Scalable nonparametric Bayesian multilevel clustering

Multilevel clustering problems where the con-tent and contextual information are jointly clustered are ubiquitous in modern datasets. Existing works on this problem are limited to small datasets due to the use of the Gibbs sampler. We address the problem of scaling up multi-level clustering under a Bayesian nonparametric setting, extending the MC2 model proposed in (Nguyen et al., 2014). We ground our approach in structured mean-field and stochastic variational inference (SVI) and develop a tree-structured SVI algorithm that exploits the interplay between content and context modeling. Our new algorithm avoids the need to repeatedly go through the corpus as in Gibbs sampler. More crucially, our method is immediately amendable to parallelization, facilitating a scalable distributed implementation on the Apache Spark platform. We conduct extensive experiments in a variety of domains including text, images, and real-world user application activities. Direct comparison with the Gibbs-sampler demonstrates that our method is an order-of-magnitude faster without loss of model quality. Our Spark-based implementation gains an-other order-of-magnitude speedup and can scale to large real-world datasets containing millions of documents and groups.

Numerical investigation of the turbulent energy budget in the wake of freely oscillating elastically mounted cylinder at low reduced velocities

We present a numerical study of the turbulent kinetic energy budget in the wake of cylinders undergoing Vortex-Induced Vibration (VIV). We show three-dimensional Large Eddy Simulations (LES) of an elastically mounted circular cylinder in the synchronization regime at Reynolds number of Re=8000. The Immersed Boundary Method (IBM) is used to account for the presence of the cylinder. The flow field in the wake is decomposed using the triple decomposition splitting the flow variables in mean, coherent and stochastic components. The energy transfer between these scales of motions are then studied and the results of the free oscillation are compared to those of a forced oscillation. The turbulent kinetic energy budget shows that the maximum amplitude of VIV is defined by the ability of the mean flow to feed energy to the coherent structures in the wake. At amplitudes above this maximum amplitude, the energy of the coherent structures needs to be fed additionally by small scale, stochastic energy in form of backscatter to sustain its motion. Furthermore, we demonstrate that the maximum amplitude of the VIV is defined by the integral length scale of the turbulence in the wake

MMSE precoder for unitary space–time codes in correlated time-varying channels

This letter addresses the problem of the design of a precoder for multiple transmit antenna communication systems with spatially and temporally correlated fading channels. By using the asymptotic (high signal-to-noise ratio) mean-square error of the channel estimates, the letter derives a precoder for unitary space-time codes that can exploit the spatiotemporal correlation in the time-varying fading channels. Simulation results illustrate that significant performance gains can be achieved by using the new precoder.

Improving the performance of the LMS and RLS algorithms for adaptive equalizer

In this paper, we present the experiment results of three adaptive equalization algorithms: least-mean-square (LMS) algorithm, discrete cosine transform-least mean square (DCT-LMS) algorithm, and recursive least square (RLS) algorithm. Based on the experiments, we obtained that the convergence rate of LMS is slow; the convergence rate of RLS is great faster while the computational price is expensive; the performance of that two parameters of DCT-LMS are between the previous two algorithms, but still not good enough. Therefore we will propose an algorithm based on H2 in a coming paper to solve the problems.

Application of fast and robust equalization in communication technology

In this paper, the authors explore the potential of several popular equalization techniques while overcoming their disadvantages. First, extensive literature survey on equalization is conducted. The focus is on popular linear equalization algorithms such as the conventional least-mean-square (LMS) algorithm , The recursive least-squares (RLS) algorithm, the filtered-X LMS algorithm and their development. To overcome the slow convergence problem while keeping the simplicity of the LMS based algorithms, an H2 optimal initialization is proposed.

A modified CM algorithm for blind equalization of MPSK signals

A new blind equalization algorithm for application to wireless communication employing MPSK signals is proposed in this paper.  Since the new cost function exploits the amplitude and phase information simultaneously, the proposed algorithm can provide a superior performance than the conventional constant modulus algorithm (CMA) which only use the amplitude knowledge in its cost function.  Theoretical analysis and numerical simulations both demonstrate that the steady-state mean square error (MSE) for the proposed algorithm is less than that of the CMA.

Adaptive blind channel equalization based on constellation information of MPSK signal

In this paper, we propose a new adaptive algorithm for the blind equalization of an FIR (finite impulse response) channel excited by an M-ary phase shift keying (MPSK) signal. Different from the conventional constant modulus algorithm (CMA), which exploits the amplitude information of the input signal, the proposed algorithm exploits the full constellation information of the input signal. Theoretical analysis shows that the new algorithm has less mean square error (MSE), namely better equalization performance, in steady state than the CMA. Numerical simulations show the effectiveness of the new algorithm

Precoder design for multi-antenna systems with temporally and spatially correlated fading channels

This paper addresses the problem of the design of a precoder for multiple transmit antenna communication systems with spatially and temporally correlated fading channels. Using the theories of matrix differential calculus, the paper derives a precoder for unitary space-time codes that can exploit the spatio-temporal correlation in the time-varying fading channels. The design criterion is based on minimizing the mean square error of the channel estimates. Computer simulation results show that a significant performance gain can be achieved by using the designed precoder.

A new blind signal separation algorithm for instantaneous MIMO system

We address the problem of adaptive blind source separation (BSS) from instantaneous multi-input multi-output (MIMO) channels. In this paper, we propose a new constant modulus (CM)-based algorithm which employ nonlinear function as the de-correlation term. Moreover, it is shown by theoretical analysis that the proposed algorithm has less mean square error (MSE), i.e., better separation performance, in steady state than the cross-correlation and constant modulus algorithm (CC-CMA). Numerical simulations show the effectiveness of the proposed result.

A new blind-equalization algorithm for an FIR SIMO system driven by MPSK signal

We consider the problem of blind equalization of a finite impulse response and single-input multiple-output system driven by an M-ary phase-shift-keying signal. The existing single-mode algorithms for this problem include the constant modulus algorithm (CMA) and the multimodulus algorithm (MMA). It has been shown that the MMA outperforms the CMA when the input signal has no more than four constellation points, i.e., Mles4. In this brief, we present a new adaptive equalization algorithm that jointly exploits the amplitude and phase information of the input signal. Theoretical analysis shows that the proposed algorithm has less mean square error, i.e., better equalization performance, at steady state than the CMA regardless of the value of M. The superior performance of our algorithm to the CMA and the MMA is validated by simulation examples

Application of radial return mapping algorithm for finite strain elastoplasticity in a hybrid finite element and particle-in-cell model

The radial return mapping algorithm within the computational context of a hybrid Finite Element and Particle-In-Cell (FE/PIC) method is constructed to allow a fluid flow FE/PIC code to be applied solid mechanic problems with large displacements and large deformations. The FE/PIC method retains the robustness of an Eulerian mesh and enables tracking of material deformation by a set of Lagrangian particles or material points. In the FE/PIC approach the particle velocities are interpolated from nodal velocities and then the particle position is updated using a suitable integration scheme, such as the 4th order Runge-Kutta scheme[1]. The strain increments are obtained from gradients of the nodal velocities at the material point positions, which are then used to evaluate the stress increment and update history variables. To obtain the stress increment from the strain increment, the nonlinear constitutive equations are solved in an incremental iterative integration scheme based on a radial return mapping algorithm[2]. A plane stress extension of a rectangular shape J2 elastoplastic material with isotropic, kinematic and combined hardening is performed as an example and for validation of the enhanced FE/PIC method. It is shown that the method is suitable for analysis of problems in crystal plasticity and metal forming. The method is specifically suitable for simulation of neighbouring microstructural phases with different constitutive equations in a multiscale material modelling framework.

Variable step-size LMS algorithm with a quotient form

An improved robust variable step-size least mean square (LMS) algorithm is developed in this paper. Unlike many existing approaches, we adjust the variable step-size using a quotient form of filtered versions of the quadratic error. The filtered estimates of the error are based on exponential windows, applying different decaying factors for the estimations in the numerator and denominator. The new algorithm, called more robust variable step-size (MRVSS), is able to reduce the sensitivity to the power of the measurement noise, and improve the steady-state performance for comparable transient behavior, with negligible increase in the computational cost. The mean convergence, the steady-state performance and the mean step-size behavior of the MRVSS algorithm are studied under a slow time-varying system model, which can be served as guidelines for the design of MRVSS algorithm in practical applications. Simulation results are demonstrated to corroborate the analytic results, and to compare MRVSS with the existing representative approaches. Superior properties of the MRVSS algorithm are indicated.

A class of modified variable step-size NLMS algorithms for system identification

This paper proposes a class of modified variable step-size normalized least mean square (VS NLMS) algorithms. The class of schemes are obtained from estimating the optimum step-size of NLMS that minimizes the mean square deviation (MSD). During the estimation, we consider the properties of the additive noise and the input excitation together. The developed class of VS NLMS algorithms have simple forms and give improved tradeoff of fast convergence rate and low misadjustment in system identification.

An approach to nonirreducible MIMO FIR channel equalization

It is known that a nonirreducible multiple-input– multiple-output finite-impulse-response channel driven by colored signals that are mutually uncorrelated and of sufficiently diverse power spectra can be identified blindly by exploiting only the second-order statistics of the measured data. In this brief, we propose an approach to dealing with the equalization of a nonirreducible channel, provided that the estimate of the channel matrix is available. Both zero-forcing and minimum-mean-square-error equalizers are developed to perform the channel equalization. The effectiveness of the approach and equalizers is demonstrated by simulation examples.

Neurophysiological responses after short-term strength training of the biceps brachii muscle

The neural adaptations that mediate the increase in strength in the early phase of a strength training program are not well understood; however, changes in neural drive and corticospinal excitability have been hypothesized. To determine the neural adaptations to strength training, we used transcranial magnetic stimulation (TMS) to compare the effect of strength training of the right elbow flexor muscles on the functional properties of the corticospinal pathway. Motorevoked potentials (MEPs) were recorded from the right biceps brachii (BB) muscle from 23 individuals (training group; n = 13 and control group; n = 10) before and after 4 weeks of progressive overload strength training at 80% of 1-repetition maximum (1 RM). The TMS was delivered at 10% of the root mean square electromyographic signal (rmsEMG) obtained from a maximal voluntary contraction (MVC) at intensities of 5% of stimulator output below active motor threshold (AMT) until saturation of the MEP (MEP maxl. Strength training resulted in a 28% (p = 0.0001) increase in 1 RM strength, and this was accompanied by a 53% increase (p = 0.05) in the amplitude of the MEP at AMT; 33% (p = 0.05) increase in MEP at 20% above AMT, and a 38% increase at MEPmax (p = 0.04). There were no significant differences in the estimated slope (p = 0.4 7) or peak slope of the stimulus-response curve for the left primary motor cortex (M1) after strength training (p = 0.61). These results demonstrate that heavy-load isotonic strength training alters neural transmission via the corticospinal pathway projecting to the motoneurons controlling BB and in part underpin the strength changes observed in this study.