Valeri Mladenov

Neural Networks for Online Classification of Hand and Finger Movements Using Surface EMG signals

Myoelectric signals (MES) are the electrical manifestation of muscular contractions and they can be used to create myoelectric prosthesis which is able to function with the amputees muscle movements. This signal recorded at the surface of the skin of the forearm has been exploited to provide recognition of the movement of the hand and finger movements of healthy subject. The objective of the paper is to describe the identification procedure, based on EMG patterns of forearm activity using various neural networks methods and to make a comparison between different intelligent computational methods of identification, which are used in this work. Then an online algorithm for movement identification and classification that utilises the trained neural networks is presented

Design of two-dimensional recursive filters by using neural networks

A new design method for two-dimensional (2-D) recursive digital filters is investigated. The design of the 2-D filter is reduced to a constrained minimization problem the solution of which is achieved by the convergence of an appropriate neural network. The method is tested on a numerical example and compared with previously published methods when applied to the same example. Advantages of the proposed method over the existing ones are discussed as well.

Artificial Neural Networks and Machine Learning – ICANN 2013

When the inputs of a regression problem are corrupted with noise, integrating out the noise process leads to biased estimates. We present a method that corrects the bias caused by the integration. The correction is proportional to the Hessian of the learned model and to the variance of the input noise. The method works for arbitrary regression models, the only requirement is two times differentiability of the respective model. The conducted experiments suggest that significant improvement can be gained using the proposed method. Nevertheless, experiments on high dimensional data highlight the limitations of the algorithm.

An interval method for finding all operating points of non-linear resistive circuits

A new method for finding the set of all d.c. operating points of non-linear electronic circuits is suggested. it is applicable in the case where the circuit equations are written in the hybrid-representation form. In the general case, the non-linear elements involved may be described by non-monotone continuous characteristics. The method suggested is based on interval analysis techniques. Unlike other non-interval methods, this approach guarantees that all operating points will be found within prescribed accuracy in a finite number of steps. The computational efficiency of the present method is illustrated by a numerical example.

Short-term load forecasting using a chaotic time series

A new approach to short-term load forecasting (STLF) in power systems is described in this paper. The method uses a chaotic time series and artificial neural network. The paper describes chaos time series analysis of daily power system peak loads. Nonlinear mapping of deterministic chaos is identified by multilayer perceptron (MLP). Using embedding dimension and delay time, an attractor in pseudo phase plane and an ANN model trained by this attractor are constructed. The proposed approach is demonstrated by an example.

On the stability of high order Sigma-Delta modulators

In this paper we present an approach for stability analysis of high order Sigma-Delta modulators. The approach is based on a parallel decomposition of the modulator. In this representation, the general N-th order modulator is transformed into decomposition of low order modulators, which interact only through the quantizer function. In the simplest case of the loop filter transfer function with real distinct poles, the low order modulators are N first order ones. The decomposition considered helps to extract the stability conditions of the N-th order modulator. They are determined by the stability conditions of each of the low order modulators but shifted with respect to the origin of the quantizer function, because of the influence of all other low order modulators. The approach is generalized for the case of repeated poles of the loop filter transfer function.

Approximation Formula for Easy Calculation of Signal-to-Noise Ratio of Sigma-Delta Modulators

The signal-to-noise ratio (SNR) is one of the most significant measures of performance of the sigma-delta modulators. An approximate formula for calculation of signal-to-noise ratio of an arbitrary sigma-delta modulator (SDM) has been proposed. Our approach for signal-to-noise ratio computation does not require modulator modeling and simulation. The proposed formula is compared with SNR calculations based on output bitstream obtained by simulations, and the reasons for small discrepancies are explained. The proposed approach is suitable for fast and precise signal-to-noise ratio computation. It is very useful in the modulator design stage, where multiple performance estimates are required.

Speech recognition using neural networks

This paper presents investigation on speech recognition classification performance when using different standard neural networks structures as a classifier. Those cases include usage of a Feed-forward Neural Network (NN) with back propagation algorithm and a Radial Basis Functions (RBF) Neural Network.

On the Stability Analysis of High Order Sigma-Delta Modulators

In this paper we present an approach for stability analysis of high order Sigma-Delta modulators. The approach is based on a parallel decomposition of the modulator. In this representation, the general N-th order modulator is transformed into decomposition of low order modulators, which interact only through the quantizer function. In the simplest case of the loop filter transfer function with real distinct poles, the low order modulators are N first order ones. The decomposition considered helps to extract the sufficient conditions for stability of the N-th order modulator. They are determined by the stability conditions of each of the low order modulators but shifted with respect to the origin of the quantizer function, because of the influence of all other low order modulators. The approach is generalized for the case of repeated poles of the loop filter transfer function.

EVALUATION OF DISTANCE BASED AMPLITUDE PANNING FOR SPATIAL AUDIO

Distance-Based Amplitude Panning (DBAP) has recently been proposed as a new technique for panning sound sources in two and three dimensional spaces spaces. In this paper, DBAP is compared with two established alternatives, Ambisonics and Vector-Based Amplitude Panning, both objectively in terms of speaker gains and their variation with source position, and subjectively using listening tests to estimate apparent source position.