Exploiting variable-width features in large vocabulary speech recognition

The use of variable-width features (prosodics, broad structural information etc.) in large vocabulary speech recognition systems is discussed. Although the value of this sort of information has been recognized in the past, previous approaches have not been widely used in speech systems because either they have not been robust enough for realistic, large vocabulary tasks or they have been limited to certain recognizer architectures. A framework for the use of variable-width features is presented which employs the N-Best algorithm with the features being applied in a post-processing phase. The framework is flexible and widely applicable, giving greater scope for exploitation of the features than previous approaches. Large vocabulary speech recognition experiments using TIMIT show that the application of variable-width features has potential benefits.

Observation of large-scale features on graphite by scanning tunnelling microscopy

Superlattice structures and rippling fringes were imaged on two separate pieces of graphite (HOPG) by scanning tunnelling microscopy (STM). We observed the corrugation conservation phenomenon on one of the superlattice structures where an overlayer does not attenuate the corrugation amplitude of the superlattice. Such a phenomenon may illustrate an implication that nanoscale defects a few layers underneath the surface may propagate through many layers without decay and form the superlattice structure on the topmost surface. Some rippling fringes with periodicities of 20 nm and 30 nm and corrugations of 0.1 nm and 0.15nm were observed in the superlattice area and in nearby regions. Such fringes are believed to be due to physical buckling of the surface. The stress required to generate such structures is estimated, and a possible cause is discussed. An equation relating the attenuation factor to the number of overlayers is proposed. © 2005 The Japan Society of Applied Physics.

Numerical analysis of the current and voltage sharing issues for resistive fault current limiter using YBCO coated conductors

YBaCuO-coated conductors offer great potential in terms of performance and cost-saving for superconducting fault current limiter (SFCL). A resistive SFCL based on coated conductors can be made from several tapes connected in parallel or in series. Ideally, the current and voltage are shared uniformly by the tapes when quench occurs. However, due to the non-uniformity of property of the tapes and the relative positions of the tapes, the currents and the voltages of the tapes are different. In this paper, a numerical model is developed to investigate the current and voltage sharing problem for the resistive SFCL. This model is able to simulate the dynamic response of YBCO tapes in normal and quench conditions. Firstly, four tapes with different Jc 's and n values in E-J power law are connected in parallel to carry the fault current. The model demonstrates how the currents are distributed among the four tapes. These four tapes are then connected in series to withstand the line voltage. In this case, the model investigates the voltage sharing between the tapes. Several factors that would affect the process of quenches are discussed including the field dependency of Jc, the magnetic coupling between the tapes and the relative positions of the tapes. © 2010 IEEE.

Integrated billing mechanisms in the Internet of Things to support information sharing and enable new business opportunities

The Internet of Things (IOT) concept and enabling technologies such as RFID offer the prospect of linking the real world of physical objects with the virtual world of information technology to improve visibility and traceability information within supply chains and across the entire lifecycles of products, as well as enabling more intuitive interactions and greater automation possibilities. There is a huge potential for savings through process optimization and profit generation within the IOT, but the sharing of financial benefits across companies remains an unsolved issue. Existing approaches towards sharing of costs and benefits have failed to scale so far. The integration of payment solutions into the IOT architecture could solve this problem. We have reviewed different possible levels of integration. Multiple payment solutions have been researched. Finally we have developed a model that meets the requirements of the IOT in relation to openness and scalability. It supports both hardware-centric and software-centric approaches to integration of payment solutions with the IOT. Different requirements concerning payment solutions within the IOT have been defined and considered in the proposed model. Possible solution providers include telcos, e-payment service providers and new players such as banks and standardization bodies. The proposed model of integrating the Internet of Things with payment solutions will lower the barrier to invoicing for the more granular visibility information generated using the IOT. Thus, it has the potential to enable recovery of the necessary investments in IOT infrastructure and accelerate adoption of the IOT, especially for projects that are only viable when multiple benefits throughout the supply chain need to be accumulated in order to achieve a Return on Investment (ROI). In a long-term perspective, it may enable IT-departments to become profit centres instead of cost centres. © 2010 - IOS Press and the authors. All rights reserved.

Multiscale object features from clustered complex wavelet coefficients

This paper introduces a method by which intuitive feature entities can be created from ILP (InterLevel Product) coefficients. The ILP transform is a pyramid of decimated complex-valued coefficients at multiple scales, derived from dual-tree complex wavelets, whose phases indicate the presence of different feature types (edges and ridges). We use an Expectation-Maximization algorithm to cluster large ILP coefficients that are spatially adjacent and similar in phase. We then demonstrate the relationship that these clusters possess with respect to observable image content, and conclude with a look at potential applications of these clusters, such as rotation- and scale-invariant object recognition. © 2005 IEEE.