Vinal Patel

Design of Adaptive Exponential Functional Link Network-Based Nonlinear Filters

A novel nonlinear filter, which incorporates the concept of exponential sinusoidal models into nonlinear filters based on functional link networks (FLNs) has been developed in this paper. The proposed filter is designed to provide improved convergence characteristics over traditional FLN filters. The conventional trigonometric FLN may be considered as a special case of the proposed adaptive exponential FLN (AEFLN). An adaptive exponential least mean square (AELMS) algorithm has been derived and the same has been successfully applied for identification of a couple of nonlinear plants. The AEFLN-based nonlinear active noise control (ANC) system has also been designed and an adaptive exponential filtered-s least mean square (AEFsLMS) algorithm has been developed to update the weights as well as the exponential factor. Simulation study has revealed the improved noise mitigation offered by the AEFLN-based nonlinear ANC system.

Compensating acoustic feedback in feed-forward active noise control systems using spline adaptive filters

Implementation of a feed-forward active noise control (ANC) system in a short duct may cause acoustic feedback between the active loudspeaker and the reference microphone. The conventional filtered-x least mean square (FxLMS) algorithm based ANC systems are not designed to handle this situation. Similarly, an FxLMS algorithm based ANC system fails to effectively mitigate noise when non-linearities are present in the system. In an endeavor to overcome these two limitations of traditional ANC systems, in this paper, we propose an adaptive infinite impulse response (IIR) spline filter for non-linear ANC. The adaptive nature of the spline activation function in the proposed scheme enables the filter to effectively switch from a linear to non-linear nature and vice versa, depending upon the scenario in which the ANC system is implemented. HighlightsANC system based on a feedback spline adaptive filter.Can compensate for acoustic feedback and nonlinearities.The nonlinearity of the filter is adaptive.Improved noise cancellation over other schemes.

Partial update even mirror fourier non-linear filters for active noise control

A non-linear active noise control (ANC) scheme, which is based on an even mirror Fourier non-linear filter has been developed in this paper. A new weight update mechanism for the proposed scheme has been suggested and the range of the learning rate which ensures stability has been derived. The noise mitigation achieved using the new scheme has been compared with that obtained using a functional link artificial neural network (FLANN) based ANC system as well as using a generalized FLANN (GFLANN) based ANC mechanism. The computational complexity of the proposed algorithm has been further reduced by using the concept of partial update signal processing. A simulation study has been carried out to evaluate the effectiveness of the new method. Improved noise reduction at reduced computational load has been provided by the new partial update ANC scheme proposed.

Modified Phase-Scheduled-Command FxLMS Algorithm for Active Sound Profiling

Active sound profiling, or active noise equalization strategies have been proposed to achieve spectral shaping of a primary disturbance signal. The control algorithms proposed to achieve such spectral shaping have either suffered from poor robustness to plant modeling uncertainties or required high levels of control effort. To improve the robustness of active sound profiling to uncertainties in the plant model, while avoiding increased control effort, a modified phase-scheduled-command filtered-x least-mean-square algorithm is proposed in this paper. The new algorithm provides improved stability, while requiring the minimum control effort. This improvement is achieved by replacing the plant model with an intelligent adaptive-hysteresis switching mechanism to allow the necessary estimation of the disturbance signal phase. The improved performance and robustness of the proposed algorithm is demonstrated through a series of simulations using measured acoustic responses.

An Improved Proportionate Delayless Multiband-Structured Subband Adaptive Feedback Canceller for Digital Hearing Aids

Acoustic feedback cancellation is one of the challenging tasks in the design of a behind the ear digital hearing aid. This feedback cancellation is usually achieved by using an adaptive filter. The finite correlation between the desired microphone input signal and the input signal to the loudspeaker results in a biased estimation of the adaptive filter, which may produce disturbances in the hearing aid. Prediction error method (PEM) has been used in literature to reduce the bias effects. The convergence of a PEM-based feedback canceller can be improved by implementing the adaptive filter in the subband domain. However, a direct subband implementation results in aliasing issues, band-edge problems, and introduces a delay due to analysis and synthesis filters. In order to reduce the aliasing and delay issues, a delayless subband implementation of a PEM-based feedback canceller is designed in this paper. A delayless multiband-structured subband implementation of the feedback canceller is also attempted to further reduce the aliasing and band-edge effects. This implementation aims at having all the subbands collectively updating the fullband adaptive filter, without the need for a subband to fullband weight conversion and offers improved feedback cancellation at reduced computational load in comparison with a delayless subband implementation of a PEM-based feedback canceller. In addition, an attempt has been made to further improve the convergence behavior by using an improved proportionate learning scheme. The improved convergence offered by the proposed scheme is evident from the simulation study. The improvement has been further quantified using a perceptual evaluation of speech quality and the proposed approach has been shown to provide enhanced speech quality.

Design of dynamic linear-in-the-parameters nonlinear filters for active noise control

Traditional active noise control (ANC) systems, which uses a fixed tap length adaptive filter as the controller may lead to non optimal noise mitigation. In addition, the conventional filtered-x least mean square algorithm based ANC schemes fail to effectively perform noise cancellation in the presence of nonlinearities in the ANC environment. In order to overcome these limitations of traditional ANC techniques, in this paper, we propose a class of dynamic nonlinear ANC systems, which adapts itself to the noise cancellation scenario. The dynamic behaviour has been achieved by developing a variable tap length and variable learning rate adaptive algorithms for functional link artificial neural network (FLANN) and generalized FLANN (GFLANN) based ANC systems. The proposed ANC schemes have been shown through a simulation study to provide an optimal convergence behaviour. This improvement has been achieved by providing a balance between the number of filter coefficients and the mean square error.

KELM-CPPpred: Kernel Extreme Learning Machine Based Prediction Model for Cell-Penetrating Peptides

Cell-penetrating peptides (CPPs) facilitate the transport of pharmacologically active molecules, such as plasmid DNA, short interfering RNA, nanoparticles, and small peptides. The accurate identification of new and unique CPPs is the initial step to gain insight into CPP activity. Experiments can provide detailed insight into the cell-penetration property of CPPs. However, the synthesis and identification of CPPs through wet-lab experiments is both resource- and time-expensive. Therefore, the development of an efficient prediction tool is essential for the identification of unique CPP prior to experiments. To this end, we developed a kernel extreme learning machine (KELM) based CPP prediction model called KELM-CPPpred. The main data set used in this study consists of 408 CPPs and an equal number of non-CPPs. The input features, used to train the proposed prediction model, include amino acid composition, dipeptide amino acid composition, pseudo amino acid composition, and the motif-based hybrid features. We further used an independent data set to validate the proposed model. In addition, we have also tested the prediction accuracy of KELM-CPPpred models with the existing artificial neural network (ANN), random forest (RF), and support vector machine (SVM) approaches on respective benchmark data sets used in the previous studies. Empirical tests showed that KELM-CPPpred outperformed existing prediction approaches based on SVM, RF, and ANN. We developed a web interface named KELM-CPPpred, which is freely available at http://sairam.people.iitgn.ac.in/KELM-CPPpred.html.

Collaborative adaptive exponential linear-in-the-parameters nonlinear filters

An adaptive exponential functional link artificial neural network (AEFLANN) based active noise control (ANC) system trained using a collaborative learning scheme has been designed in this paper. In the proposed approach, separate learning mechanism is used for updating the weights of the linear portion of the AEFLANN and its non-linear section. The outputs of the linear and non-linear sections are suitably combined and the update mechanism involves the update of weights of linear and non-linear portions, the combination parameter and the adaptive exponential factor. Simulation study shows enhanced noise cancellation in comparison with other non-linear ANC schemes compared.

Generalized spline nonlinear adaptive filters

A new spline based generalized non-linear filter.Adaptive spline function is used as basis function.Provides improved classification accuracy.Successfully applied in dynamic system identification. A new nonlinear filter, which employs an adaptive spline function as the basis function is designed in this paper. The input signal to this filter is used to generate suitable parameters to update the control points in a spline function. The update rule for updating the control points have been derived and a mean square analysis has been carried out. The output of the spline functions are suitably combined together to obtain the filter response. This filter is called the generalized spline nonlinear adaptive filter (GSNAF). The proposed GSNAF is similar to a functional link artificial neural network (FLANN), considering a functional expansion using spline basis functions. GSNAF has been shown to offer improved accuracy in benchmark classification scenarios and provide enhanced modeling accuracy in single input single output as well as in multiple input multiple output dynamic system identification cases.

Design of hybrid nonlinear spline adaptive filters for active noise control

In this paper, we focus on the problem of removing noise in the acoustic domain. To this end, we introduce a class of hybrid nonlinear spline filters, which are designed as a cascade of an adaptive spline function and a single layer adaptive nonlinear network. The adaptive nonlinear networks employed in this work are the functional link network and the even mirror Fourier nonlinear network. Suitable update rules, which not only update the adaptive weights of the nonlinear networks, but also introduce adaptability in the developed spline function are derived. The proposed nonlinear filters have been successfully applied to nonlinear system identification as well as nonlinear active noise control. The new filters have been shown to outperform other popular nonlinear filters.