Asymmetric cross-coupled differential pair configuration to realize neuron activation function and its derivative

In this brief, we present a new circuit technique to generate the sigmoid neuron activation function (NAF) and its derivative (DNAF). The circuit makes use of transistor asymmetry in cross-coupled differential pair to obtain the derivative. The asymmetry is introduced through external control signal, as and when required. This results in the efficient utilization of the hard-ware by realizing NAF and DNAF using the same building blocks. The operation of the circuit is presented in the subthreshold region for ultra low-power applications. The proposed circuit has been experimentally prototyped and characterized as a proof of concept on the 1.5-mum AMI technology.

Sequential split vector quantization of LSF parameters using conditional PDF

A better performing product code vector quantization (VQ) method is proposed for coding the line spectrum frequency (LSF) parameters; the method is referred to as sequential split vector quantization (SeSVQ). The split sub-vectors of the full LSF vector are quantized in sequence and thus uses conditional distribution derived from the previous quantized sub-vectors. Unlike the traditional split vector quantization (SVQ) method, SeSVQ exploits the inter sub-vector correlation and thus provides improved rate-distortion performance, but at the expense of higher memory. We investigate the quantization performance of SeSVQ over traditional SVQ and transform domain split VQ (TrSVQ) methods. Compared to SVQ, SeSVQ saves 1 bit and nearly 3 bits, for telephone-band and wide-band speech coding applications respectively.

On the performance analysis of a class of neuron circuits

A new performance metric, Peak-Error Ratio (PER) has been presented to benchmark the performance of a class of neuron circuits to realize neuron activation function (NAF) and its derivative (DNAF). Neuron circuits, biased in subthreshold region, based on the asymmetric cross-coupled differential pair configuration and conventional configuration of applying small external offset voltage at the input have been compared on the basis of PER. It is shown that the technique of using transistor asymmetry in a cross-coupled differential pair performs on-par with that of applying external offset voltage. The neuron circuits have been experimentally prototyped and characterized as a proof of concept on the 1.5 mu m AMI technology.

On the performance analysis of a class of neuron circuits

A new performance metric, Peak-Error Ratio (PER) has been presented to benchmark the performance of a class of neuron circuits to realize neuron activation function (NAF) and its derivative (DNAF). Neuron circuits, biased in subthreshold region, based on the asymmetric cross-coupled differential pair configuration and conventional configuration of applying small external offset voltage at the input have been compared on the basis of PER. It is shown that the technique of using transistor asymmetry in a cross-coupled differential pair performs on-par with that of applying external offset voltage. The neuron circuits have been experimentally prototyped and characterized as a proof of concept on the 1.5 mu m AMI technology.

Low PAPR full-diversity space-frequency codes for MIMO-OFDM systems

Use of precoding transforms such as Hadamard Transforms and Phase Alteration for Peak to Average Power Ratio (PAPR) reduction in OFDM systems are well known. In this paper we propose use of Inverse Discrete Fourier Transform (IDFT) and Hadamard transform as precoding transforms in MIMO-OFDM systems to achieve low peak to average power ratio (PAPR). We show that while our approach using IDFT does not disturb the diversity gains of the MIMO-OFDM systems (spatial, temporal and frequency diversity gains), it offers a better trade-off between PAPR reduction and ML decoding complexity compared to that of the Hadamard transform precoding. We study in detail the amount of PAPR reduction achieved for the following two recently proposed full-diversity Space-Frequency coded MIMO-OFDM systems using both the IDFT and the Hadamard transform: (i) W. Su. Z. Safar, M. Olfat, K. J. R. Liu (IEEE Trans. on Signal Processing, Nov. 2003), and (ii) W. Su, Z. Safar, K. J. R. Liu (IEEE Trans. on Information Theory, Jan. 2005).

The Line Spectral Frequency Model of a Finite-Length Sequence

The line spectral frequency (LSF) of a causal finite length sequence is a frequency at which the spectrum of the sequence annihilates or the magnitude spectrum has a spectral null. A causal finite-length sequencewith (L + 1) samples having exactly L-LSFs, is referred as an Annihilating (AH) sequence. Using some spectral properties of finite-length sequences, and some model parameters, we develop spectral decomposition structures, which are used to translate any finite-length sequence to an equivalent set of AH-sequences defined by LSFs and some complex constants. This alternate representation format of any finite-length sequence is referred as its LSF-Model. For a finite-length sequence, one can obtain multiple LSF-Models by varying the model parameters. The LSF-Model, in time domain can be used to synthesize any arbitrary causal finite-length sequence in terms of its characteristic AH-sequences. In the frequency domain, the LSF-Model can be used to obtain the spectral samples of the sequence as a linear combination of spectra of its characteristic AH-sequences. We also summarize the utility of the LSF-Model in practical discrete signal processing systems.

Performance analysis of a suprathreshold, stochastic resonance based nonlinear detector

In this paper we introduce a nonlinear detector based on the phenomenon of suprathreshold stochastic resonance (SSR). We first present a model (an array of 1-bit quantizers) that demonstrates the SSR phenomenon. We then use this as a pre-processor to the conventional matched filter. We employ the Neyman-Pearson(NP) detection strategy and compare the performances of the matched filter, the SSR-based detector and the optimal detector. Although the proposed detector is non-optimal, for non-Gaussian noises with heavy tails (leptokurtic) it shows better performance than the matched filter. In situations where the noise is known to be leptokurtic without the availability of the exact knowledge of its distribution, the proposed detector turns out to be a better choice than the matched filter.

Estimation of signals in colored non gaussian noise based on Gaussian Mixture models

Non-Gaussianity of signals/noise often results in significant performance degradation for systems, which are designed using the Gaussian assumption. So non-Gaussian signals/noise require a different modelling and processing approach. In this paper, we discuss a new Bayesian estimation technique for non-Gaussian signals corrupted by colored non Gaussian noise. The method is based on using zero mean finite Gaussian Mixture Models (GMMs) for signal and noise. The estimation is done using an adaptive non-causal nonlinear filtering technique. The method involves deriving an estimator in terms of the GMM parameters, which are in turn estimated using the EM algorithm. The proposed filter is of finite length and offers computational feasibility. The simulations show that the proposed method gives a significant improvement compared to the linear filter for a wide variety of noise conditions, including impulsive noise. We also claim that the estimation of signal using the correlation with past and future samples leads to reduced mean squared error as compared to signal estimation based on past samples only.

A cost effective pipelined divider for double precision floating point number

he growth of high-performance application in computer graphics, signal processing and scientific computing is a key driver for high performance, fixed latency; pipelined floating point dividers. Solutions available in the literature use large lookup table for double precision floating point operations.In this paper, we propose a cost effective, fixed latency pipelined divider using modified Taylor-series expansion for double precision floating point operations. We reduce chip area by using a smaller lookup table. We show that the latency of the proposed divider is 49.4 times the latency of a full-adder. The proposed divider reduces chip area by about 81% than the pipelined divider in [9] which is based on modified Taylor-series.

A speech-music discriminator using HILN model based features

We propose a simple speech music discriminator that uses features based on HILN(Harmonics, Individual Lines and Noise) model. We have been able to test the strength of the feature set on a standard database of 66 files and get an accuracy of around 97%. We also have tested on sung queries and polyphonic music and have got very good results. The current algorithm is being used to discriminate between sung queries and played (using an instrument like flute) queries for a Query by Humming(QBH) system currently under development in the lab.

Dynamic programming based optimum non-uniform samples for speech reconstruction and coding

Non-uniform sampling of a signal is formulated as an optimization problem which minimizes the reconstruction signal error. Dynamic programming (DP) has been used to solve this problem efficiently for a finite duration signal. Further, the optimum samples are quantized to realize a speech coder. The quantizer and the DP based optimum search for non-uniform samples (DP-NUS) can be combined in a closed-loop manner, which provides distinct advantage over the open-loop formulation. The DP-NUS formulation provides a useful control over the trade-off between bitrate and performance (reconstruction error). It is shown that 5-10 dB SNR improvement is possible using DP-NUS compared to extrema sampling approach. In addition, the close-loop DP-NUS gives a 4-5 dB improvement in reconstruction error.

Comparative study of filter-bank mean-energy distance for automated segmentation of speech signals

This paper describes a method of automated segmentation of speech assuming the signal is continuously time varying rather than the traditional short time stationary model. It has been shown that this representation gives comparable if not marginally better results than the other techniques for automated segmentation. A formulation of the 'Bach' (music semitonal) frequency scale filter-bank is proposed. A comparative study has been made of the performances using Mel, Bark and Bach scale filter banks considering this model. The preliminary results show up to 80 % matches within 20 ms of the manually segmented data, without any information of the content of the text and without any language dependence. 'Bach' filters are seen to marginally outperform the other filters.

Classification of magnetic resonance brain images using wavelets as input to support vector machine and neural network

In this paper. we propose a novel method using wavelets as input to neural network self-organizing maps and support vector machine for classification of magnetic resonance (MR) images of the human brain. The proposed method classifies MR brain images as either normal or abnormal. We have tested the proposed approach using a dataset of 52 MR brain images. Good classification percentage of more than 94% was achieved using the neural network self-organizing maps (SOM) and 98% front support vector machine. We observed that the classification rate is high for a Support vector machine classifier compared to self-organizing map-based approach.

Filter design using radial basis function neural network and genetic algorithm for improved operational health monitoring

The problem of denoising damage indicator signals for improved operational health monitoring of systems is addressed by applying soft computing methods to design filters. Since measured data in operational settings is contaminated with noise and outliers, pattern recognition algorithms for fault detection and isolation can give false alarms. A direct approach to improving the fault detection and isolation is to remove noise and outliers from time series of measured data or damage indicators before performing fault detection and isolation. Many popular signal-processing approaches do not work well with damage indicator signals, which can contain sudden changes due to abrupt faults and non-Gaussian outliers. Signal-processing algorithms based on radial basis function (RBF) neural network and weighted recursive median (WRM) filters are explored for denoising simulated time series. The RBF neural network filter is developed using a K-means clustering algorithm and is much less computationally expensive to develop than feedforward neural networks trained using backpropagation. The nonlinear multimodal integer-programming problem of selecting optimal integer weights of the WRM filter is solved using genetic algorithm. Numerical results are obtained for helicopter rotor structural damage indicators based on simulated frequencies. Test signals consider low order polynomial growth of damage indicators with time to simulate gradual or incipient faults and step changes in the signal to simulate abrupt faults. Noise and outliers are added to the test signals. The WRM and RBF filters result in a noise reduction of 54 - 71 and 59 - 73% for the test signals considered in this study, respectively. Their performance is much better than the moving average FIR filter, which causes significant feature distortion and has poor outlier removal capabilities and shows the potential of soft computing methods for specific signal-processing applications.

Multi-Pattern Viterbi Algorithm for joint decoding of multiple speech patterns

Joint decoding of multiple speech patterns so as to improve speech recognition performance is important, especially in the presence of noise. In this paper, we propose a Multi-Pattern Viterbi algorithm (MPVA) to jointly decode and recognize multiple speech patterns for automatic speech recognition (ASR). The MPVA is a generalization of the Viterbi Algorithm to jointly decode multiple patterns given a Hidden Markov Model (HMM). Unlike the previously proposed two stage Constrained Multi-Pattern Viterbi Algorithm (CMPVA),the MPVA is a single stage algorithm. MPVA has the advantage that it cart be extended to connected word recognition (CWR) and continuous speech recognition (CSR) problems. MPVA is shown to provide better speech recognition performance than the earlier techniques: using only two repetitions of noisy speech patterns (-5 dB SNR, 10% burst noise), the word error rate using MPVA decreased by 28.5%, when compared to using individual decoding. (C) 2010 Elsevier B.V. All rights reserved.