Clustering quality and topology preservation in fast learning SOMs

The Self-Organizing Map (SOM) is a popular unsupervised neural network able to provide effective clustering and data visualization for data represented in multidimensional input spaces. In this paper, we describe Fast Learning SOM (FLSOM) which adopts a learning algorithm that improves the performance of the standard SOM with respect to the convergence time in the training phase. We show that FLSOM also improves the quality of the map by providing better clustering quality and topology preservation of multidimensional input data. Several tests have been carried out on different multidimensional datasets, which demonstrate better performances of the algorithm in comparison with the original SOM.

Factors mediating the routinisation of e-learning within a traditional university education environment

Technology-enhanced or Computer Aided Learning (e-learning) can be institutionally integrated and supported by learning management systems or Virtual Learning Environments (VLEs) to offer efficiency gains, effectiveness and scalability of the e-leaning paradigm. However this can only be achieved through integration of pedagogically intelligent approaches and lesson preparation tools environment and VLE that is well accepted by both the students and teachers. This paper critically explores some of the issues relevant to scalable routinisation of e-learning at the tertiary level, typically first year university undergraduates, with the teaching of Relational Data Analysis (RDA), as supported by multimedia authoring, as a case study. The paper concludes that blended learning approaches which balance the deployment of e-learning with other modalities of learning delivery such as instructor–mediated group learning etc offer the most flexible and scalable route to e-learning but that this requires the graceful integration of platforms for multimedia production, distribution and delivery through advanced interactive spaces that provoke learner engagement and promote learning autonomy and group learning facilitated by a cooperative-creative learning environment that remains open to personal exploration of constructivist-constructionist pathways to learning.

Conceptualizing change in the Portuguese footwear industry: regional voices and cognitive spaces of agency

Expanding national services sectors and global competition aggravate current and perceived future market pressures on traditional manufacturing industries. These perceptions of change have provoked a growing intensification of geo-political discourses on technological innovation and ‘learning’, and calls for competency in design among other professional skills. However, these political discourses on innovation and learning have paralleled public concerns with the apparent ‘growth pains’ from factory closures and subsequent increases in unemployment, and its debilitating social and economic implications for local and regional development. In this respect the following investigation sets out to conceptualize change through the complementary and differing perceptions of industry and regional actors’ experiences or narratives, linking these perceptions to their structure-determined spheres of agent-environment interactivity. It aims to determine whether agents’ differing perceptions of industry transformation can have a role in the legitimization of their interests in, and in sustaining their organizational influence over the process of industry-regional transformation. It argues that industry and regional agent perceptions are among the cognitive aspects of agent-environment interactivity that permeate agency. It stresses agents’ ability to reason and manipulate their work environments to preserve their self-regulating interests in, and task representative influence over the multi-jurisdictional space of industry-regional transformation. The contributions of this investigation suggest that agents’ varied perceptions of industry and regional change inform or compete for influence over the redirection of regional, industry and business strategies. This claim offers a greater appreciation for the reflexive and complex institutional dimensions of industry planning and development, and the political responsibility to socially just forms of regional development. It positions the outcomes of this investigation at the nexus of intensifying geo-political discourses on the efficiency and equity of territorial development in Europe.

Application of complex extreme learning machine to multiclass classification problems with high dimensionality: a THz spectra classification problem

We extend extreme learning machine (ELM) classifiers to complex Reproducing Kernel Hilbert Spaces (RKHS) where the input/output variables as well as the optimization variables are complex-valued. A new family of classifiers, called complex-valued ELM (CELM) suitable for complex-valued multiple-input–multiple-output processing is introduced. In the proposed method, the associated Lagrangian is computed using induced RKHS kernels, adopting a Wirtinger calculus approach formulated as a constrained optimization problem similarly to the conventional ELM classifier formulation. When training the CELM, the Karush–Khun–Tuker (KKT) theorem is used to solve the dual optimization problem that consists of satisfying simultaneously smallest training error as well as smallest norm of output weights criteria. The proposed formulation also addresses aspects of quaternary classification within a Clifford algebra context. For 2D complex-valued inputs, user-defined complex-coupled hyper-planes divide the classifier input space into four partitions. For 3D complex-valued inputs, the formulation generates three pairs of complex-coupled hyper-planes through orthogonal projections. The six hyper-planes then divide the 3D space into eight partitions. It is shown that the CELM problem formulation is equivalent to solving six real-valued ELM tasks, which are induced by projecting the chosen complex kernel across the different user-defined coordinate planes. A classification example of powdered samples on the basis of their terahertz spectral signatures is used to demonstrate the advantages of the CELM classifiers compared to their SVM counterparts. The proposed classifiers retain the advantages of their ELM counterparts, in that they can perform multiclass classification with lower computational complexity than SVM classifiers. Furthermore, because of their ability to perform classification tasks fast, the proposed formulations are of interest to real-time applications.