Radu Mirsu - Academia.edu (original) (raw)
Papers by Radu Mirsu
International Conference on Systems, Jul 14, 2011
This paper presents a problem of transportation planning that is able to achieve higher performan... more This paper presents a problem of transportation planning that is able to achieve higher performance by means of genetic optimization [2]. The paper introduces a model for simulating and evaluating the costs of a transportation system. It also describes a new way to use genetic optimization for a problem that requires adaptive chromosome length [1].
International Conference on Systems, Jul 22, 2010
This paper presents a modified version of the p-Delta learning algorithm. The algorithm can be us... more This paper presents a modified version of the p-Delta learning algorithm. The algorithm can be used for training a parallel perceptron regardless of the application. There are three significant changes from the original algorithm: an adaptive learning rate, a conscience mechanism and an adaptive margin enhancement mechanism. The changes offer an improved speed, stability and noise margin at the expense of complexity. The higher complexity can be a drawback if the algorithm is intended to be implemented in hardware.
Advances in Electrical and Computer Engineering, 2012
ABSTRACT
The hybrid Boost-L converter is suitable for practical applications that require a step-up conver... more The hybrid Boost-L converter is suitable for practical applications that require a step-up converter with wide conversion ratios. The filter stress is higher compared with the traditional Boost converter. The filter effort is reduced when using a multiphase configuration and interleaved switching. The two-phase converter is analyzed, simulated and the results are presented in this paper. Also a comparison with the single-phase Boost converter is made.
2021 IEEE 27th International Symposium for Design and Technology in Electronic Packaging (SIITME), 2021
2018 IEEE 18th International Power Electronics and Motion Control Conference (PEMC), 2018
A novel hybrid Buck-L converter with one active switch and three diodes is presented in this pape... more A novel hybrid Buck-L converter with one active switch and three diodes is presented in this paper. Compared to other topologies, the proposed converter is shown to be better suited in applications where a small difference between input and output voltage is required. It is shown that the proposed converter exhibits excellent efficiency over a wide range of the duty cycle. A DC analysis is performed and the static characteristics are derived, then device stresses are evaluated and compared to other buck topologies showing the superiority of the proposed solution. Then design equations are provided, and finally the converter is simulated and experimentally tested. The experimental results accurately confirmed the theoretical considerations and validated the feasibility of the proposed hybrid Buck topology.
This paper presents a modified version of the p-Delta learning algorithm. The algorithm can be us... more This paper presents a modified version of the p-Delta learning algorithm. The algorithm can be used for training a parallel perceptron regardless of the application. There are three significant changes from the original algorithm: an adaptive learning rate, a conscience mechanism and an adaptive margin enhancement mechanism. The changes offer an improved speed, stability and noise margin at the expense of complexity. The higher complexity can be a drawback if the algorithm is intended to be implemented in hardware.
2019 42nd International Conference on Telecommunications and Signal Processing (TSP), 2019
Deep Neural Networks and their associated learning paradigm, Deep Learning, represent one of the ... more Deep Neural Networks and their associated learning paradigm, Deep Learning, represent one of the hottest approaches used in image understanding and recognition. As their performances depends quasi-linearly on the amount of available data, the typical case studies in the literature assume the availability of huge datasets. This paper proposes to analyze several deep neural networks (trained from the scratch or pre-trained), test their efficiency in the problem of hand gesture recognition, and compare the results to a state-of-the-art classical method, the bag of features, for the case of small databases.
Energies, 2019
A new hybrid step-up converter suitable in applications where large conversion ratios are needed ... more A new hybrid step-up converter suitable in applications where large conversion ratios are needed is presented. The topology is still simple, containing only one transistor and three diodes. A detailed dc and ac analysis is performed and all design equations are provided. Compared to other topologies of the same type, the proposed converter exhibits lower or equal current and voltage stresses. A state space model is provided including the conduction losses and based on it the audio susceptibility and the control to output function are derived. As the converter is still of second order with the control to output transfer function exhibiting a right half plane zero, controller design is practically the same like in a Boost topology. It is shown how the proposed converter can be used in a photovoltaic system for performing the maximum power point tracking algorithm using the perturb and observe method. All theoretical considerations are verified through simulation and finally validated ...
2016 12th IEEE International Symposium on Electronics and Telecommunications (ISETC), 2016
The hybrid Boost-L converter is suitable for practical applications that require a step-up conver... more The hybrid Boost-L converter is suitable for practical applications that require a step-up converter with wide conversion ratios. The filter stress is higher compared with the traditional Boost converter. The filter effort is reduced when using a multiphase configuration and interleaved switching. The two-phase converter is analyzed, simulated and the results are presented in this paper. Also a comparison with the single-phase Boost converter is made.
This paper presents a problem of transportation planning that is able to achieve higher performan... more This paper presents a problem of transportation planning that is able to achieve higher performance by means of genetic optimization [2]. The paper introduces a model for simulating and evaluating the costs of a transportation system. It also describes a new way to use genetic optimization for a problem that requires adaptive chromosome length [1].
2010 9th International Symposium on Electronics and Telecommunications, 2010
Spiking Neural Networks are the last generation of neural models. Because the model is recent, ve... more Spiking Neural Networks are the last generation of neural models. Because the model is recent, very few dedicated simulation frameworks exist. This paper proposes a simulation framework developed in MATLAB that can be useful at: designing the network, uploading input stimuli, simulating the network, processing and displaying the results. The framework can be run on a network of computers by exploiting the parallelism of the model. The result is an improved performance by reducing simulation time.
Advances in Electrical and Computer Engineering, 2012
ABSTRACT
Sensors, 2020
Gesture recognition is an intensively researched area for several reasons. One of the most import... more Gesture recognition is an intensively researched area for several reasons. One of the most important reasons is because of this technology’s numerous application in various domains (e.g., robotics, games, medicine, automotive, etc.) Additionally, the introduction of three-dimensional (3D) image acquisition techniques (e.g., stereovision, projected-light, time-of-flight, etc.) overcomes the limitations of traditional two-dimensional (2D) approaches. Combined with the larger availability of 3D sensors (e.g., Microsoft Kinect, Intel RealSense, photonic mixer device (PMD), CamCube, etc.), recent interest in this domain has sparked. Moreover, in many computer vision tasks, the traditional statistic top approaches were outperformed by deep neural network-based solutions. In view of these considerations, we proposed a deep neural network solution by employing PointNet architecture for the problem of hand gesture recognition using depth data produced by a time of flight (ToF) sensor. We cre...
International Conference on Systems, Jul 14, 2011
This paper presents a problem of transportation planning that is able to achieve higher performan... more This paper presents a problem of transportation planning that is able to achieve higher performance by means of genetic optimization [2]. The paper introduces a model for simulating and evaluating the costs of a transportation system. It also describes a new way to use genetic optimization for a problem that requires adaptive chromosome length [1].
International Conference on Systems, Jul 22, 2010
This paper presents a modified version of the p-Delta learning algorithm. The algorithm can be us... more This paper presents a modified version of the p-Delta learning algorithm. The algorithm can be used for training a parallel perceptron regardless of the application. There are three significant changes from the original algorithm: an adaptive learning rate, a conscience mechanism and an adaptive margin enhancement mechanism. The changes offer an improved speed, stability and noise margin at the expense of complexity. The higher complexity can be a drawback if the algorithm is intended to be implemented in hardware.
Advances in Electrical and Computer Engineering, 2012
ABSTRACT
The hybrid Boost-L converter is suitable for practical applications that require a step-up conver... more The hybrid Boost-L converter is suitable for practical applications that require a step-up converter with wide conversion ratios. The filter stress is higher compared with the traditional Boost converter. The filter effort is reduced when using a multiphase configuration and interleaved switching. The two-phase converter is analyzed, simulated and the results are presented in this paper. Also a comparison with the single-phase Boost converter is made.
2021 IEEE 27th International Symposium for Design and Technology in Electronic Packaging (SIITME), 2021
2018 IEEE 18th International Power Electronics and Motion Control Conference (PEMC), 2018
A novel hybrid Buck-L converter with one active switch and three diodes is presented in this pape... more A novel hybrid Buck-L converter with one active switch and three diodes is presented in this paper. Compared to other topologies, the proposed converter is shown to be better suited in applications where a small difference between input and output voltage is required. It is shown that the proposed converter exhibits excellent efficiency over a wide range of the duty cycle. A DC analysis is performed and the static characteristics are derived, then device stresses are evaluated and compared to other buck topologies showing the superiority of the proposed solution. Then design equations are provided, and finally the converter is simulated and experimentally tested. The experimental results accurately confirmed the theoretical considerations and validated the feasibility of the proposed hybrid Buck topology.
This paper presents a modified version of the p-Delta learning algorithm. The algorithm can be us... more This paper presents a modified version of the p-Delta learning algorithm. The algorithm can be used for training a parallel perceptron regardless of the application. There are three significant changes from the original algorithm: an adaptive learning rate, a conscience mechanism and an adaptive margin enhancement mechanism. The changes offer an improved speed, stability and noise margin at the expense of complexity. The higher complexity can be a drawback if the algorithm is intended to be implemented in hardware.
2019 42nd International Conference on Telecommunications and Signal Processing (TSP), 2019
Deep Neural Networks and their associated learning paradigm, Deep Learning, represent one of the ... more Deep Neural Networks and their associated learning paradigm, Deep Learning, represent one of the hottest approaches used in image understanding and recognition. As their performances depends quasi-linearly on the amount of available data, the typical case studies in the literature assume the availability of huge datasets. This paper proposes to analyze several deep neural networks (trained from the scratch or pre-trained), test their efficiency in the problem of hand gesture recognition, and compare the results to a state-of-the-art classical method, the bag of features, for the case of small databases.
Energies, 2019
A new hybrid step-up converter suitable in applications where large conversion ratios are needed ... more A new hybrid step-up converter suitable in applications where large conversion ratios are needed is presented. The topology is still simple, containing only one transistor and three diodes. A detailed dc and ac analysis is performed and all design equations are provided. Compared to other topologies of the same type, the proposed converter exhibits lower or equal current and voltage stresses. A state space model is provided including the conduction losses and based on it the audio susceptibility and the control to output function are derived. As the converter is still of second order with the control to output transfer function exhibiting a right half plane zero, controller design is practically the same like in a Boost topology. It is shown how the proposed converter can be used in a photovoltaic system for performing the maximum power point tracking algorithm using the perturb and observe method. All theoretical considerations are verified through simulation and finally validated ...
2016 12th IEEE International Symposium on Electronics and Telecommunications (ISETC), 2016
The hybrid Boost-L converter is suitable for practical applications that require a step-up conver... more The hybrid Boost-L converter is suitable for practical applications that require a step-up converter with wide conversion ratios. The filter stress is higher compared with the traditional Boost converter. The filter effort is reduced when using a multiphase configuration and interleaved switching. The two-phase converter is analyzed, simulated and the results are presented in this paper. Also a comparison with the single-phase Boost converter is made.
This paper presents a problem of transportation planning that is able to achieve higher performan... more This paper presents a problem of transportation planning that is able to achieve higher performance by means of genetic optimization [2]. The paper introduces a model for simulating and evaluating the costs of a transportation system. It also describes a new way to use genetic optimization for a problem that requires adaptive chromosome length [1].
2010 9th International Symposium on Electronics and Telecommunications, 2010
Spiking Neural Networks are the last generation of neural models. Because the model is recent, ve... more Spiking Neural Networks are the last generation of neural models. Because the model is recent, very few dedicated simulation frameworks exist. This paper proposes a simulation framework developed in MATLAB that can be useful at: designing the network, uploading input stimuli, simulating the network, processing and displaying the results. The framework can be run on a network of computers by exploiting the parallelism of the model. The result is an improved performance by reducing simulation time.
Advances in Electrical and Computer Engineering, 2012
ABSTRACT
Sensors, 2020
Gesture recognition is an intensively researched area for several reasons. One of the most import... more Gesture recognition is an intensively researched area for several reasons. One of the most important reasons is because of this technology’s numerous application in various domains (e.g., robotics, games, medicine, automotive, etc.) Additionally, the introduction of three-dimensional (3D) image acquisition techniques (e.g., stereovision, projected-light, time-of-flight, etc.) overcomes the limitations of traditional two-dimensional (2D) approaches. Combined with the larger availability of 3D sensors (e.g., Microsoft Kinect, Intel RealSense, photonic mixer device (PMD), CamCube, etc.), recent interest in this domain has sparked. Moreover, in many computer vision tasks, the traditional statistic top approaches were outperformed by deep neural network-based solutions. In view of these considerations, we proposed a deep neural network solution by employing PointNet architecture for the problem of hand gesture recognition using depth data produced by a time of flight (ToF) sensor. We cre...