{ A new vector quantization method is proposed which incrementally generates a suitable codebook. During the generation process new vectors are inserted in areas of the input vector space where the quantization error is especially high. A one-dimensional topological neighborhood makes it possible to interpolate new vectors from existing ones. Vectors not contributing to error minimization are r...

This paper considers coordinated user scheduling in a multi-user two-hop multiinput multi-output relay system with limited feedback. The proposed scheme utilizes a quantized transmit correlation and channel quality information, enabling to achieve both interference mitigation and multi-user diversity (MUD) gain. To this end, we first investigate the effect of quantization error on the statistic...

A technique for efficient coding of homogeneous textures is presented here. The technique is based on the use of Stochastic Vector Quantization and provides very high compression with graceful degradation. To encode the image, a linear prediction filter is computed. Then, the prediction error is encoded using a Stochastic Vector Quantization approach. To decode the image, the prediction error i...

An initialization mechanism is presented for Kohonen neural network implemented in CMOS technology. Proper selection of initial values of neurons’ weights has a large influence on speed of the learning algorithm and finally on the quantization error of the network, which for different initial parameters can vary even by several orders of magnitude. Experiments with the software model of designe...

The second order Sigma-Delta (Σ∆) scheme with linear quantization rule is analyzed for quantizing finite unit-norm tight frame expansions for Rd. Approximation error estimates are derived, and it is shown that for certain choices of frames the quantization error is of order 1/N2, where N is the frame size. However, in contrast to the setting of bandlimited functions there are many situations wh...

This communication studies the quantization effects on the steady-state performance of a fixed-point implementation of the Least Mean Squares (LMS) adaptive algorithm. Based on experimental observations, we introduce a new intermediate mode of operation and develop a simplified theoretical approach to explain the behaviour caused by quantization effects in this mode. We also review the stall mo...

Discriminative training of the feature space using the minimum phone error objective function has been shown to yield remarkable accuracy improvements. These gains, however, come at a high cost of memory. In this paper we present techniques that maintain fMPE performance while reducing the required memory by approximately 94%. This is achieved by designing a quantization methodology which minim...

State-of-the-art image compression and reconstruction techniques utilize wavelets. Recently published research demonstrated that a genetic algorithm (GA) is capable of evolving non-wavelet transforms that consistently outperform wavelets when applied to a broad class of images under conditions subject to quantization error. This paper describes new results that build upon previous research by d...

The paper presents a non-linear quantization method for detail components in the JPEG2000 standard. The quantization step sizes are determined by actual statistics of the wavelet coefficients. Mean and standard deviation are the two statistical parameters used to obtain the step sizes. Moreover, weighted mean of the coefficients lying within the step size is chosen as quantized value contrary t...

Dithering represents a well-established, attractive technique for minimizing the audibility of the error produced during requantization of digital audio signals. However, it is also known that, as dither is applied as controlled, additive random noise prior requantisation, it also increases the overall quantization noise level and audibility. In this work, a novel dither generator is presented,...