The evolution of learning systems:to Bayes or not to be

Bayesian algorithms pose a limit to the performance learning algorithms can achieve. Natural selection should guide the evolution of information processing systems towards those limits. What can we learn from this evolution and what properties do the intermediate stages have? While this question is too general to permit any answer, progress can be made by restricting the class of information processing systems under study. We present analytical and numerical results for the evolution of on-line algorithms for learning from examples for neural network classifiers, which might include or not a hidden layer. The analytical results are obtained by solving a variational problem to determine the learning algorithm that leads to maximum generalization ability. Simulations using evolutionary programming, for programs that implement learning algorithms, confirm and expand the results. The principal result is not just that the evolution is towards a Bayesian limit. Indeed it is essentially reached. In addition we find that evolution is driven by the discovery of useful structures or combinations of variables and operators. In different runs the temporal order of the discovery of such combinations is unique. The main result is that combinations that signal the surprise brought by an example arise always before combinations that serve to gauge the performance of the learning algorithm. This latter structures can be used to implement annealing schedules. The temporal ordering can be understood analytically as well by doing the functional optimization in restricted functional spaces. We also show that there is data suggesting that the appearance of these traits also follows the same temporal ordering in biological systems. © 2006 American Institute of Physics.

Perception of concurrent sentences with harmonic or frequency-shifted voiced excitation:performance of human listeners and of computational models based on autocorrelation

Keyword identification in one of two simultaneous sentences is improved when the sentences differ in F0, particularly when they are almost continuously voiced. Sentences of this kind were recorded, monotonised using PSOLA, and re-synthesised to give a range of harmonic ?F0s (0, 1, 3, and 10 semitones). They were additionally re-synthesised by LPC with the LPC residual frequency shifted by 25% of F0, to give excitation with inharmonic but regularly spaced components. Perceptual identification of frequency-shifted sentences showed a similar large improvement with nominal ?F0 as seen for harmonic sentences, although overall performance was about 10% poorer. We compared performance with that of two autocorrelation-based computational models comprising four stages: (i) peripheral frequency selectivity and half-wave rectification; (ii) within-channel periodicity extraction; (iii) identification of the two major peaks in the summary autocorrelation function (SACF); (iv) a template-based approach to speech recognition using dynamic time warping. One model sampled the correlogram at the target-F0 period and performed spectral matching; the other deselected channels dominated by the interferer and performed matching on the short-lag portion of the residual SACF. Both models reproduced the monotonic increase observed in human performance with increasing ?F0 for the harmonic stimuli, but not for the frequency-shifted stimuli. A revised version of the spectral-matching model, which groups patterns of periodicity that lie on a curve in the frequency-delay plane, showed a closer match to the perceptual data for frequency-shifted sentences. The results extend the range of phenomena originally attributed to harmonic processing to grouping by common spectral pattern.

The black middle class: middle class Afro-Caribbeans: a racial fraction of the British middle class or a class fraction of a racial group

This research examines the relationship between 'race' and class in Britain. This is achieved by considering how these two concepts articulate in the overall structuring of class relationships in a society which is typified by the incorporatation of black labour into a majority white society, This relationship is examined through an investigation of those black workers who occupy a position in the objectively defined middle class. The basic theme underlying this research is that 'race, in the form of structural racism, plays a significant role at two levels. Firstly, it serves to structure the class position of black labour in Britain. Secondly, it serves to determine the type of race, class and political consciousness generated by black labour. The study was carried out in the London area. Occupation was used as an indicator of 'objective' class position when selecting respondents to be included in the two survey populations required for the research. A 'network' approach was used to actually locate the respondents. In-depth interviews were carried out with all the respondents. The study concludes that the concepts of 'race' and class are not independent of each other in the overall structuring of class relationships between black and white labour. It is argued that the inter-relationship identified between these two concepts serves to highlight the fact that the structural position of black labour, the type of consciousness generated and the type of decisions taken by those who took part in the research are to a large extent a result of the structural constraints deriving from the effects of structural racism in Britain.

Statistical guidelines for the analysis of data obtained from one or both eyes

Purpose - Measurements obtained from the right and left eye of a subject are often correlated whereas many statistical tests assume observations in a sample are independent. Hence, data collected from both eyes cannot be combined without taking this correlation into account. Current practice is reviewed with reference to articles published in three optometry journals, viz., Ophthalmic and Physiological Optics (OPO), Optometry and Vision Science (OVS), Clinical and Experimental Optometry (CEO) during the period 2009–2012. Recent findings - Of the 230 articles reviewed, 148/230 (64%) obtained data from one eye and 82/230 (36%) from both eyes. Of the 148 one-eye articles, the right eye, left eye, a randomly selected eye, the better eye, the worse or diseased eye, or the dominant eye were all used as selection criteria. Of the 82 two-eye articles, the analysis utilized data from: (1) one eye only rejecting data from the adjacent eye, (2) both eyes separately, (3) both eyes taking into account the correlation between eyes, or (4) both eyes using one eye as a treated or diseased eye, the other acting as a control. In a proportion of studies, data were combined from both eyes without correction. Summary - It is suggested that: (1) investigators should consider whether it is advantageous to collect data from both eyes, (2) if one eye is studied and both are eligible, then it should be chosen at random, and (3) two-eye data can be analysed incorporating eyes as a ‘within subjects’ factor.

Estimating the SCAN*PRO model of store sales: HB, FM or just OLS?

In this paper we investigate whether consideration of store-level heterogeneity in marketing mix effects improves the accuracy of the marketing mix elasticities, fit, and forecasting accuracy of the widely-applied SCAN*PRO model of store sales. Models with continuous and discrete representations of heterogeneity, estimated using hierarchical Bayes (HB) and finite mixture (FM) techniques, respectively, are empirically compared to the original model, which does not account for store-level heterogeneity in marketing mix effects, and is estimated using ordinary least squares (OLS). The empirical comparisons are conducted in two contexts: Dutch store-level scanner data for the shampoo product category, and an extensive simulation experiment. The simulation investigates how between- and within-segment variance in marketing mix effects, error variance, the number of weeks of data, and the number of stores impact the accuracy of marketing mix elasticities, model fit, and forecasting accuracy. Contrary to expectations, accommodating store-level heterogeneity does not improve the accuracy of marketing mix elasticities relative to the homogeneous SCAN*PRO model, suggesting that little may be lost by employing the original homogeneous SCAN*PRO model estimated using ordinary least squares. Improvements in fit and forecasting accuracy are also fairly modest. We pursue an explanation for this result since research in other contexts has shown clear advantages from assuming some type of heterogeneity in market response models. In an Afterthought section, we comment on the controversial nature of our result, distinguishing factors inherent to household-level data and associated models vs. general store-level data and associated models vs. the unique SCAN*PRO model specification.

The social and entrepreneurial orientation of small businesses:mutually exclusive or reinforcing? (INTERACTIVE PAPER)

Principal Topic - As argued by Acs and Phillips (2002) it is not only “the creation of wealth (entrepreneurship)” but also “the reconstitution of wealth (philanthropy)”, which has been essential for the inherent dynamism of the market economy (Ibid., p.201). However, we understand little about the entrepreneurship – philanthropy link in institutional contexts that differ from that of leading developed market economies. Accordingly our research agenda is to investigate the entrepreneurship-philanthropy nexus in a very different context of Lithuania, a country which shed a command economy system twenty years ago. In particular, we are interested to see if the cluster of attitudes and strategies of firms conducive to entrepreneurship, i.e. their entrepreneurial orientation (Covin & Slevin, 1989), is consistent or contradictory with philanthropy? In other words, is philanthropy strongly associated with some core components of entrepreneurship, or is it an entrepreneurial anomaly, relying on a minority of economic actors that provide important links with wider, non-economic communities. Method - The study draws on 270 randomly sampled, phone interviews with owners and ownermanagers of small and medium-sized enterprises (SMEs), i.e. firms with less than 250 employees. Interviews were conducted in Lithuania during January- March, 2008. Our results are based on confirmatory factor analysis combined with regression analysis. Results and Implications - Despite the legacy of informal institutions that is conducive neither to entrepreneurship nor to civic society, we found that by now, (i) the companies that score highest on entrepreneurial orientation construct, (ii) that perform best and those (iii) that have foreign owners are also most likely to declare their commitment to philanthropy. Our findings that most entrepreneurial firms are also involved in philanthropy are consistent with the perspective on the pattern of development in an entrepreneurial economy as outlined by Acs and Phillips (2002).

Extracellular Nm23H1 stimulates neurite outgrowth from dorsal root ganglia neurons in vitro independently of nerve growth factor supplementation or its nucleoside diphosphate kinase activity

The nucleoside diphosphate (NDP) kinase, Nm23H1, is a highly expressed during neuronal development, whilst induced over-expression in neuronal cells results in increased neurite outgrowth. Extracellular Nm23H1 affects the survival, proliferation and differentiation of non-neuronal cells. Therefore, this study has examined whether extracellular Nm23H1 regulates nerve growth. We have immobilised recombinant Nm23H1 proteins to defined locations of culture plates, which were then seeded with explants of embryonic chick dorsal root ganglia (DRG) or dissociated adult rat DRG neurons. The substratum-bound extracellular Nm23H1 was stimulatory for neurite outgrowth from chick DRG explants in a concentration-dependent manner. On high concentrations of Nm23H1, chick DRG neurite outgrowth was extensive and effectively limited to the location of the Nm23H1, i.e. neuronal growth cones turned away from adjacent collagen-coated substrata. Nm23H1-coated substrata also significantly enhanced rat DRG neuronal cell adhesion and neurite outgrowth in comparison to collagen-coated substrata. These effects were independent of NGF supplementation. Recombinant Nm23H1 (H118F), which does not possess NDP kinase activity, exhibited the same activity as the wild-type protein. Hence, a novel neuro-stimulatory activity for extracellular Nm23H1 has been identified in vitro, which may function in developing neuronal systems. © 2010 Elsevier Inc.

The Role of the big five personality traits, proactive behaviour, and socialisation influences in newcomer adjustment

During the early stages of employment, newly hired employees find out what their new organisations are like. Their first impressions are extremely important in determining the course of subsequent attitudes and behaviour. Recently, a considerable progress has been made towards the understanding of adjustment process, however, the literature remain divided along a number of fronts. Moreover, newcomer research has been conducted independent and irrespective of newcomer personality individual differences. This seems to be a critical oversight because there is overlap in predictions involving these constructs. The current research extended the previous one by examining these multiple antecedents, including Big Five personality traits of newcomer to the tandem process of adjustment as well as outcomes that immediate, or ?proximal? to the process of newcomer adjustment. Following a cross sectional pilot study of recent college graduate, a three- wave longitudinal study of newcomers in seven organisations examined Big Five personality traits, proactive behaviour, and socialisation influence (formal training, leaders, co-workers) as antecedents of proximal adjustment outcomes (group integration, political knowledge of organisation, and task performance). The main study results suggested that personality traits were related to proximal adjustment outcomes, specifically, Conscientiousness was positively related to all proximal adjustment outcomes. Openness to experience was related to task performance and political knowledge. Group integration is independently related to Agreeableness, Extraversion and Neuroticism. The socialisation influence moderate these relations, for example, leader socialisation moderate Conscientiousness as it relates to political knowledge and group integration, while co-worker moderate Extraversion as it relates to task performance. Finally, it was found that, the relationship between proximal adjustment outcomes and the personality dimensions Openness was mediated by proactive behaviour. Overall, the results suggested that individual differences have a role in newcomer adjustment as it facilitate the socialisation influence, and Big Five was one of the key determinants of newcomer adjustment.