Athanasios Koutras

University of Peloponnese | UOP · Electrical & Computer Engineering

In this paper we present a novel solution to the convolutive and post non-linear Blind Speech Separation (NLBSS) problem based on a neural network topology. The non-linear separating functions are chosen to be a mixture of parametric sigmoid functions. The estimation of the separating filter coefficients and the parameters of the separating non-lin...

In this paper a robust regions-of-suspicion (ROS) diagnosis system on mammograms, recognizing all types of abnormalities is presented and evaluated. A new type of statistical descriptors, based on Independent Component Analysis (ICA), derive the source regions that generate the observed ROS in mammograms. The reduced set of linear transformation co...

This paper describes the theoretical background of a new data-driven approach to encephalographic single-trial (ST) data analysis. Temporal constrained source extraction using sparse decomposition identifies signal topographies that closely match the shape characteristics of a reference signal, one response for each ST. The correlations between the...

A computer-aided method for the classification of regions of suspicion on digitized mammograms is presented. The method employs features extracted by a novel technique based on independent component analysis. We concentrate our approach on finding a set of independent source regions that generate the observed regions. The coefficients of the linear...

We present a novel blind signal extraction (BSE) method for robust speech recognition in a real room environment under the coexistence of simultaneous interfering non-speech sources. The proposed method is capable of extracting the target speaker's voice based on a maximum kurtosis criterion. Extensive phoneme recognition experiments have proved th...

An important part of planning a trip involves not only search for accommodation, but for food and beverage spots as well, to enhance the visitor’s travel experience, especially when the trip involves visits to major urban centers. The agony and/or joy of tourists to meet the desired destinations can be seen nowadays in the continuous and overwhelmi...

Mammography still remains the foremost effective procedure for early diagnosis of breast cancer. To compensate with the increasing number of readings that radiologists have to undertake, Computer-aided Diagnosis (CAD) systems have become a significant assistance tool, that are used to help identify abnormal and normal regions of interest in mammogr...

In this paper we present a novel approach to the person identification problem using rhythmic brain activity of spindles from whole night EEG recordings. The proposed system consists of a feature extraction module and a K-NN based classifier. Different types of features from time, frequency and wavelet domain are used to highlight the topographic,...

Mammography is still the most effective procedure for early diagnosis of the breast cancer. Computer-aided Diagnosis (CAD) systems can be very helpful in this direction for radiologists to recognize abnormal and normal regions of interest in digital mammograms faster than traditional screening program. In this work, we propose a new method for brea...

Music Emotion Recognition (MER) is an important topic in music understanding, recommendation and retrieval that has gained great attention in the last years due to the constantly increasing number of people accessing digital musical content. In this paper we propose a new song emotion recognition system that takes into consideration the song’s genr...

In this paper we propose and evaluate linear and nonlinear prediction models based on Artificial Neural Networks (ANN) for tourism demand in accommodation industry. For efficient forecasting the Multilayer Perceptron (MLP), Support Vector regression (SVR) and Linear Regression (LR) have been used that utilize two different feature sets for training...

In this paper a robust regions-of-suspicion (ROS) diagnosis system on mammograms, recognizing all types of abnormalities is presented and evaluated. A new type of descriptors, based on Shapelet decomposition, estimate the source images that generate the observed ROS in mammograms. The Shapelet decomposition coefficients can be used to efficiently d...

Accurate tourism demand forecasting systems are very important in tourism planning, especially in high tourist countries and regions within. In this paper we investigate the problem of accurate tourism demand prediction using nonlinear regression techniques based on Artificial Neural Networks (ANN). The relative accuracy of the Multilayer Perceptro...

In this paper we propose a new technique to forecast tourism demand based on Independent Component Analysis. The proposed method uses Dynamic Embedding (DE) to transform the time series in a higher dimensional space, where Independent Component Analysis is performed to estimate the independent components (sources). Prediction is then applied using...

In this paper a robust regions-of-suspicion (ROS) diagnosis system on mammograms, recognizing all types of abnormalities is presented and evaluated. A new type of descriptors, based on Shapelet decomposition, derive the source images that generate the observed ROS in mammograms. The Shapelet decomposition coefficients can be used efficiently to det...

Independent Component Analysis (ICA) is a new signal processing technique that extracts a set of underlying sources or components from a set of measurements or signals without incorporating any a-priori information. In brain signal analysis area, ICA has already been used as an artifact rejection technique, or to analyze Event Related Potentials (E...

In this paper we present a new approach to the prediction of Cardiotocogram (CTG) missing values. Our approach uses Dynamic Embedding (DE) to transform the CTGs in a higher dimensional space, combined with Independent Component Analysis (ICA) to estimate the Independent Components (ICs) that best describe them. Prediction is then performed in the I...

Concurrent studies of central and autonomic activity are considered useful in elucidating the relationship between the two systems and indicating the centripetal feedback of peripheral changes. The SSR (sympathetic skin response) is one index of autonomic arousal. In our study we examine the possible relationships between central and autonomic resp...

A computer aided neural network classification of regions of suspicion (ROS) on digitized mammograms is presented in this paper which employs features extracted by a new technique based on independent component analysis. Our approach is concentrated in finding a set of independent source regions that generate the observed mammograms. The coefficien...

In this paper, we examine the robustness of Blind Signal Separation (BSS) in the time as well as the frequency domain, for separating competing speakers in real reverberant environments. The separation network's learning rule is based on the Maximum Likelihood Estimation criterion and was tested in real room situations in a noise-free reverberant e...

In this paper we present a novel solution to the convolutive and post non-linear Blind Speech Separation (NLBSS) problem. The non-linear separating functions are chosen to be a mixture of parametric sigmoid functions and higher order odd polynomial functions. The estimation of the separating filter coefficients and the parameters of the separating...

The Cardiotocogram (CTG), the continuous recording of fetal heart rate and maternal contractions during labor, has been introduced in the clinical routine eliminating fetus mortality and warning obstetricians about the health status of fetus and the occurrence of problems during antenatal as well as during labour. The instantaneous FHR (beats/min)...

In this paper we propose a new Blind Separation method for speech signals in the wavelet domain. Our approach consists in transforming the basic speech separation problem into the wavelet domain by decomposing the signal space in two orthogonal subspaces, where successful separation is achieved by combining information from both subspaces. Extensiv...

In this paper we present a novel method for recognizing all kinds of abnormalities in digital mammograms using Independent Component Analysis mixture models and two sets of statistical features based on texture analysis.Our approach is concentrated on finding the ICA mixture model parameters that describe in an exclusive and effective way the abnor...

In this paper we present a novel neural network topology capable of separating simultaneous signals transferred through a memoryless non-linear path. The employed neural network is a two-layer perceptron that uses parametric non-linearities in the hidden neurons. The non-linearities are formed using a mixture of sigmoidal non-linear functions and p...

In this paper we present a new feature extraction technique for digital mammograms. Our approach uses Independent Component Analysis to find the source regions that generate the observed regions of suspicion in mammograms. The linear transformation coefficients, which result from the source regions, are used as features that describe the observed...

Dimensionality reduction of the feature vector is one of the most important preprocessing techniques in visual recognition which reduces significantly the computational load and improves the performance of the learning algorithms. In this paper, we investigate the efficiency of two linear feature reduction methods, based on Principal Component Anal...

: - In this paper, we examine the robustness of a Blind Signal Separation (BSS) technique in the time domain, based on a recurrent neural network, for separating multiple competing speakers in real reverberant environments. The separation network's learning rule is based on the Maximum Likelihood Estimation criterion and was tested in real room sit...

In this paper we present a new on-line Blind Signal Separation method capable to separate convolutive speech signals of moving speakers in highly reverberant rooms. The separation network used is a recurrent network which performs separation of convolutive speech mixtures in the time domain, without any prior knowledge of the propagation media, bas...

In this paper we present a neural network approach to the Blind Signal Separation problem of simultaneous speech signals in highly reverberant noisy rooms. The separation networks that are used, are recurrent and feedforward neural networks, along with a proposed hybrid network. These networks perform separation of convolutive speech mixtures in th...

In this paper it is shown that a Blind Signal Separation (BSS) method in the frequency domain (FDBSS) improves significantly the speaker Signal to Interference Ratio (SIR) and the phoneme recognition score of a continuous speech, speaker-independent acoustic decoder in a multi-simultaneous-speaker office environment. Specifically, the efficiency of...

In this paper it is shown experimentally that a new blind signal separation method in the frequency domain improves significantly the speaker signal to interference ratio (SIR) and the phoneme recognition score of a continuous speech, speaker-independent acoustic decoder in a two-simultaneousspeaker environment. The implemented two-sensor separatio...

In this paper it is shown that a blind signal separation (BSS) method in the frequency domain (FDBSS) improves significantly the speaker signal to interference ratio (SIR) and the phoneme recognition score of a continuous speech, speaker-independent acoustic decoder in a multi-simultaneous-speaker office environment. Specifically, the efficiency of...