A5046865970- Quantum Mechanics and Non-Hermitian Physics
- Quantum and electron transport phenomena
- Quantum chaos and dynamical systems
- Atmospheric and Environmental Gas Dynamics
- Semiconductor Quantum Structures and Devices
- Energy Load and Power Forecasting
- Advanced Numerical Methods in Computational Mathematics
- Matrix Theory and Algorithms
- Wind Energy Research and Development
- Quantum, superfluid, helium dynamics
- Meteorological Phenomena and Simulations
- Image Processing Techniques and Applications
- Solar Radiation and Photovoltaics
- Advanced Image and Video Retrieval Techniques
- Elasticity and Material Modeling
- Quantum optics and atomic interactions
- Advanced Optimization Algorithms Research
- Numerical Methods and Algorithms
- AI in cancer detection
- Advanced Research in Systems and Signal Processing
- Advanced Physical and Chemical Molecular Interactions
- Numerical methods in engineering
- Experimental Learning in Engineering
- Wind and Air Flow Studies
- Digital Imaging for Blood Diseases
University of Debrecen
2015-2024
B.I. Stepanov Institute of Physics
2017-2022
National Academy of Sciences of Belarus
2013-2021
Eötvös Loránd University
2006
Skin cancer is among the deadliest variants of if not recognized and treated in time. This work focuses on identification this disease using an ensemble state-of-the-art deep learning approaches. More specifically, we propose aggregation robust convolutional neural networks (CNNs) into one net architecture, where final classification achieved based weighted output member CNNs. Since our framework realized within a single all parameters CNNs weights applied fusion can be determined by...
Diabetic retinopathy (DR) and especially diabetic macular edema (DME) are common causes of vision loss as complications diabetes. In this work, we consider an ensemble that organizes a convolutional neural network (CNN) traditional hand-crafted features into single architecture for retinal image classification. This approach allows the joint training CNN fine-tuning weights handcrafted to provide final prediction. Our solution is dedicated automatic classification fundus images according...
In this paper, we propose a deep convolutional neural network framework to classify dermoscopy images into seven classes. With taking the advantage that these classes can be merged two (healthy/diseased) ones train part of regarding binary task only. Then, confidences classification are used tune multi-class confidence values provided by other network, since solved more accurately. For both tasks GoogLeNet Inception-v3, however, any CNN architectures could applied for purposes. The whole is...
Abstract Accurate and reliable forecasting of total cloud cover (TCC) is vital for many areas such as astronomy, energy demand production, or agriculture. Most meteorological centres issue ensemble forecasts TCC; however, these are often uncalibrated exhibit worse forecast skill than other weather variables. Hence, some form post-processing strongly required to improve predictive performance. As TCC observations usually reported on a discrete scale taking just nine different values called...
Abstract. Although, by now, ensemble-based probabilistic forecasting is the most advanced approach to weather prediction, ensemble forecasts still suffer from a lack of calibration and/or display systematic bias, thus requiring some post-processing improve their forecast skill. Here, we focus on visibility, quantity that plays crucial role in, for example, aviation and road safety or ship navigation, propose parametric model where predictive distribution mixture gamma truncated normal...
By the end of 2021, renewable energy share global electricity capacity reached 38.3% and new installations were dominated by wind solar energy, showing increases 12.7% 18.5% respectively. However, both photovoltaic sources are highly volatile, making planning difficult for grid operators; thuss, accurate forecasts corresponding weather variables essential reliable predictions. The most advanced approach in prediction is ensemble method, which opens door probabilistic forecasting. often...
<p>In 2020, 36.6 % of the total electricity demand world was covered by renewable sources, whereas in EU (UK included) this share reached 49.3 %. A substantial part green energy is produced wind farms, where accurate short range power predictions are required for successful integration into electrical grid. Accurate require forecasts corresponding weather quantity, state-of-the-art method probabilistic approach based on ensemble forecasts. However, often uncalibrated and might...
We investigate the effect of statistical post-processing on probabilistic skill discomfort index (DI) and indoor wet-bulb globe temperature (WBGTid) ensemble forecasts, both calculated from corresponding forecasts dew point temperature. Two different methodological approaches to calibration are compared. In first case, we start with joint then create calibrated samples DI WBGTid using obtained bivariate predictive distributions. This approach is compared direct heat forecasts. For this...
By the end of 2022, renewable energy share global electricity capacity reached 40.3% and new installations were dominated by solar energy, showing a increase 21.7%. Due to high volatility photovoltaic sources, their successful integration into electrical grid requires accurate short-term power forecasts. These forecasts are obtained from predictions irradiance, where most advanced method is probabilistic approach based on ensemble  However, often underdispersive subject systematic...