Ed Keedwell

A hybrid genetic algorithm for the design of water distribution networks

Genetic algorithms are currently one of the state-of-the-art techniques for the optimisation of engineering systems including water network design and rehabilitation. They are capable of finding near optimal cost solutions to these problems given certain cost and hydraulic parameters. However, many forms of genetic algorithms rely on random starting points that are often poor solutions and the problem of how to efficiently provide good initial estimates of solution sets automatically is still an ongoing research topic. This paper proposes a novel method, known as CANDA-GA, which uses a heuristic-based, local representative cellular automata approach to provide a good initial population for genetic algorithm runs. CANDA-GA is applied to three networks, one taken from the literature and two taken from industry. The results show that the proposed method consistently outperforms the conventional non-heuristic-based GA approach in terms of producing more economically designed water distribution networks.

Discovering Gene Networks with a Neural-Genetic Hybrid

Recent advances in biology (namely, DNA arrays) allow an unprecedented view of the biochemical mechanisms contained within a cell. However, this technology raises new challenges for computer scientists and biologists alike, as the data created by these arrays is often highly complex. One of the challenges is the elucidation of the regulatory connections and interactions between genes, proteins and other gene products. In this paper, a novel method is described for determining gene interactions in temporal gene expression data using genetic algorithms combined with a neural network component. Experiments conducted on real-world temporal gene expression data sets confirm that the approach is capable of finding gene networks that fit the data. A further repeated approach shows that those genes significantly involved in interaction with other genes can be highlighted and hypothetical gene networks and circuits proposed for further laboratory testing.

Battle of the Water Networks II

The Battle of the Water Networks II (BWN-II) is the latest of a series of competitions related to the design and operation of water distribution systems (WDSs) undertaken within the Water Distribution Systems Analysis (WDSA) Symposium series. The BWN-II problem specification involved a broadly defined design and operation problem for an existing network that has to be upgraded for increased future demands, and the addition of a new development area. The design decisions involved addition of new and parallel pipes, storage, operational controls for pumps and valves, and sizing of backup power supply. Design criteria involved hydraulic, water quality, reliability, and environmental performance measures. Fourteen teams participated in the Battle and presented their results at the 14th Water Distribution Systems Analysis conference in Adelaide, Australia, September 2012. This paper summarizes the approaches used by the participants and the results they obtained. Given the complexity of the BWN-II problem and the innovative methods required to deal with the multiobjective, high dimensional and computationally demanding nature of the problem, this paper represents a snap-shot of state of the art methods for the design and operation of water distribution systems. A general finding of this paper is that there is benefit in using a combination of heuristic engineering experience and sophisticated optimization algorithms when tackling complex real-world water distribution system design problems

Deductive sort and climbing sort: New methods for non-dominated sorting

In recent years an increasing number of real-world many-dimensional optimisation problems have been identified across the spectrum of research fields. Many popular evolutionary algorithms use non-dominance as a measure for selecting solutions for future generations. The process of sorting populations into non-dominated fronts is usually the controlling order of computational complexity and can be expensive for large populations or for a high number of objectives. This paper presents two novel methods for non-dominated sorting: deductive sort and climbing sort. The two new methods are compared to the fast non-dominated sort of NSGA-II and the non-dominated rank sort of the omni-optimizer. The results demonstrate the improved efficiencies of the deductive sort and the reductions in comparisons that can be made when applying inferred dominance relationships defined in this paper.

A novel cellular automata based approach to storm sewer design

Optimal storm sewer design aims at minimizing capital investment on infrastructure whilst ensuring good system performance under specified design criteria. An innovative sewer design approach based on cellular automata (CA) principles is introduced in this paper. Cellular automata have been applied as computational simulation devices in various scientific fields. However, some recent research has indicated that CA can also be a viable and efficient optimization engine. This engine is heuristic and largely relies on the key properties of CA: locality, homogeneity, and parallelism. In the proposed approach, the CA-based optimizer is combined with a sewer hydraulic simulator, the EPA Storm Water Management Model (SWMM). At each optimization step, according to a set of transition rules, the optimizer updates all decision variables simultaneously based on the hydraulic situation within each neighbourhood. Two sewer networks (one small artificial network and one large real network) have been tested in this study. The CA optimizer demonstrated its ability to obtain near-optimal solutions in a remarkably small number of computational steps in a comparison of its performance with that of a genetic algorithm.

Genetic algorithms for gene expression analysis

The major problem for current gene expression analysis techniques is how to identify the handful of genes which contribute to a disease from the thousands of genes measured on gene chips (microarrays). The use of a novel neural-genetic hybrid algorithm for gene expression analysis is described here. The genetic algorithm identifies possible gene combinations for classification and then uses the output from a neural network to determine the fitness of these combinations. Normal mutation and crossover operations are used to find increasingly fit combinations. Experiments on artificial and real-world gene expression databases are reported. The results from the algorithm are also explored for biological plausibility and confirm that the algorithm is a powerful alternative to standard data mining techniques in this domain.

Single-layer artificial neural networks for gene expression analysis

Gene expression datasets are being produced in increasing quantities and made available on the web. Several thousands of genes are usually measured for their mRNA expression levels per sample using Affymetrix gene chips and Stanford microarrays, for instance. Such datasets are normally separated into distinct, objectively measured classes, typically disease states or other objectively measured phenotypes. A major problem for current gene expression analysis is, given the disparity between the number of genes measured (typically, thousands) and number of individuals sampled (typically, dozens), how to identify the handful of genes which, individually or in combination, help classify individuals. Previous approaches when faced with the dimensionality of the problem have tended to use unsupervised or supervised techniques that result in smaller clusters of genes, but clusters by themselves do not yield classification rules. This is especially the case with temporal microarray data, which represents the activity of genes within a cell, tissue or organism over time. The expression levels of some genes at a particular time-point can be controlled by the expression levels of other genes at a previous time-point. It is the extraction of these temporal connections within the data that is of great interest to biomolecular scientists and researchers within the pharmaceutical industry. If these so-called gene networks can be found that explain disease inception and progression, drugs can be designed to target specific genes so that the disease either does not progress or is even eradicated from an individual. In this paper we describe novel experiments using single-layer artificial neural networks for modelling both non-temporal (classificatory) and temporal microarray data.

Markov chain hyper-heuristic (MCHH): an online selective hyper-heuristic for multi-objective continuous problems

In this paper we present the Markov chain Hyper-heuristic (MCHH), a novel online selective hyper-heuristic which employs reinforcement learning and Markov chains to provide an adaptive heuristic selection method. Experiments are conducted to demonstrate the efficacy of the method and comparisons are made with standard heuristics, a random hyper-heuristic and a multi-objective hyper-heuristic from the literature. The approaches are compared on a small number of evaluations of the multi-objective DTLZ test problems to reflect the computational limitations of expensive optimisation problems. The results demonstrate the MCHH robust and reliable performance on these problems.

A novel evolutionary meta-heuristic for the multi-objective optimization of real-world water distribution networks

Genetic algorithms are currently one of the state-of-the-art meta-heuristic techniques for the optimization of large engineering systems such as the design and rehabilitation of water distribution networks. They are capable of finding near-optimal cost solutions to these problems given certain cost and hydraulic parameters. Recently, multi-objective genetic algorithms have become prevalent in the water industry due to the conflicting nature of these hydraulic and cost objectives. The Pareto-front of solutions can aid decision makers in the water industry as it provides a set of design solutions which can be examined by experienced engineers. However, multi-objective genetic algorithms tend to require a large number of objective function evaluations to arrive at an acceptable Pareto-front. This article investigates a novel hybrid cellular automaton and genetic approach to multi-objective optimization (known as CAMOGA). The proposed method is applied to two large, real-world networks taken from the UK water industry. The results show that the proposed cellular automaton approach can provide a good approximation of the Pareto-front with very few network simulations, and that CAMOGA outperforms the standard multi-objective genetic algorithm in terms of efficiency in discovering similar Pareto-fronts.

Ant colony optimisation to identify genetic variant association with type 2 diabetes

Around 1.8 million people in the UK have type 2 diabetes, representing about 90% of all diabetes cases in the UK. Genome wide association studies have recently implicated several new genes that are likely to be associated with this disease. However, common genetic variants so far identified only explain a small proportion of the heritability of type 2 diabetes. The interaction of two or more gene variants, may explain a further element of this heritability but full interaction analyses are currently highly computationally burdensome or infeasible. For this reason this study investigates an ant colony optimisation (ACO) approach for its ability to identify common gene variants associated with type 2 diabetes, including putative epistatic interactions. This study uses a dataset comprising 15,309 common (>5% minor allele frequency) SNPs from chromosome 16, genotyped in 1924 type 2 diabetes cases and 2938 controls. This chromosome contains two previously determined associations, one of which is replicated in additional samples. Although no epistatic interactions have been previously reported on this dataset, we demonstrate that ACO can be used to discover single SNP and plausible epistatic associations from this dataset and is shown to be both accurate and computationally tractable on large, real datasets of SNPs with no expert knowledge included in the algorithm.