Relative predictive value of intraventricular dysynchrony in determining long-term outcome in patients with known or suspected coronary disease

Improvement of intra-ventricular dysynchrony (IVD) in pts undergoing bi-ventricular pacing is associated with clinical improvementbut little isknownabout the relationship between IVD and prognosis.We sought whether IVD influences long-term outcome in pts with known or suspected coronary disease (CAD). Tissue Doppler imaging was performed in 184 pts (aged 61±10 years, 67% male) prior to dobutamine echo. From velocity curves the interval between QRS onset and max systolic velocity (Ts) was measured in basal septal, lateral, inferior and anterior segments. The maximal difference in Ts between segments (TsMax) was used as a measure of IVD. The standard deviation (TsSD) between all segments and the septal-lateral difference (TsSL) were also calculated. Pts were followed up for a median interval of 5 years and a Cox model used for survival analysis. The medianwall motion index (WMI) was 1.3 (IQR 1.0–1.8) at rest and 1.4 (IQR 1.3–1.9) at stress. The table shows IVD parameters. Forty-one deaths occurred during follow-up. Pts who died during follow-up, compared to survivors, showed greater IVD. WMI at rest (p = 0.03) and peak stress (p = 0.02), TsSD (p = 0.06), TsSL (p = 0.02) and TsMax (p = 0.05) but not QRS width were univariate predictors of mortality. TsSL was the only independent predictor of death (p = 0.01). Therefore, IVD is common in pts with known or suspected CAD. Pts with more IVD have reduced long-term survival, independent of WMI.

Work at hand: an exploration of gesture in the context of work and everyday life to inform the design of gestural input devices

This research explores gestures used in the context of activities in the workplace and in everyday life in order to understand requirements and devise concepts for the design of gestural information appliances. A collaborative method of video interaction analysis devised to suit design explorations, the Video Card Game, was used to capture and analyse how gesture is used in the context of six different domains: the dentist's office; PDA and mobile phone use; the experimental biologist's laboratory; a city ferry service; a video cassette player repair shop; and a factory flowmeter assembly station. Findings are presented in the form of gestural themes, derived from the tradition of qualitative analysis but bearing some similarity to Alexandrian patterns. Implications for the design of gestural devices are discussed.

Efficient training of RBF networks for classification.

Radial Basis Function networks with linear outputs are often used in regression problems because they can be substantially faster to train than Multi-layer Perceptrons. For classification problems, the use of linear outputs is less appropriate as the outputs are not guaranteed to represent probabilities. We show how RBFs with logistic and softmax outputs can be trained efficiently using the Fisher scoring algorithm. This approach can be used with any model which consists of a generalised linear output function applied to a model which is linear in its parameters. We compare this approach with standard non-linear optimisation algorithms on a number of datasets.

Bayesian training of mixture density networks

Mixture Density Networks (MDNs) are a well-established method for modelling the conditional probability density which is useful for complex multi-valued functions where regression methods (such as MLPs) fail. In this paper we extend earlier research of a regularisation method for a special case of MDNs to the general case using evidence based regularisation and we show how the Hessian of the MDN error function can be evaluated using R-propagation. The method is tested on two data sets and compared with early stopping.

Efficient training of RBF networks for classification

Radial Basis Function networks with linear outputs are often used in regression problems because they can be substantially faster to train than Multi-layer Perceptrons. For classification problems, the use of linear outputs is less appropriate as the outputs are not guaranteed to represent probabilities. In this paper we show how RBFs with logistic and softmax outputs can be trained efficiently using algorithms derived from Generalised Linear Models. This approach is compared with standard non-linear optimisation algorithms on a number of datasets.

Sensing characteristics of a novel two-section long-period grating

The behavior of a temperature self-compensating, fiber, long-period grating (LPG) device is studied. This device consists of a single 325-µm-period LPG recorded across two sections of a single-mode B-Ge-codoped fiber—one section bare and the other coated with a 1-µm thickness of Ag. This structure generates two attenuation bands associated with the eighth and ninth cladding modes, which are spectrally close together (~60 nm). The attenuation band associated with the Ag-coated section is unaffected by changes in the refractive index of the surrounding medium and can be used to compensate for the temperature of the bare-fiber section. The sensor has a resolution of ±1.0 × 10-3 for the refractive index and ±0.3 °C for the temperature. The effect of bending on the spectral characteristics of the two attenuation bands was found to be nonlinear, with the Ag-coated LPG having the greater sensitivity.

Implementation of vectorial bend sensors using long-period gratings UV-inscribed in special shape fibres

We report the implementation of vector bending sensors using long-period gratings (LPGs) UV-inscribed in flat-clad, four-core and D-shaped fibres. Our experiments reveal a strong fibre-orientation dependence of the spectral response when such LPGs are subjected to dynamic bending, which provided an opportunity to realize curvature measurement with direction recognition.

Discriminative training of the hidden vector state model for semantic parsing

In this paper, we discuss how discriminative training can be applied to the hidden vector state (HVS) model in different task domains. The HVS model is a discrete hidden Markov model (HMM) in which each HMM state represents the state of a push-down automaton with a finite stack size. In previous applications, maximum-likelihood estimation (MLE) is used to derive the parameters of the HVS model. However, MLE makes a number of assumptions and unfortunately some of these assumptions do not hold. Discriminative training, without making such assumptions, can improve the performance of the HVS model by discriminating the correct hypothesis from the competing hypotheses. Experiments have been conducted in two domains: the travel domain for the semantic parsing task using the DARPA Communicator data and the Air Travel Information Services (ATIS) data and the bioinformatics domain for the information extraction task using the GENIA corpus. The results demonstrate modest improvements of the performance of the HVS model using discriminative training. In the travel domain, discriminative training of the HVS model gives a relative error reduction rate of 31 percent in F-measure when compared with MLE on the DARPA Communicator data and 9 percent on the ATIS data. In the bioinformatics domain, a relative error reduction rate of 4 percent in F-measure is achieved on the GENIA corpus.

Impact of power allocation strategies in long-haul few-mode fiber transmission systems

We report for the first time on the limitations in the operational power range of few-mode fiber based transmission systems, employing 28Gbaud quadrature phase shift keying transponders, over 1,600km. It is demonstrated that if an additional mode is used on a preexisting few-mode transmission link, and allowed to optimize its performance, it will have a significant impact on the pre-existing mode. In particular, we show that for low mode coupling strengths (weak coupling regime), the newly added variable power mode does not considerably impact the fixed power existing mode, with performance penalties less than 2dB (in Q-factor). On the other hand, as mode coupling strength is increased (strong coupling regime), the individual launch power optimization significantly degrades the system performance, with penalties up to ∼6dB. Our results further suggest that mutual power optimization, of both fixed power and variable power modes, reduces power allocation related penalties to less than 3dB, for any given coupling strength, for both high and low differential mode delays. © 2013 Optical Society of America.