A5071607319- Embedded Systems Design Techniques
- Music and Audio Processing
- Peer-to-Peer Network Technologies
- Parallel Computing and Optimization Techniques
- ECG Monitoring and Analysis
- Hungarian Social, Economic and Educational Studies
- Data Mining Algorithms and Applications
- Hand Gesture Recognition Systems
- Speech and Audio Processing
- Mobile Ad Hoc Networks
- Phonocardiography and Auscultation Techniques
- Network Traffic and Congestion Control
- Blind Source Separation Techniques
- Software-Defined Networks and 5G
- Caching and Content Delivery
- Time Series Analysis and Forecasting
- Handwritten Text Recognition Techniques
- Image Processing and 3D Reconstruction
- Plasma Diagnostics and Applications
- Low-power high-performance VLSI design
- Data Management and Algorithms
- Video Analysis and Summarization
- Mobile Agent-Based Network Management
- Muscle activation and electromyography studies
- Rough Sets and Fuzzy Logic
University of Wisconsin–Madison
2023
Pázmány Péter Catholic University
2006-2022
Budapest University of Technology and Economics
2006-2021
Hungarian National Bank
2020
Obuda University
2020
Nokia (Finland)
2017
Nokia (United States)
2016
Catholic University of America
2007
Budapest Institute
1999-2005
Eötvös Loránd University
1993-2005
This paper presents a complex heuristic method for the evaluation of fetal heart sounds, applying simultaneously several algorithms, where autocorrelation technique has been completed with wavelet transform and matching pursuit methods. In this way, more reliable rate variability can be achieved further parameters cardiac operation assessed in addition to conventional cardiotocographic examination. also comprises those parameters, which investigated only long-term or continuous monitoring,...
One of the most important problems in data mining is frequent itemset mining. It requires very large computation and I/O traffic capacity. For that reason several parallel distributed algorithms were developed. Recently mapreduce processing paradigm unavoidable porting current to focus. In this paper a substantial their implementations are introduced investi-gated. An algorithm improvement also proposed analyzed.
Abstract Targeted spatial transcriptomics methods capture the topology of cell types and states in tissues at single cell- subcellular resolution by measuring expression a predefined set genes. The selection an optimal probed genes is crucial for capturing interpreting signals present tissue. However, current selections often rely on marker genes, precluding them from detecting continuous or novel states. We Spapros, end-to-end probe pipeline that optimizes both specificity type...
Abstract In this paper, we present a fast and efficient technique for detecting identifying the slippage twisting motion of touching objects. This kind action cannot be detected with tactile sensors sensing only normal (perpendicular) component forces acting between surfaces. Our approach utilizes an integrated sensing–processing–actuating system comprising: (1) A 2 × taxel (tactile pixel) array mounted on two‐fingered robot hand, (2) 64 CNN‐UM (Cellular Neural Network‐Universal machine),...
The main advantage of the primary storage based on linear transformer scheme is ground potential capacitor bodies during shot, allowing exclusion total output voltage insulation highest stages, and to trigger all stages simultaneously by using an external pulse. problem was build a fast driver (LTD), providing both high efficiency energy transfer into secondary turn current rise time below /spl sim/1 CLs, because relatively inductance caps switches. key elements LTD described are HAEFELY...
Surface Electromyography (sEMG/EMG) is to record muscles' electrical activity from a restricted area of the skin by using electrodes. The sEMG-based gesture recognition extremely sensitive inter-session and inter-subject variances. We propose model deep-learning-based domain adaptation method approximate shift for accuracy enhancement. Analysis performed on sparse HighDensity (HD) sEMG public datasets validate that our approach outperforms state-of-the-art methods.
Presents an improved pass-gate adiabatic charge-recovery logic (IPGL) which is efficient in power consumption and requires only moderate reversibility overhead. The energy efficiency achieved by completely eliminating non-adiabatic loss during the charge phase of clock, overhead kept low allowing partial at nodes with small capacitance. We also show how to implement flip-flops propose binary counter. dissipation comparison shows, that IPGL outperforms previous families, it dissipates less...
We propose a novel dual-rail energy efficient adiabatic charge recovery logic. completely eliminate nonadiabatic loss during the phase, but allow partial reversibility only at nodes with small capacitance, in order to keep overhead low. apply transistor size optimization for given frequency alleviate problem of increasing higher operating frequencies. In case high path is introduced provide complete recovery. The simulation on test circuits threshold voltage 0.8 V showed 34% power reduction...
The switches that are designed for application in high pulsed power drivers should meet some hard requirements, the main of them being small inductance and big charge flowing through switch. In order to reduce increase it is preferable divide switched current into multiple parallel channels. Such channels could be obtained railgap three-electrode switches, or surface due electrical field disturbance which occurs when a trigger pulse rising as -10/sup 13/ V/s applied Multi channel discharge...
In this paper, we investigate dimension reduction possibilities of multidimensional time series data and introduce a graph based clustering approach using the cross-correlation between series. The proposed solution consists two main steps: introducing novel similarity measure for measuring cross-correlations graph-based technique. These parts are both compared to existing techniques, including noise tolerance our performs better in noisy environment. is applied performance metrics specific...
A novel compact method for fetal home monitoring optimized long data acquisition time has already been presented. The involves preprocessing a disturbed acoustic signal received on the maternal abdomen. present work summarizes phenomena encountered during analysis of incoming PCG signal, and describes further improvements made to system in use including application MP determining heartbeat lengths identifying murmurs between heart sounds.
This article proposes a novel mathematical model of logical relationship among glyphs belonging to the same grapheme. Its research belongs computational paleography that is field in applied computer science. The proposed grapheme presented four layers from bottom up namely as Topology, Visual Identity, Phonetic, and Semantic Layer. In Topology Layer, unique glyph defined by set topological properties. When trying describe relation various glyphs, their properties must be examined higher...
This paper presents a machine learning approach to explore the phenetic relations of historical scripts and their glyphs. Its first step is identification observable topological transformations in development glyphs, with use these transformations, method collects possible cognate glyphs by minimizing necessary between In investigations, phonetic properties graphemes were consistently considered. The second our selecting similarity groups differences properties. third multidimensional...
Identification in real-time streaming data using plain template-matching algorithms is difficult due to the undefined frame position on-line compared frame-based features' of templates. We have investigated use short-time Fourier-spectrum, Walsh-spectrum and signal energy ratio succeeding features from frame-shift point view. The last two aim improve calculation speed larger sets records. For further simplifying operations comparison stage, a quantization step was applied spectrum values...
This paper presents a heuristics-based, successive approximation method for heartbeat detection in fetal phonocardiographic signals, which takes advantage on some priori biological knowledge. The algorithm reaches above 90% terms of accuracy heart rate identification. It is computationally effective and easily parallelizable. A signal qualification also presented.
Due to the rapid technology advancement in integrated circuit era, need for high computation performance together with increasing complexity and manufacturing costs has raised demand high-performance configurable designs; therefore, Application-Specific Instruction Set Processors (ASIPs) are widely used SoC design.The automated generation of software tools ASIPs is a commonly technique, but hardware model less frequently applied terms final RTL implementations.Contrary this,...
We present an integer linear programming (ILP) formulation and a heuristic scheduling approach for high-level synthesis to synthesize two-level pipeline datapaths using four-phase adiabatic logic. Adiabatic CMOS logic that relies on charge recovery is attractive achieve low energy dissipation. It complements voltage-scaling approaches, while its inherent structure makes it most suitable signal processing applications. However, the differences between static logic, such as phase clock...
This paper presents the first results achieved on a LTD module built at CEG (France-DGA) in collaboration with HCEI (Russia-Tomsk). The main purpose of this work is to evaluate concept term efficiency, cost, reliability, etc. for applications multimegajoule inductive storage generators. based an voltage adder level capacitor. size components (capacitor, switch, magnetic core, ...) allows obtain very compact system running air. stage includes 2 capacitors HEAFELY (75 kV-5.651 /spl mu/F),...