Sparse kernel density construction using orthogonal forward regression with leave-one-out test score and local regularization

This paper presents an efficient construction algorithm for obtaining sparse kernel density estimates based on a regression approach that directly optimizes model generalization capability. Computational efficiency of the density construction is ensured using an orthogonal forward regression, and the algorithm incrementally minimizes the leave-one-out test score. A local regularization method is incorporated naturally into the density construction process to further enforce sparsity. An additional advantage of the proposed algorithm is that it is fully automatic and the user is not required to specify any criterion to terminate the density construction procedure. This is in contrast to an existing state-of-art kernel density estimation method using the support vector machine (SVM), where the user is required to specify some critical algorithm parameter. Several examples are included to demonstrate the ability of the proposed algorithm to effectively construct a very sparse kernel density estimate with comparable accuracy to that of the full sample optimized Parzen window density estimate. Our experimental results also demonstrate that the proposed algorithm compares favorably with the SVM method, in terms of both test accuracy and sparsity, for constructing kernel density estimates.

Fast kernel classifier construction using orthogonal forward selection to minimise leave-one-out misclassification rate

We propose a simple yet computationally efficient construction algorithm for two-class kernel classifiers. In order to optimise classifier's generalisation capability, an orthogonal forward selection procedure is used to select kernels one by one by minimising the leave-one-out (LOO) misclassification rate directly. It is shown that the computation of the LOO misclassification rate is very efficient owing to orthogonalisation. Examples are used to demonstrate that the proposed algorithm is a viable alternative to construct sparse two-class kernel classifiers in terms of performance and computational efficiency.

A fast linear-in-the-parameters classifier construction algorithm using orthogonal forward selection to minimize leave-one-out misclassification rate

We propose a simple and computationally efficient construction algorithm for two class linear-in-the-parameters classifiers. In order to optimize model generalization, a forward orthogonal selection (OFS) procedure is used for minimizing the leave-one-out (LOO) misclassification rate directly. An analytic formula and a set of forward recursive updating formula of the LOO misclassification rate are developed and applied in the proposed algorithm. Numerical examples are used to demonstrate that the proposed algorithm is an excellent alternative approach to construct sparse two class classifiers in terms of performance and computational efficiency.

Backward elimination model construction for regression and classification using leave-one-out criteria

A fundamental principle in practical nonlinear data modeling is the parsimonious principle of constructing the minimal model that explains the training data well. Leave-one-out (LOO) cross validation is often used to estimate generalization errors by choosing amongst different network architectures (M. Stone, "Cross validatory choice and assessment of statistical predictions", J. R. Stast. Soc., Ser. B, 36, pp. 117-147, 1974). Based upon the minimization of LOO criteria of either the mean squares of LOO errors or the LOO misclassification rate respectively, we present two backward elimination algorithms as model post-processing procedures for regression and classification problems. The proposed backward elimination procedures exploit an orthogonalization procedure to enable the orthogonality between the subspace as spanned by the pruned model and the deleted regressor. Subsequently, it is shown that the LOO criteria used in both algorithms can be calculated via some analytic recursive formula, as derived in this contribution, without actually splitting the estimation data set so as to reduce computational expense. Compared to most other model construction methods, the proposed algorithms are advantageous in several aspects; (i) There are no tuning parameters to be optimized through an extra validation data set; (ii) The procedure is fully automatic without an additional stopping criteria; and (iii) The model structure selection is directly based on model generalization performance. The illustrative examples on regression and classification are used to demonstrate that the proposed algorithms are viable post-processing methods to prune a model to gain extra sparsity and improved generalization.

Interpretation of ambiguity in children: a prospective study of associations with anxiety and parental interpretations

Interpretation of ambiguity is consistently associated with anxiety in children, however, the temporal relationship between interpretation and anxiety remains unclear as do the developmental origins of interpretative biases. This study set out to test a model of the development of interpretative biases in a prospective study of 110 children aged 5–9 years of age. Children and their parents were assessed three times, annually, on measures of anxiety and interpretation of ambiguous scenarios (including, for parents, both their own interpretations and their expectations regarding their child). Three models were constructed to assess associations between parent and child anxiety and threat and distress cognitions and expectancies. The three models were all a reasonable fit of the data, and supported conclusions that: (i) children’s threat and distress cognitions were stable over time and were significantly associated with anxiety, (ii) parents’ threat and distress cognitions and expectancies significantly predicted child threat cognitions at some time points, and (iii) parental anxiety significantly predicted parents cognitions, which predicted parental expectancies at some time points. Parental expectancies were also significantly predicted by child cognitions. The findings varied depending on assessment time point and whether threat or distress cognitions were being considered. The findings support the notion that child and parent cognitive processes, in particular parental expectations, may be a useful target in the treatment or prevention of anxiety disorders in children.

Parental incarceration, attachment and child psychopathology

Theory and evidence relating parental incarceration, attachment, and psychopathology are reviewed. Parental incarceration is a strong risk factor for long-lasting psychopathology, including antisocial and internalizing outcomes. Parental incarceration might threaten children's attachment security because of parent-child separation, confusing communication about parental absence, restricted contact with incarcerated parents, and unstable caregiving arrangements. Parental incarceration can also cause economic strain, reduced supervision, stigma, home and school moves, and other negative life events for children. Thus, there are multiple possible mechanisms whereby parental incarceration might increase risk for child psychopathology. Maternal incarceration tends to cause more disruption for children than paternal incarceration and may lead to greater risk for insecure attachment and psychopathology. Children's prior attachment relations and other life experiences are likely to be of great importance for understanding children's reactions to parental incarceration. Several hypotheses are presented about how prior insecure attachment and social adversity might interact with parental incarceration and contribute to psychopathology. Carefully designed longitudinal studies, randomized controlled trials, and cross-national comparative research are required to test these hypotheses.

The parental overprotection scale: associations with child and parental anxiety

Background: Parental overprotection has commonly been implicated in the development and maintenance of childhood anxiety disorders. Overprotection has been assessed using questionnaire and observational methods interchangeably; however, the extent to which these methods access the same construct has received little attention. Edwards, 2008 and Edwards et al., 2010 developed a promising parent-report measure of overprotection (OP) and reported that, with parents of pre-school children, the measure correlated with observational assessments and predicted changes in child anxiety symptoms. We aimed to validate the use of the OP measure with mothers of children in middle childhood, and examine its association with child and parental anxiety. Methods: Mothers of 90 children (60 clinically anxious, 30 non-anxious) aged 7–12 years completed the measure and engaged in a series of mildly stressful tasks with their child. Results: The internal reliability of the measure was good and scores correlated significantly with observations of maternal overprotection in a challenging puzzle task. Contrary to expectations, OP was not significantly associated with child anxiety status or symptoms, but was significantly associated with maternal anxiety symptoms. Limitations: Participants were predominantly from affluent social groups and of non-minority status. Overprotection is a broad construct, the use of specific sub-dimensions of behavioural constructs may be preferable. Conclusions: The findings support the use of the OP measure to assess parental overprotection among 7–12 year-old children; however, they suggest that parental responses may be more closely related to the degree of parental rather than child anxiety.

A radial basis function network classifier to maximise leave-one-out mutual information

tWe develop an orthogonal forward selection (OFS) approach to construct radial basis function (RBF)network classifiers for two-class problems. Our approach integrates several concepts in probabilisticmodelling, including cross validation, mutual information and Bayesian hyperparameter fitting. At eachstage of the OFS procedure, one model term is selected by maximising the leave-one-out mutual infor-mation (LOOMI) between the classifier’s predicted class labels and the true class labels. We derive theformula of LOOMI within the OFS framework so that the LOOMI can be evaluated efficiently for modelterm selection. Furthermore, a Bayesian procedure of hyperparameter fitting is also integrated into theeach stage of the OFS to infer the l2-norm based local regularisation parameter from the data. Since eachforward stage is effectively fitting of a one-variable model, this task is very fast. The classifier construc-tion procedure is automatically terminated without the need of using additional stopping criterion toyield very sparse RBF classifiers with excellent classification generalisation performance, which is par-ticular useful for the noisy data sets with highly overlapping class distribution. A number of benchmarkexamples are employed to demonstrate the effectiveness of our proposed approach.

Parental death as a vital conjuncture? Intergenerational care and responsibility following bereavement in Senegal

This article explores the ways that parental death represents a 'vital conjuncture' for Serer young people that reconfigures and potentially transforms intergenerational caring responsibilities in different spatial and temporal contexts. Drawing on semi-structured interviews with young people (aged 15-27), family members, religious and community leaders and professionals in rural and urban Senegal, I explore young people's responses to parental death. 'Continuing bonds' with the deceased were expressed through memories evoked in homespace, shared family practices and gendered responsibilities to 'take care of' bereaved family members, to cultivate inherited farmland and to fulfil the wishes of the deceased. Parental death could reconfigure intergenerational care and lead to shifts in power dynamics, as eldest sons asserted their position of authority. While care-giving roles were associated with agency, the low social status accorded to young women's paid and unpaid domestic work undermined their efforts. The research contributes to understandings of gendered nuances in the experience of bereavement and continuing bonds and provides insight into intra-household decision-making processes, ownership and control of assets. Analysis of the culturally specific meanings of relationships and a young person's social location within hierarchies of gender, age, sibling birth order and wider socio-cultural norms and practices is needed.

A tale of two Montreal communities: parental perspectives on their children's multi-lingual and multi-literate development

This comparative inquiry examines the multi-/bilingual nature and cultural diversity of two distinctly different linguistic and ethnic communities in Montreal – English speakers and Chinese speakers – with a focus on the multi/bilingual and multi/biliterate development of children from these two communities who attend French-language schools, by choice in one case and by law in the other. In both of these communities, children traditionally achieve academic success. The authors approach this investigation from the perspective of the parents’ aspirations and expectations for, and their support of and involvement in, their children’s education. These two communities share key similarities and differences that, when considered together, help to clarify a number of issues involving multi/biliteracy development, socio-economic and linguistic capital, minority/majority language status, mother-tongue support, home–school continuities, and linguistic identity.