Selection procedures for module discovery : exploring evolutionary algorithms for cognitive science

Evolutionary algorithms are playing an increasingly important role as search methods in cognitive science domains. In this study, methodological issues in the use of evolutionary algorithms were investigated via simulations in which procedures were systematically varied to modify the selection pressures on populations of evolving agents. Traditional roulette wheel, tournament, and variations of these selection algorithms were compared on the “needle-in-a-haystack” problem developed by Hinton and Nowlan in their 1987 study of the Baldwin effect. The task is an important one for cognitive science, as it demonstrates the power of learning as a local search technique in smoothing a fitness landscape that lacks gradient information. One aspect that has continued to foster interest in the problem is the observation of residual learning ability in simulated populations even after long periods of time. Effective evolutionary algorithms balance their search effort between broad exploration of the search space and in-depth exploitation of promising solutions already found. Issues discussed include the differential effects of rank and proportional selection, the tradeoff between migration of populations towards good solutions and maintenance of diversity, and the development of measures that illustrate how each selection algorithm affects the search process over generations. We show that both roulette wheel and tournament algorithms can be modified to appropriately balance search between exploration and exploitation, and effectively eliminate residual learning in this problem.

A continent-wide study reveals clear relationships between regional abiotic conditions and post-dispersal seed predation

Aim Large-scale patterns linking energy availability, biological productivity and diversity form a central focus of ecology. Despite evidence that the activity and abundance of animals may be limited by climatic variables associated with regional biological productivity (e.g. mean annual precipitation and annual actual evapotranspiration), it is unclear whether plant–granivore interactions are themselves influenced by these climatic factors across broad spatial extents. We evaluated whether climatic conditions that are known to alter the abundance and activity of granivorous animals also affect rates of seed removal. Location Eleven sites across temperate North America. Methods We used a common protocol to assess the removal of the same seed species (Avena sativa) over a 2-day period. Model selection via the Akaike information criterion was used to determine a set of candidate binomial generalized linear mixed models that evaluated the relationship between local climatic data and post-dispersal seed predation. Results Annual actual evapotranspiration was the single best predictor of the proportion of seeds removed. Annual actual evapotranspiration and mean annual precipitation were both positively related to mean seed removal and were included in four and three of the top five models, respectively. Annual temperature range was also positively related to seed removal and was an explanatory variable in three of the top four models. Main conclusions Our work provides the first evidence that energy and precipitation, which are known to affect consumer abundance and activity, also translate to strong, predictable patterns of seed predation across a continent. More generally, these findings suggest that future changes in temperature and precipitation could have widespread consequences for plant species composition in grasslands, through impacts on plant recruitment.

Sex-specific roost selection by bats in clearfell harvested plantation forest : improved knowledge advises management

To maximize energetic savings, female bats often roost communally whilst pregnant or with non-volant dependents, whereas male bats more often roost alone; however, differences in selection of roosts by sex have not often been investigated. Better understanding of female colony locations could focus management to protect the majority of bats. New Zealand's long-tailed bat (Chalinolobus tuberculatus) roost in exotic plantation forest, where sex-specific roost selection has not been investigated, and therefore such management is not possible. We investigated sex-specific roost selection by long-tailed bats for the first time. Roosts and paired nonroosts were characterized testing predictions that males and females select roosts that differ from non-roosts, and males and females select different roosts. Females and males chose Pinus radiata roosts that differed from non-roost trees. Results suggest each sex chose roosts that maximized energetic savings. Female bats used roosts closer to water sources, that warmed earlier in the day, which allowed maintenance of high temperatures. Males appeared to choose roosts that allowed torpor use for long periods of the day. Males may be less selective with their roost locations than females, as they roosted further from water sources. This could allow persistence of male bats in marginal habitat. As all female long-tailed bats chose roosts within 150 m of waterways, management to protect bats could be focused here. To protect bats least able to escape when roosts are harvested, harvest of forest stands selected by female bats as roost sites should be planned when bats are not heavily pregnant nor have non-volant dependents.

Home range and habitat selection by a threatened bat in exotic plantation forest

Individuals' home ranges are constrained by resource distribution and density, population size, and energetic requirements. Consequently, home ranges and habitat selection may vary between individuals of different sex and reproductive conditions. Whilst home ranges of bats are well-studied in native habitats, they are often not well understood in modified landscapes, particularly exotic plantation forests. Although Chalinolobus tuberculatus (Vespertilionidae, Chiroptera) are present in plantation forests throughout New Zealand their home ranges have only been studied in native forest and forest-agricultural mosaic and no studies of habitat selection that included males had occurred in any habitat type. Therefore, we investigated C. tuberculatus home range and habitat selection within exotic plantation forest. Home range sizes did not differ between bats of different reproductive states. Bats selected home ranges with higher proportions of relatively old forest than was available. Males selected edges with open unplanted areas within their home ranges, which females avoided. We suggest males use these edges, highly profitable foraging areas with early evening peaks in invertebrate abundance, to maintain relatively low energetic demands. Females require longer periods of invertebrate activity to fulfil their needs so select older stands for foraging, where invertebrate activity is higher. These results highlight additional understanding gained when data are not pooled across sexes. Mitigation for harvest operations could include ensuring that areas suitable for foraging and roosting are located within a radius equal to the home range of this bat species.

A hybrid quality function deployment and cybernetic analytic network process model for project manager selection

Identifying appropriate decision criteria and making optimal decisions in a structured way is a complex process. This paper presents an approach for doing this in the form of a hybrid Quality Function Deployment (QFD) and Cybernetic Analytic Network Process (CANP) model for project manager selection. This involves the use of QFD to translate the owner's project management expectations into selection criteria and the CANP to weight the expectations and selection criteria. The supermatrix approach then prioritises the candidates with respect to the overall decision-making goal. A case study is used to demonstrate the use of the model in selecting a renovation project manager. This involves the development of 18 selection criteria in response to the owner's three main expectations of time, cost and quality.

Feature selection using expected attainable discrimination

We propose expected attainable discrimination (EAD) as a measure to select discrete valued features for reliable discrimination between two classes of data. EAD is an average of the area under the ROC curves obtained when a simple histogram probability density model is trained and tested on many random partitions of a data set. EAD can be incorporated into various stepwise search methods to determine promising subsets of features, particularly when misclassification costs are difficult or impossible to specify. Experimental application to the problem of risk prediction in pregnancy is described.

Growth medium selection and its economic impact on plasmid DNA production

Current developments in gene medicine and vaccination studies are utilizing plasmid DNA (pDNA) as the vector. For this reason, there has been an increasing trend towards larger and larger doses of pDNA utilized in human trials: from 100-1000 μg in 2002 to 500-5000 μg in 2005. The increasing demand of pDNA has created the need to revolutionalize current production levels under optimum economy. In this work, different standard media (LB, TB and SOC) for culturing recombinant Escherichia coli DH5α harbouring pUC19 were compared to a medium optimised for pDNA production. Lab scale fermentations using the standard media showed that the highest pDNA volumetric and specific yields were for TB (11.4 μg/ml and 6.3 μg/mg dry cell mass respectively) and the lowest was for LB (2.8 μg/ml and 3.3 μg/mg dry cell mass respectively). A fourth medium, PDMR, designed by modifying a stoichiometrically-formulated medium with an optimised carbon source concentration and carbon to nitrogen ratio displayed pDNA volumetric and specific yields of 23.8 μg/ml and 11.2 μg/mg dry cell mass respectively. However, it is the economic advantages of the optimised medium that makes it so attractive. Keeping all variables constant except medium and using LB as a base scenario (100 medium cost [MC] units/mg pDNA), the optimised PDMR medium yielded pDNA at a cost of only 27 MC units/mg pDNA. These results show that greater amounts of pDNA can be obtained more economically with minimal extra effort simply by using a medium optimised for pDNA production.

Rethinking board composition: Balancing competence and compatibility in director selection

Board composition is critical to board effectiveness. Shaping an effective board begins with the selection of directors. While much attention has been paid to the skills and qualifications directors require, there has been less focus on the necessity for board members to interact and work well together. This exploratory study offers insights into what qualities directors look for when selecting new members and the approach adopted to identify and select them. The findings of 10 in-depth interviews with Australian directors suggest new members are selected both on competencies and compatibility. Yet not all selection approaches adequately assess candidates for these two criteria. As a result many appointments fail to realise the selection criteria reducing capacity to reach its full potential.