A5056129670- Advanced Data Compression Techniques
- Advanced Adaptive Filtering Techniques
- Image and Signal Denoising Methods
- Speech and Audio Processing
- Voice and Speech Disorders
- Digital Filter Design and Implementation
- Music and Audio Processing
- Sensor Technology and Measurement Systems
- Blind Source Separation Techniques
- Respiratory and Cough-Related Research
- Mineral Processing and Grinding
- Advanced Control Systems Optimization
- Radiomics and Machine Learning in Medical Imaging
- Phonocardiography and Auscultation Techniques
- Advanced Control Systems Design
- Context-Aware Activity Recognition Systems
- Cell Image Analysis Techniques
- Cardiac, Anesthesia and Surgical Outcomes
- Fractional Differential Equations Solutions
- Industrial Automation and Control Systems
- Speech Recognition and Synthesis
- Emergency and Acute Care Studies
- Wireless Sensor Networks for Data Analysis
- Inflammatory Myopathies and Dermatomyositis
- COVID-19 diagnosis using AI
Washington University in St. Louis
2015-2025
Luxembourg Institute of Health
2022-2024
University of Luxembourg
2020-2023
Barnes-Jewish Hospital
2005-2020
University of Belgrade
2009-2019
University of Nis
2006-2019
Mining and Metallurgy Institute Bor
2010-2019
Pidstryhach Institute for Applied Problems of Mechanics and Mathematics
2019
University of Castilla-La Mancha
2019
Northwestern University
2019
COVID-19 heavily affects breathing and voice causes symptoms that make patients' voices distinctive, creating recognizable audio signatures. Initial studies have already suggested the potential of using as a screening solution. In this article we present dataset voice, cough recordings collected from individuals infected by SARS-CoV-2 virus, well non-infected subjects via large scale crowdsourced campaign. We describe preliminary results for detection patterns standard acoustic features...
In recent years, human activity recognition (HAR) has gained importance in several domains such as surveillance, recognizing indoor and outdoor activities, providing active assisted living environments smart homes healthcare services. all these scenarios, audio-, video- image-based processing algorithms have been applied well systems using wearable sensors. This scoping review focuses on audio- video-based for applications. We provide a comprehensive overview of technologies discuss their...
We provide in this paper a comprehensive comparison of various transfer learning strategies and deep architectures for computer-aided classification adult-type diffuse gliomas. evaluate the generalizability out-of-domain ImageNet representations target domain histopathological images, study impact in-domain adaptation using self-supervised multi-task approaches pretraining models medium-to-large scale datasets images. A semi-supervised approach is furthermore proposed, where fine-tuned are...
New language technologies are coming, thanks to the huge and competing private investment fuelling rapid progress; we can either understand foresee their effects, or be taken by surprise spend our time trying catch up. This report scketches out some transformative new that likely fundamentally change use of language. Some these may feel unrealistically futuristic far-fetched, but a central purpose this - wider LITHME network is illustrate mostly just logical development maturation currently...
Hemophagocytic lymphohistiocytosis (HLH) and rhabdomyolysis are rare complications of typhoid fever from Salmonella enterica serovar Typhi. Herein, we describe the clinical features in a 21-year-old female India who presented to intensive care unit with fever, severe pancytopenia, rhabdomyolysis.
Objective To develop a vocal biomarker for fatigue monitoring in people with COVID-19. Design Prospective cohort study. Setting Predi-COVID data between May 2020 and 2021. Participants A total of 1772 voice recordings were used to train an AI-based algorithm predict fatigue, stratified by gender smartphone’s operating system (Android/iOS). The collected from 296 participants tracked 2 weeks following SARS-CoV-2 infection. Primary secondary outcome measures Four machine learning algorithms...
Seven state-of-the-art machine learning techniques for estimation of construction costs reinforced-concrete and prestressed concrete bridges are investigated in this paper, including artificial neural networks (ANN) ensembles ANNs, regression tree (random forests, boosted bagged trees), support vector (SVR) method, Gaussian process (GPR). A database design characteristics 181 prestressed-concrete is created model training evaluation.
Introduction: The complex health, social, and economic consequences of tobacco smoking underscore the importance incorporating reliable scalable data collection on status habits into research across various disciplines. Given that impacts voice production, we aimed to develop a gender language-specific vocal biomarker status. Methods: Leveraging from Colive Voice study, used statistical analysis methods quantify effects characteristics. Various feature extraction combined with machine...
People with COVID-19 can experience impairing symptoms that require enhanced surveillance. Our objective was to train an artificial intelligence-based model predict the presence of and derive a digital vocal biomarker for easily quantitatively monitoring symptom resolution. We used data from 272 participants in prospective Predi-COVID cohort study recruited between May 2020 2021. A total 6473 voice features were derived recordings reading standardized pre-specified text. Models trained...
Linear prediction is extensively used in modeling, compression, coding, and generation of speech signal. Various formulations linear are available, both time frequency domain, which start from different assumptions but result the same solution. In this letter, we propose a novel, generalized formulation optimal low-order using fractional (non-integer) derivatives. The proposed derivative allows for definition predictor with versatile behavior based on order derivative. We derive closed-form...
Previous studies of nonlinear prediction speech have been mostly focused on short-term prediction. This paper presents long-term based second-order Volterra filters. It will be shown that the presented predictor can outperform conventional linear techniques in terms gain and "whiter" residuals.
The one-parameter fractional linear prediction (FLP) is presented and the closed-form expressions for evaluation of FLP coefficients are derived. Contrary to classical first-order (LP) that uses one previous sample predictor coefficient, model derived using memory two, three or four samples, while not increasing number coefficients. LP only a special case proposed when order derivative tends zero. Based on numerical experiments test signals (sine waves), real-data (speech electrocardiogram),...
A compression method based on non-uniform binary scalar quantization, designed for the memoryless Laplacian source with zero-mean and unit variance, is analyzed in this paper. Two quantizer design approaches are presented that investigate effect of clipping aim reducing quantization noise, where minimal mean-squared error distortion used to determine optimal factor. detailed comparison both models provided, performance evaluation a wide dynamic range input data variances also performed. The...
Flotation deinking is one of the most widely used techniques for separation ink particles from cellulose fibers during process paper recycling. It a complex influenced by variety factors, and difficult to represent usually results in models that are inconvenient implement and/or interpret. In this paper, comprehensive study several machine learning methods prediction flotation performance carried out, including support vector regression, regression tree ensembles (random forests boosting)...
In this paper, we investigate unsupervised acoustic model training approaches for dysarthric-speech recognition.These models are first, frame-based Gaussian posteriorgrams, obtained from Vector Quantization (VQ), second, so-called Acoustic Unit Descriptors (AUDs), which hidden Markov of phone-like units, that trained in an fashion, and, third, posteriorgrams computed on the AUDs.Experiments were carried out a database collected home automation task and containing nine speakers, seven...