Tandem system adaptation using multiple linear feature transforms

Adaptation to speaker and environment changes is an essential part of current automatic speech recognition (ASR) systems. In recent years the use of multi-layer percpetrons (MLPs) has become increasingly common in ASR systems. A standard approach to handling speaker differences when using MLPs is to apply a global speaker-specific constrained MLLR (CMLLR) transform to the features prior to training or using the MLP. This paper considers the situation when there are both speaker and channel, communication link, differences in the data. A more powerful transform, front-end CMLLR (FE-CMLLR), is applied to the inputs to the MLP to represent the channel differences. Though global, these FE-CMLLR transforms vary from time-instance to time-instance. Experiments on a channel distorted dialect Arabic conversational speech recognition task indicates the usefulness of adapting MLP features using both CMLLR and FE-CMLLR transforms. © 2013 IEEE.

Optimal displacement mechanisms beneath shallow foundations on linear-elastic perfectly plastic soil

An energy method for a linear-elastic perfectly plastic method utilising the von Mises yield criterion with associated flow developed in 2013 by McMahon and co-workers is used to compare the ellipsoidal cavity-expansion mechanism, from the same work, and the displacement fields of other research by Levin, in 1995, and Osman and Bolton, in 2005, which utilise the Hill and Prandtl mechanisms respectively. The energy method was also used with a mechanism produced by performing a linear-elastic finite-element analysis in Abaqus. At small values of settlement and soil rigidity the elastic mechanism provides the lowest upper-bound solution, and matches well with finite-element analysis results published in the literature. At typical footing working loads and settlements the cavity-expansion mechanism produces a more optimal solution than the displacement fields within the Hill and Prandtl mechanisms, and also matches well with the published finite-element analysis results in this range. Beyond these loads, at greater footing settlements, or soil rigidity, the Prandtl mechanism is shown to be the most appropriate.

Kernelized log linear models for continuous speech recognition

Large margin criteria and discriminative models are two effective improvements for HMM-based speech recognition. This paper proposed a large margin trained log linear model with kernels for CSR. To avoid explicitly computing in the high dimensional feature space and to achieve the nonlinear decision boundaries, a kernel based training and decoding framework is proposed in this work. To make the system robust to noise a kernel adaptation scheme is also presented. Previous work in this area is extended in two directions. First, most kernels for CSR focus on measuring the similarity between two observation sequences. The proposed joint kernels defined a similarity between two observation-label sequence pairs on the sentence level. Second, this paper addresses how to efficiently employ kernels in large margin training and decoding with lattices. To the best of our knowledge, this is the first attempt at using large margin kernel-based log linear models for CSR. The model is evaluated on a noise corrupted continuous digit task: AURORA 2.0. © 2013 IEEE.

Non linear constitutive models for lattice materials

We use a computational homogenisation approach to derive a non linear constitutive model for lattice materials. A representative volume element (RVE) of the lattice is modelled by means of discrete structural elements, and macroscopic stress-strain relationships are numerically evaluated after applying appropriate periodic boundary conditions to the RVE. The influence of the choice of the RVE on the predictions of the model is discussed. The model has been used for the analysis of the hexagonal and the triangulated lattices subjected to large strains. The fidelity of the model has been demonstrated by analysing a plate with a central hole under prescribed in plane compressive and tensile loads, and then comparing the results from the discrete and the homogenised models. © 2013 Elsevier Ltd.

Experimental evaluation of layout designs for 3 × 3 MIMO-enabled radio-over-fiber distributed antenna systems

This paper experimentally demonstrates that, for two representative indoor distributed antenna system (DAS) scenarios, existing radio-over-fiber (RoF) DAS installations can enhance the capacity advantages of broadband 3 × 3 multiple-input-multiple-output (MIMO) radio services without requiring additional fibers or multiplexing schemes. This is true for both single-and multiple-user cases with a single base station and multiple base stations. First, a theoretical example is used to illustrate that there is a negligible improvement in signal-to-noise ratio (SNR) when using a MIMO DAS with all N spatial streams replicated at N RAUs, compared with a MIMO DAS with only one of the N streams replicated at each RAU for N ≤ 4. It is then experimentally confirmed that a 3 × 3 MIMO DAS offers improved capacity and throughput compared with a 3 × 3 MIMO collocated antenna system (CAS) for the single-user case in two typical indoor DAS scenarios, i.e., one with significant line-of-sight (LOS) propagation and the other with entirely non-line-of-sight (NLOS) propagation. The improvement in capacity is 3.2% and 4.1%, respectively. Then, experimental channel measurements confirm that there is a negligible capacity increase in the 3 × 3 configuration with three spatial streams per antenna unit over the 3 × 3 configuration with a single spatial stream per antenna unit. The former layout is observed to provide an increase of ∼1% in the median channel capacity in both the single-and multiple-user scenarios. With 20 users and three base stations, a MIMO DAS using the latter layout offers median aggregate capacities of 259 and 233 bit/s/Hz for the LOS and NLOS scenarios, respectively. It is concluded that DAS installations can further enhance the capacity offered to multiple users by multiple 3 × 3 MIMO-enabled base stations. Further, designing future DAS systems to support broadband 3 × 3 MIMO systems may not require significant upgrades to existing installations for small numbers of spatial streams. © 2013 IEEE.

Global linear stability of the boundary-layer flow over a rotating sphere

We consider the linear global stability of the boundary-layer flow over a rotating sphere. Our results suggest that a self-excited linear global mode can exist when the sphere rotates sufficiently fast, with properties fixed by the flow at latitudes between approximately 55°-65° from the pole (depending on the rotation rate). A neutral curve for global linear instabilities is presented with critical Reynolds number consistent with existing experimentally measured values for the appearance of turbulence. The existence of an unstable linear global mode is in contrast to the literature on the rotating disk, where it is expected that nonlinearity is required to prompt the transition to turbulence. Despite both being susceptible to local absolute instabilities, we conclude that the transition mechanism for the rotating-sphere flow may be different to that for the rotating disk. © 2014 Elsevier Masson SAS. All rights reserved.

Preloaded hopping with linear multi-modal actuation

For more dexterous and agile legged robot locomotion, alternative actuation has been one of the most long-awaited technologies. The goal of this paper is to investigate the use of newly developed actuator, the so-called Linear Multi-Modal Actuator (LMMA), in the context of legged robot locomotion, and analyze the behavioral performance of it. The LMMA consists of three discrete couplings which enable the system to switch between different mechanical dynamics such as instantaneous switches between series elastic and fully actuated dynamics. To test this actuator for legged locomotion, this paper introduces a one-legged robot platform we developed to implement the actuator, and explains a novel control strategy for hopping, i.e. 'preloaded hopping control'. This control strategy takes advantage of the coupling mechanism of the LMMA to preload the series elasticity during the flight phase to improve the energy efficiency of hopping locomotion. This paper shows a series of experimental results that compare the control strategy with a simple sinusoidal actuation strategy to discuss the benefits and challenges of the proposed approach. © 2013 IEEE.

Linear multimodal actuation through discrete coupling

Due to technological limitations, robot actuators are often designed for specific tasks with narrow performance goals, whereas a wide range of behaviors is necessary for autonomous robots in uncertain complex environments. In an effort to increase the versatility of actuators, we introduce a new concept of multimodal actuation (MMA) that employs dynamic coupling in the form of clutches and brakes to change its mode of operation. The dynamic coupling allows motors and passive elements such as springs to be engaged and disengaged within a single actuator. We apply the concept to a linear series elastic actuator which uses friction brakes controlled online for the dynamic coupling. With this prototype, we are able to demonstrate several modes of operation including stiff position control, series elastic actuation as well as the possibility to store and release energy in a controlled manner for explosive tasks such as jumping. In this paper, we model the proposed concept of actuation and show a systematic performance analysis of the physical prototype that we developed in our laboratory. © 1996-2012 IEEE.

Linear multi-modal actuation through discrete coupling

Due to technological limitations robot actuators are often designed for specific tasks with narrow performance goals, whereas a wide range of output and behaviours is necessary for robots to operate autonomously in uncertain complex environments. We present a design framework that employs dynamic couplings in the form of brakes and clutches to increase the performance and diversity of linear actuators. The couplings are used to switch between a diverse range of discrete modes of operation within a single actuator. We also provide a design solution for miniaturized couplings that use dry friction to produce rapid switching and high braking forces. The couplings are designed so that once engaged or disengaged no extra energy is consumed. We apply the design framework and coupling design to a linear series elastic actuator (SEA) and show that this relatively simple implementation increases the performance and adds new behaviours to the standard design. Through a number of performance tests we are able to show rapid switching between a high and a low impedance output mode; that the actuator's spring can be charged to produce short bursts of high output power; and that the actuator has additional passive and rigid modes that consume no power once activated. Robots using actuators from this design framework would see a vast increase in their behavioural diversity and improvements in their performance not yet possible with conventional actuator design. © 2012 IEEE.

Linear dimensionality reduction: Survey, insights, and generalizations

© 2015 John P. Cunningham and Zoubin Ghahramani. Linear dimensionality reduction methods are a cornerstone of analyzing high dimensional data, due to their simple geometric interpretations and typically attractive computational properties. These methods capture many data features of interest, such as covariance, dynamical structure, correlation between data sets, input-output relationships, and margin between data classes. Methods have been developed with a variety of names and motivations in many fields, and perhaps as a result the connections between all these methods have not been highlighted. Here we survey methods from this disparate literature as optimization programs over matrix manifolds. We discuss principal component analysis, factor analysis, linear multidimensional scaling, Fisher's linear discriminant analysis, canonical correlations analysis, maximum autocorrelation factors, slow feature analysis, sufficient dimensionality reduction, undercomplete independent component analysis, linear regression, distance metric learning, and more. This optimization framework gives insight to some rarely discussed shortcomings of well-known methods, such as the suboptimality of certain eigenvector solutions. Modern techniques for optimization over matrix manifolds enable a generic linear dimensionality reduction solver, which accepts as input data and an objective to be optimized, and returns, as output, an optimal low-dimensional projection of the data. This simple optimization framework further allows straightforward generalizations and novel variants of classical methods, which we demonstrate here by creating an orthogonal-projection canonical correlations analysis. More broadly, this survey and generic solver suggest that linear dimensionality reduction can move toward becoming a blackbox, objective-agnostic numerical technology.

A non-linear perturbation model considering catchment wetness and its application in river flow forecasting

A non-linear perturbation model for river flow forecasting is developed, based on consideration of catchment wetness using an antecedent precipitation index (API). Catchment seasonality, of the form accounted for in the linear perturbation model (the LPM), and non-linear behaviour both in the runoff generation mechanism and in the flow routing processes are represented by a constrained nan-linear model, the NLPM-API. A total of ten catchments, across a range of climatic conditions and catchment area magnitudes, located in China and in other countries, were selected for testing daily rainfall-runoff forecasting with this model. It was found that the NLPM-API model was significantly more efficient than the original linear perturbation model (the LPM). However, restric tion of explicit nan-linearity to the runoff generation process, in the simpler LPM-API form of the model, did not produce a significantly lower value of the efficiency in flood forecasting, in terms of the model efficiency index R-2. (C) 1997 Elsevier Science B.V.

INFLUENCE OF LINEAR ALKYLBENZENE SULFONATE (LAS) AS ORGANIC COSOLVENT ON LEACHING BEHAVIOR OF PCDD/FS FROM FLY-ASH AND SOIL

The leaching of polychlorinated dibenzo-p-dioxins and dibenzofurans (PCDD/Fs) was measured in soil and standard fly ash column eluted with pure water and linear alkylbenzene sulfonate (LAS)- water. The data obtained were used to evaluate the leachability of PCDD/Fs from waste dump like incineration residual slag and fly ash deposition. The leaching rate was shown to be increased significantly by using LAS water. The leachate contents of PCDD/Fs were above their known water solubility. Concentration of PCDD/Fs in the leachates as well as the relative leaching (calculated on the fly ash content) increased with increasing chlorinating degree and decreasing water solubility. LAS above the critical micelle concentration (CMC) probably enhances PCDD/Fs solubility.

GENOTOXIC EFFECTS OF LINEAR ALKYL BENZENE SULFONATE, SODIUM PENTACHLOROPHENATE AND DICHROMATE ON TETRAHYMENA-PYRIFORMIS

DNA in macro- and micronuclei of Tetrahymena pyriformis treated with linear alkyl benzene sulfonate (LAS) and sodium pentachlorophenate (PCP-Na) were determined by microspectrophotometry. The effects on rate of formation of macronuclear DNA extrusion bodies were also studied. We found DNA content of micronuclei in 0.14 ppm LAS and 0.9 ppb PCP-Na was lower than in that of the control, and LAS was able to increase the formation rate of macronuclear DNA extrusion bodies (the formation rate was 54% in 11.3 ppm LAS and 25.6% in 16.7 ppm dichromate). We concluded that 0.14 ppm LAS (below the maximum acceptable toxicant concentration) was genotoxic, whereas 0.014 ppm LAS was not. Dichromate 0.05 ppm and 0.9 ppb PCP-Na, equal to and below the maximum acceptable toxicant concentration, respectively, were potentially genetoxic.

Weak measurement of qubit oscillations with strong response detectors: Violation of the fundamental bound imposed on linear detectors

Qubit measurement by mesoscopic charge detectors has received great interest in the community of mesoscopic transport and solid-state quantum computation, and some controversial issues still remain unresolved. In this work, we revisit the continuous weak measurement of a solid-state qubit by single electron transistors (SETs) in nonlinear-response regime. For two SET models typically used in the literature, we find that the signal-to-noise ratio can violate the universal upper bound "4," which is imposed quantum mechanically on linear-response detectors. This different result can be understood by means of the cross correlation of the detector currents by viewing the two junctions of the single SET as two detectors. Possible limitation of the potential-scattering approach to this result is also discussed.