This paper presents an image and video indexing approach that combines face detection and face recognition methods. images of a database or frames of a video sequence are scanned for faces by a neural Network-based fa...
详细信息
ISBN:
(纸本)0780370414
This paper presents an image and video indexing approach that combines face detection and face recognition methods. images of a database or frames of a video sequence are scanned for faces by a neural Network-based face detector. The extracted faces are then grouped into clusters by a combination of a face recognition method using pseudo two-dimensional Hidden Markov Models and a k-means clustering algorithm. Each resulting main cluster consists of the face images of one person. In a subsequent step the detected faces are labeled as one of the different people in the video sequence or the image database and the occurrence of the people can be evaluated. The results of the proposed approach on a TV broadcast news sequence are presented. It is demonstrated that the system is able to discriminate between three different newscasters and an interviewed person.
This paper proposes a new method, using neural networks, of adapting phone HMMs to noisy speech. The neural networks are designed to map clean speech HMMs to noise-adapted HMMs, using noise HMMs and signal-to-noise ra...
详细信息
ISBN:
(纸本)0780370414
This paper proposes a new method, using neural networks, of adapting phone HMMs to noisy speech. The neural networks are designed to map clean speech HMMs to noise-adapted HMMs, using noise HMMs and signal-to-noise ratios (SNRs) as inputs, and are trained to minimize the mean square error between the output HMMs and the target noise-adapted HMMs. In evaluation, the proposed method was used to recognize noisy broadcast-news speech in speaker-dependent and -independent modes. The trained networks were confirmed to be effective in recognizing new speakers under new noise and various SNR conditions.
This paper presents the performance results of a recently developed minimal radial basis function neural network referred to as Minimal Resource Allocation Network (MRAN) for equalization of a highly nonlinear magneti...
详细信息
ISBN:
(纸本)0780370414
This paper presents the performance results of a recently developed minimal radial basis function neural network referred to as Minimal Resource Allocation Network (MRAN) for equalization of a highly nonlinear magnetic recording data storage channel. Using a realistic magnetic channel model, MRAN equalizer's performance has been studied in the presence of channel impairments alike partial erasure, additive white gaussian noise and jitter and width variance. Compared with the earlier neural equalizers, MRAN equalizer has better performance in terms of higher signal to Distortion Ratios (SDR).
In this paper, automatic classification of QAM signals including 64-state QAM and 256-state QAM is discussed.: Three layer neural networks whose input data is the histogram distribution of instantaneous amplitude at s...
详细信息
ISBN:
(纸本)0780370414
In this paper, automatic classification of QAM signals including 64-state QAM and 256-state QAM is discussed.: Three layer neural networks whose input data is the histogram distribution of instantaneous amplitude at symbol points is used for the classification. The evaluations of classification performance are carried out for both cases in which the synchronization of symbol timing is assured at the receiver and not assured. Good classification results are obtained by the computer simulations at SNR greater than or equal to 10dB. The influence of the number of symbol points which are used for the calculation of histogram is also discussed.
This work considers the practical situation where adaptive systems arc subject to a saturation nonlinearity at the output of the adaptive filter. Such is the case in active control of noise and vibration. A new adapti...
详细信息
ISBN:
(纸本)0780370414
This work considers the practical situation where adaptive systems arc subject to a saturation nonlinearity at the output of the adaptive filter. Such is the case in active control of noise and vibration. A new adaptive algorithm is proposed which implements the true stochastic gradient approach to the nonlinear problem. Deterministic nonlinear recursions are derived which model the mean weight and mean square error behaviors. The steady-state behavior is also studied. The practical aspects of nonlinearity estimation and hardware implementation are addressed. It is shown that the new algorithm outperforms the LMS algorithm even for considerable errors in estimating the nonlinearity parameters.
This paper discusses several issues related to blind source separation in nonlinear models. Specifically, separability results show that separation in the general case is impossible, however, for specific nonlinear mo...
详细信息
ISBN:
(纸本)0780370414
This paper discusses several issues related to blind source separation in nonlinear models. Specifically, separability results show that separation in the general case is impossible, however, for specific nonlinear models the problem does have a solution. A specific set of parametric nonlinear mixtures is considered, this set has the Lie group structure. In the parameter set, a group operation is defined and a relative gradient is defined. The latter is applied to design stochastic algorithms for which the equivariance property is shown.
We have developed a method suitable for reconstructing spatio-temporal activities of neural sources using MEG data. Our method is based on an adaptive beamformer technique. It extends a beamformer originally proposed ...
详细信息
ISBN:
(纸本)0780370414
We have developed a method suitable for reconstructing spatio-temporal activities of neural sources using MEG data. Our method is based on an adaptive beamformer technique. It extends a beamformer originally proposed by Borgiotti and Kaplan to a vector beamformer formulation in which three sets of weight vectors are used to detect the source activity in three orthogonal directions. The weight vectors of this vector-extension of the Borgiotti-Kaplan beamformer are then projected onto the signal subspace of the measurement covariance matrix to obtain a final form of the proposed beamformer's weight vectors. Our numerical experiments demonstrated the effectiveness of the proposed beamformer.
This paper deals with a new approach to detect the structure (i.e. determination of the number of hidden units) of a feedforward neural network (FNN). This approach is based on the principle that any FNN could be repr...
详细信息
ISBN:
(纸本)0780370414
This paper deals with a new approach to detect the structure (i.e. determination of the number of hidden units) of a feedforward neural network (FNN). This approach is based on the principle that any FNN could be represented by a Volterra series such as a nonlinear input-output model. The new proposed algorithm is based on the following three steps: first, we develop the nonlinear activation function of the hidden layer's neurons in a Taylor expansion, secondly we express the neural network output as a NARX (nonlinear auto regressive with exogenous input) model and finally, by appropriately using the nonlinear order selection algorithm proposed by Kortmann-Unbehauen, we select the most relevant signals on the NARX model obtained. Starting from the output layer, this pruning procedure is performed on each node in each layer. Using this new algorithm with the standard backpropagation (SBP) and over various initial conditions, we perform Monte Carlo experiments leading to a drastic reduction in the nonsignificant network hidden layer neurons.
Chaotic signals generated by iterating nonlinear difference equations may be useful models for many natural phenomena. In this paper we propose a family of chaotic models for signalprocessing applications. The chaoti...
详细信息
ISBN:
(纸本)0780370414
Chaotic signals generated by iterating nonlinear difference equations may be useful models for many natural phenomena. In this paper we propose a family of chaotic models for signalprocessing applications. The chaotic signals generated by this family of first order difference equations have autocorrelations identical to stochastic first-order autoregressive (AR) processes. After considering the huge computational cost and the inconsistency of the optimal model estimator in the maximum-likelihood (NIL) sense we propose low cost, suboptimal estimation approaches. Computer simulations show the good performance of the proposed modeling approach.
The objective of this paper is the accurate prediction of segmental duration in a Spanish text-to-speech system. There are many parameters that affect duration, but not all of them are always relevant. We present a co...
详细信息
ISBN:
(纸本)0780370414
The objective of this paper is the accurate prediction of segmental duration in a Spanish text-to-speech system. There are many parameters that affect duration, but not all of them are always relevant. We present a complete environment in which to decide which parameters are more relevant and the best way to code them. This work is the continuation of [1], where all efforts were dedicated to an unrestricted-domain database for a male voice. In this case, we are considering a female voice in a restricted-domain environment. This restricted-domain offers several advantages to the modeling: the variation in the different patterns is reduced, and so most of the decisions we have made about the parameters are now based in more significant results. So, the conclusions that we present now show clearly which parameters are best. The system is based in a neural network absolutely configurable.
暂无评论