Modified radial basis function neural network using output transformation

A modified radial basis function (RBF) neural network and its identification algorithm based on observational data with heterogeneous noise are introduced. The transformed system output of Box-Cox is represented by the RBF neural network. To identify the model from observational data, the singular value decomposition of the full regression matrix consisting of basis functions formed by system input data is initially carried out and a new fast identification method is then developed using Gauss-Newton algorithm to derive the required Box-Cox transformation, based on a maximum likelihood estimator (MLE) for a model base spanned by the largest eigenvectors. Finally, the Box-Cox transformation-based RBF neural network, with good generalisation and sparsity, is identified based on the derived optimal Box-Cox transformation and an orthogonal forward regression algorithm using a pseudo-PRESS statistic to select a sparse RBF model with good generalisation. The proposed algorithm and its efficacy are demonstrated with numerical examples.

The effects of maternal postnatal depression and child sex on academic performance at age 16 years: a developmental approach

Background: Postnatal depression (PND) is associated with poor cognitive functioning in infancy and the early school years; long-term effects on academic outcome are not known. Method: Children of postnatally depressed (N = 50) and non-depressed mothers (N = 39), studied from infancy, were followed up at 16 years. We examined the effects on General Certificate of Secondary Education (GCSE) exam performance of maternal depression (postnatal and subsequent) and IQ, child sex and earlier cognitive development, and mother–child interactions, using structural equation modelling (SEM). Results: Boys, but not girls, of PND mothers had poorer GCSE results than control children. This was principally accounted for by effects on early child cognitive functioning, which showed strong continuity from infancy. PND had continuing negative effects on maternal interactions through childhood, and these also contributed to poorer GCSE performance. Neither chronic, nor recent, exposure to maternal depression had significant effects. Conclusions: The adverse effects of PND on male infants’ cognitive functioning may persist through development. Continuing difficulties in mother–child interactions are also important, suggesting that both early intervention and continuing monitoring of mothers with PND may be warranted.

Teaching assistants and pupils' academic and social engagement in mainstream schools: insights from systematic literature reviews

The last 20 years have seen a huge expansion in the additional adults working in classrooms in the UK, USA, and other countries. This paper presents the findings of a series of systematic literature reviews about teaching assistants. The first two reviews focused on stakeholder perceptions of teaching assistant contributions to academic and social engagement. Stakeholders were pupils, teachers, TAs, headteachers and parents. Perceptions focused on four principal contributions that teaching assistants contribute to: pupils’ academic and socio-academic engagement; inclusion; maintenance of stakeholder relations; and support for the teacher. The third review explored training. Against a background of patchy training provision both in the UK and the USA, strong claims are made for the benefits to TAs of training provided, particularly in building confidence and skills. The conclusions include implications for further training and the need for further research to gain an in-depth understanding as to precisely the manner in which TAs engage with children.

Input/output linearization using dynamic recurrent neural networks

This paper brings together two areas of research that have received considerable attention during the last years, namely feedback linearization and neural networks. A proposition that guarantees the Input/Output (I/O) linearization of nonlinear control affine systems with Dynamic Recurrent Neural Networks (DRNNs) is formulated and proved. The proposition and the linearization procedure are illustrated with the simulation of a single link manipulator.

Performance analysis of the LMS blind minimum-output-energy CDMA detector

Adaptive filters used in code division multiple access (CDMA) receivers to counter interference have been formulated both with and without the assumption of training symbols being transmitted. They are known as training-based and blind detectors respectively. We show that the convergence behaviour of the blind minimum-output-energy (MOE) detector can be quite easily derived, unlike what was implied by the procedure outlined in a previous paper. The simplification results from the observation that the correlation matrix determining convergence performance can be made symmetric, after which many standard results from the literature on least mean square (LMS) filters apply immediately.