=0.602) when applied on UK data for comparison on patients with organ confined and extra prostatic extension conditions, patients at the two most frequently observed pathological stages. The use of replicate lookup tables built from British data shows an improvement in the classification, but the overall predictive quality remains low (c-index=0.610). Comparing a range of classifiers shows that BNs generally outperform other methods. Using the four variables from Partin tables, naive Bayes is the best classifier for the prediction of each class label (AUC=0.662 for OC). When two additional variables are added, the results of LR (0.675), artificial neural networks (0.656) and BN methods (0.679) are overall improved. BNs show higher AUCs than the other methods when the number of variables raises
The predictive quality of Partin tables can be described as low to moderate on U.K. data. This means that following the predictions generated by Partin tables, many patients would received an inappropriate treatment, generally associated with a deterioration of their quality of life. In addition to demographic differences between U.K. and the original U.S. population, the methodology and in particular LR present limitations. BN represents a promising alternative to LR from which prostate cancer staging can benefit. Heuristic search for structure learning and the inclusion of more variables are elements that further improve BN models quality.
Artificial intelligence in medicine 12/2011; 55(1):25-35. · 1.65 Impact Factor
[show abstract][hide abstract] ABSTRACT: The paper presents a novel approach to optimising cancer chemotherapy with respect to conflicting treatment objectives aimed at reducing the number of cancerous cells and at limiting the amounts of anti-cancer drugs used. The approach is based on the Particle Swarm Optimisation (PSO) algorithm that decomposes a multi-objective optimisation problem into several scalar aggregation problems, thereby reducing its complexity and enabling an effective application of Computational Intelligence techniques. The novelty of the algorithm is in providing particles in the swarm with information from a set of defined neighbours and leaders that assists in finding versatile chemotherapeutic treatments.
[show abstract][hide abstract] ABSTRACT: In many situations, data is scattered across different sites, making the modeling process difficult or sometimes impossible. Some applications could benefit from collaborations between organisations but data security or privacy policies often act as a barrier to data mining on such contexts. In this paper, we present a novel approach to learning Bayesian Networks (BN) structures from multiple datasets, based on the use of Ensembles and an Island Model Genetic Algorithm (IMGA). The proposed design ensures no data is shared during the process and can fit many applications.
13th Annual Genetic and Evolutionary Computation Conference, GECCO 2011, Companion Material Proceedings, Dublin, Ireland, July 12-16, 2011; 01/2011
[show abstract][hide abstract] ABSTRACT: Algorithms for learning the structure of Bayesian Networks (BN) from data are the focus of intense research interest. Search-and-score algorithms using nature-inspired metaheuristics are an important strand of this research; however performance is variable and strongly problem-dependent. In this paper we use fitness landscape analysis to explain empirically- observed performance differences between particular search- and-score algorithms on two well-studied benchmark problems. We investigate the average landscape discovered by random walks around optimal points in the space of BN node orderings. Differences in algorithm performance are explained in terms of these landscapes, which in turn are related to properties of the BN structures. These initial findings suggest that fitness landscape analysis is a promising approach for explaining existing empirical performance comparisons with further potential for understanding the relative difficulty of benchmark problems and the robustness of particular algorithms.
Proceedings of the IEEE Congress on Evolutionary Computation, CEC 2011, New Orleans, LA, USA, 5-8 June, 2011; 01/2011
[show abstract][hide abstract] ABSTRACT: In, we introduced Smart Multi-Objective Particle Swarm Optimisation using Decomposition (SDMOPSO). The method uses the decomposition approach proposed in Multi-Objective Evolutionary Algorithms based on Decomposition (MOEA/D), whereby a multi-objective problem (MOP) is represented as several scalar aggregation problems. The scalar aggregation problems are viewed as particles in a swarm; each particle assigns weights to every optimisation objective. The problem is solved then as a Multi-Objective Particle Swarm Optimisation (MOPSO), in which every particle uses information from a set of defined neighbours. This work customize SDMOSPO to cover binary problems and applies the proposed binary method on the channel selection problem for Brain-Computer Interfaces (BCI).
Computational Intelligence (UKCI), 2010 UK Workshop on; 10/2010
[show abstract][hide abstract] ABSTRACT: The operation of drilling rigs is highly expensive. It is therefore important to be able to identify and analyse variables affecting rig operations. We investigate the use of Genetic Algorithms and Ant Colony Optimisation to induce a Bayesian Network model for the real world problem of Rig Operations Management and confirm the validity of our previous model. We explore the relative performances of different search and scoring heuristics and consider trade-offs between best network score and computation time from an industry standpoint. Finally, we analyse edge-discovery statistics over repeated runs to explain observed differences between the algorithms.
Computational Intelligence (UKCI), 2010 UK Workshop on; 10/2010
[show abstract][hide abstract] ABSTRACT: The operation of drilling rigs is highly expensive. It is therefore important to be able to identify and analyse factors affecting rig operations. We investigate the use of two Genetic Algorithms, K2GA and ChainGA, to induce a Bayesian Network model for the real world problem of Rig Operations Management. We sample from a unique dataset derived from the commercial market intelligence databases assembled by ODS-Petrodata Ltd. We observe a trade-off between K2GA, which finds significantly better scoring networks on our dataset, and ChainGA, which uses only one quarter of the computation time. We analyse the best structures produced from an industry standpoint and conclude by outlining a few potential applications of the models to support rig operations.
Evolutionary Computation (CEC), 2010 IEEE Congress on; 08/2010
[show abstract][hide abstract] ABSTRACT: Cancer treatment by chemotherapy involves multiple applications of toxic drugs over a period of time. Optimising the schedule of these treatments can improve the outcome for the patient. A schedule of treatment and its effect on the tumour can be simulated by a mathematical growth model. However, when used in conjunction with a black-box optimisation algorithm such as an Evolutionary Algorithm (EA) to search for effective treatment schedules, the frequent use of the model can become computationally onerous. One approach to improve the efficiency of EAs is to use `fitness inheritance', in which, for a proportion of candidate solutions, simple means are used to estimate the fitness, rather than use the computationally intensive model. We investigate two versions of fitness inheritance for the chemotherapy schedule optimisation problem, and demonstrate the significant improvement in efficiency that can be achieved. In particular, concerning the two main types of fitness inheritance (Averaged and Proportional), we find that the Averaged Inheritance strategy is highly effective in this case, and is strongly recommended for use in further investigations of chemotherapy optimisation using population-based search.
Evolutionary Computation (CEC), 2010 IEEE Congress on; 08/2010