Biswas, Rupak, Bailey, David H., Kutler, Paul et al. · NASA STI Repository (National Aeronautics and Space Administration)
This paper examines the applications most commonly run on the supercomputers at the Numerical Aerospace Simulation (NAS) facility. It analyzes the extent to which such applications are fundamentally oriented to vector computers, and whether or not they can be efficiently implemented on hierarchical memory machines, such as systems with cache memories and highly parallel, distributed memory systems.
Cited by 0Open access
David H. Bailey, Rupak Biswas, Rob F. VanderWijngaart et al. · NASA Technical Reports Server (NASA) · 1997
This paper examines the applications most commonly run on the supercomputers at the Numerical Aerospace Simulation (NAS) facility. It analyzes the extent to which such applications are fundamentally oriented to vector computers, and whether or not they can be efficiently implemented on hierarchical memory machines, such as systems with cache memories and highly parallel, distributed memory systems.
Cited by 0
Beril Sırmaçek · InTech eBooks · 2018
This book is prepared as a combination of the manuscripts submitted by respected mathematicians and scientists around the world. As an editor, I truly enjoyed reading each manuscript. Not only will the methods and explanations help you to understand more about graph theory, but I also hope you will find it joyful to discover ways that you can apply graph theory in your scientific field. I believe the book can be read from the beginning to the end at once. However, the book can also be used as a reference guide in order to turn back to it when it is needed. I have to mention that this book assumes the reader to have a basic knowledge about graph theory. The very basics of the theory and terms are not explained at the beginner level. I hope this book will support many applied and research scientists from different scientific fields.
Cited by 5
Beril Sirmacek · Directory of Open access Books (OAPEN Foundation) · 2018
This book is prepared as a combination of the manuscripts submitted by respected mathematicians and scientists around the world. As an editor, I truly enjoyed reading each manuscript. Not only will the methods and explanations help you to understand more about graph theory, but I also hope you will find it joyful to discover ways that you can apply graph theory in your scientific field. I believe the book can be read from the beginning to the end at once. However, the book can also be used as a reference guide in order to turn back to it when it is needed. I have to mention that this book assumes the reader to have a basic knowledge about graph theory. The very basics of the theory and terms are not explained at the beginner level. I hope this book will support many applied and research scientists from different scientific fields.
Cited by 0Open access
Galety, Mohammad Gouse 1976-, Natarajan, Arul Kumar 1986-, Sriharsha, A. V. · Advances in systems analysis, software engineering, and high performance computing book series · 2023
Assist in understanding and applying the fundamentals of data structures and their many implementations. Data structures play a critical role in creating efficient applications. To improve one's ability to solve problems by utilizing proper data formats. Upon successful completion of reading this textbook, students will have gained the ability to: Recognize and use the most fundamental types of data structures and algorithms. Be familiar with the advantages and disadvantages of various data structures. Use the proper data structure in the given problem's solution context. Improve your ability to program to address a particular problem. To distinguish between programmers and non-programmers, it is necessary to know about Data Structures and Algorithms
Cited by 3
Shengjun Zhang · 2022
Networked multi-agent systems have attracted lots of researchers to develop algorithms, techniques, and applications.A multi-agent networked system consists of more than one subsystem (agent) to cooperately solve a global problem with only local computations and communications in a fully distributed manner. These networked systems have been investigated in various different areas including signal processing, control system, and machine learning. We can see massive applications using networked systems in reality, for example, persistent surveillance, healthcare, factory manufacturing, data mining, machine learning, power system, transportation system, and many other areas. Considering the nature of those mentioned applications, traditional centralized control and optimization algorithms which require both higher communication and computational capacities are not suitable. Additionally, compared to distributed control and optimization approaches, centralized control, and optimization algorithms cannot be scaled into systems with a large number of agents, or guarantee performance and security. All of the limitations of centralized control and optimization algorithms motivate us to investigate and develop new distributed control and optimization algorithms in networked systems. Moreover, convergence rate and analysis are crucial in control and optimization literature, which motivates us to investigate how to analyze and accerlate the convergence of distributed optimization algorithms.
Cited by 0Open access
N/A · IEEE RESOURCE CENTERS · 2024
Cited by 0Open access
A. Kaveh, Kiarash Biabani Hamedani · Studies in computational intelligence · 2022
Cited by 9
Lightcap AI, Alpay, Faruk · 2025
Cited by 0Open access
Nada, Ali · Hogskolan Ihalmstad (Halmstad University) · 2025
The increasing demand for high-speed, reliable wireless communication has driven the adoption of massive multiple-input multiple-output (MIMO) systems, which leverage a large number of antennas to enhance performance. Despite their potential, massive MIMO systems introduce substantial computational challenges, for instance in uplink detection. This thesis addresses these challenges by exploring software acceleration techniques to optimize the performance of massive MIMO systems. In the context of uplink detection, the study focuses on linear detection algorithms, such as zero-forcing (ZF) and minimum mean square error (MMSE) and employs graphics processing units (GPUs) to accelerate matrix operations. Techniques such as block Cholesky and QR decompositions were implemented to reduce computational overhead. The results demonstrate significant reductions in execution time and improvements in scalability, achieving notable speedups while balancing precision and performance trade-offs. Additionally, the thesis investigates the application of convolutional neural networks (CNNs) for channel state information (CSI)-based positioning. By optimizing CNN architectures and employing pruning techniques, the study enhances localization accuracy while minimizing computational requirements. These advancements enable precise positioning in resource-constrained environments, supporting advanced applications in 5G and beyond. The thesis also proposes future work directions, emphasizing the potential of hardware-based implementations using the dataflow model of computation within the compute abstraction layer (CAL) framework. By modelling algorithms as actor networks with explicit data dependencies, CAL facilitates efficient and scalable hardware designs, particularly for field-programmable gate arrays (FPGAs). This approach offers a promising pathway to address the increasing computational demands of next-generation massive MIMO systems.
Cited by 0Open access
Hasni Arezki · theses.fr (ABES) · 2024
A dynamical system models how a system evolves over time, governed by specific laws or equations. Ensuring the stability of such systems, under a given control action, is vital for predicting their long-term behavior. Stability analysis methods, such as Lyapunov’s theory and Input-to-State Stability (ISS), provide essential tools to assess whether a system will remain stable. When direct measurements of all system states are not possible, estimation techniques, such as observers and estimators, play a crucial role in reconstructing internal states from available measurements, enabling effective feedback control. Then, developing advanced estimation schemes and observer design methods is the main motivation of this thesis. Towards this end, three major contributions are proposed in this thesis, as detailed below: egin{enumerate} \item Analysis of Robust Moving Horizon Estimation (MHE) Schemes: Since the incremental exponential input/output-to-state stability (i-EIOSS) property is required to synthesize the parameters of the cost function in the MHE context, then, first, two numerical design procedures are proposed to ensure that a nonlinear system achieves i-EIOSS property and to compute the associated i-EIOSS parameters. Then, the robust stability of moving horizon estimation (MHE) is proven. Novel design conditions are established and advanced prediction techniques are introduced. \item Contributions to LMI-based and high-gain-based Observers: This chapter is split into two parts: the first part deals with observer design for nonlinear systems via Linear Matrix Inequalities (LMIs). The main goal consists of showing that for some families of nonlinear systems, the LMI-based observer design techniques always provide exponential convergent observer. The second part deals with high-gain observer methodology. A novel design method is proposed for systems with arbitrary nonlinear structures contrary to the standard results on high-gain observer methodology developed for triangular nonlinearities. \item Contributions to observer design for nonlinear systems with delayed measurements: The main idea behind consists of using a dynamic extension technique to transform a system with delayed nonlinear outputs into a system with linear outputs and a delay-dependent integral term in the dynamic process. Based on this transformation, novel and less conservative synthesis conditions are established, ensuring the exponential convergence of the observer despite the large values of the delay in the output measurements.
Cited by 0Open access
Lewis McMillan · St Andrews Research Repository (St Andrews Research Repository) · 2020
The Monte Carlo radiation transfer (MCRT) method can simulate the transport of light through turbid media. MCRT allows the modelling of multiple anisotropic scattering events, as well as a range of microphysics such as polarisation and fluorescence. This thesis concerns the development of several MCRT algorithms to solve various biophotonic and medically-related problems including modelling of tissue ablation and autofluorescent signals. An extension of the MCRT method through a theoretical quasi-wave/particle model is also demonstrated, allowing beam shapes with arbitrary phase profiles to be propagated. Tissue ablation can be used to treat acne scarring, Rhinophyma, and it can also be used to help enhance topical drug delivery. Currently the depth of ablation is not easily elucidated from a given laser or laser power setting. Therefore, a numerical tissue ablation model is developed using a combination of MCRT, a heat diffusion model, and a numerical tissue damage model to assess ablation crater depth and thermal damage to the surrounding tissue. Autofluorescence is the natural fluorescence of biological structures in tissue. Autofluorescence can be used as a biomarker of several diseases including: cardiovascular diseases, Alzheimers, and diabetes. However, the origin of the autofluorescence signal is not completely clear. The effect of tissue optics on the signal, which fluorophores contribute to the signal and by how much, and how different locations on the body can affect the signal are not well understood. This thesis presents a study of the effect of tissue optics on the autofluorescent signal. As part of this study, AmoebaMCRT was created to determine the relative concentrations of fluorophores for a given autofluorescent signal. Finally, we developed an extension to the MCRT method which allows the simulation of quasi-wave/particles. This method relies on the Huygens-Fresnel principle and the tracking of the phase of each individual photon packet. The extension, φMC, allows the modelling of complex beams that require the wave properties of light such as arbitrary order Bessel beams and Gaussian beams. We then use φMC to predict which beam, Bessel or Gaussian, performs “better" in a highly turbid medium.
Cited by 0Open access
Lobna El Harrouchi, Hanae Moussaoui, Mounir Karmoudi et al. · 2025
Deep learning has revolutionized artificial intelligence, but with very sensitive systems that exploit minute patterns inside data, able to solve impossible problems with fine nuance beyond anything that is even comprehensible by humans just a few years ago. The objective of this survey is to present an update on the state-of-the-art architectures, learning methods and computational techniques for GANs using relevant literature as source of information. We investigate Convolutional (CNNs), Recurrent (RNNs) and Long-Short-Term Memory networks, Transformers and attention mechanisms, Generative Adversarial Networks (GANs), Graph Neural Networks (GNNs), diffusion models, Deep Reinforcement Learning (DRL) and meta-learning. Theoretical backgrounds, mathematical definitions and training process are provided for each family, followed by association of their pros and cons to typical application domains as well as popular benchmarks. We also discuss cross-cutting issues interpretability, robustness, efficiency, ethics and privacy and summarize emerging research directions. To connect theory with practice, we add a comparative summary, case studies, and notes on implementation. Our aim with this work is to provide guidance to researchers, practitioners, and policy makers looking to responsibly harness advanced deep learning in practice.
Cited by 0
Venkata Ramana T, C. Balakrishnan, ХАДУЕВА Я.А. ХАДУЕВА Я.А. et al. · 2025
This paper features a detailed review of the current tools and algorithms and techniques that are used in the visualization of data and have greatly enhanced the analysis and interpretation of data. It includes overviews of Tableau, Power BI, D3.js and Paraview, Matplotlib and it provides an insight of their usage in various domains. The paper explores dimensionality reduction techniques including t-SNE, UMAP and PCA, clustering techniques including K-means and DBSCAN and graph visualization techniques. Focusing on the development and the present state, it also discusses new approaches, such as immersive visualization, real-time data streaming, and multimodal integration. These tools and techniques are transforming fields like data science, Artificial Intelligence, and scientific research, leading to better decision making, better methods of sharing data, enhanced insights. These advancements are being complemented by the use of AI, machine learning, and quantum computing in expanding the visualization capacities and encouraging further investigation and development.
Cited by 0
Arthur Charpentier · OpenEdition (OpenEdition) · 2020
Cited by 0Open access
Qiang Liu, Jialu Fan, Jin‐Xi Zhang et al. · IEEE Transactions on Industrial Informatics · 2023
Intelligent manufacturing has promoted the development of Industry 4.0 and enabled the manufacturing industry to gradually move into the stage of intelligence with the rapid development of the Internet of Things and the Industrial Internet. An intelligent manufacturing system is a manufacturing system that can automatically adapt to changing environments and varying process requirements with minimal supervision and assistance from operators. Therefore, intelligent manufacturing has become a recognized core high technology to enhance the overall competitiveness of the manufacturing industry. The goal of intelligent manufacturing is to make production resources form a circular network with the characteristics of autonomy, adjustability, and configurability, to develop production processes flexibly, and to realize the efficiency of individual customization. For example, by analyzing the factory floor data, equipment monitored data, and the enterprise manufacturing database, it could help to store, explore, and make complex decisions for the manufacturing system. To achieve this goal, modern information technologies, such as artificial intelligence, big data, cloud computing, and mobile Internet, modeling, control, and optimization need to be integrated and collaborated with the physical resources of the manufacturing process, which triggers new theory, solution algorithms, and application scenarios.
Cited by 4Open access
Yufei Chen · DR-NTU (Nanyang Technological University) · 2025
The quality of medical images is easily affected by various types of noise, which can affect accurate diagnosis and subsequent analysis. To systematically evaluate the validity of noise-reducing methods, this study focuses on two general modalities: Computed Tomography (CT) and Magnetic Resonance Imaging (MRI). The physical origins of noise in these modalities are first analyzed. Controllable simulations on clean images are conducted to generate a benchmark dataset. A range of denoising algorithms, from image processing-based traditional filters to state-of-the-art deep learning models, is then applied. Their performance is assessed through quantitative metrics like Peak Signal to Noise Ratio (PSNR) and Structural Similarity Index (SSIM). Furthermore, we directly evaluate the clinical utility of denoising by measuring its impact on the accuracy of a downstream segmentation task. Results demonstrate that deep learning-based methods outperform other methods in restoring image quality and in enhancing segmentation precision. The Transformer-based Restormer has the best performance.
Cited by 0
SILVESTRI, SIMONE · IRIS Research product catalog (Sapienza University of Rome) · 2009
Cited by 0
Manuel Lang · Repository for Publications and Research Data (ETH Zurich) · 2014
Videos have become an important part in our daily lives. Thankfully, the recent shift to an entire digital video pipeline allowed the introduction of improved tools and algorithms for professional video processing. Therefore, despite its complexity, video processing has never been easier to use and more available to the general public than it is today. However, due to the rapid technological advances new challenges in video processing have emerged. The ability to capture high definition content has often outpaced the progress of computing hardware. Furthermore, with the renaissance of stereoscopic 3D a new dimension is added to video processing algorithms. This additionally increases computational demands as well as it introduces its own sets of problems. Moreover, the ubiquitous of currently available displays results in a wide range of possible video output formats. Viewers consume video content on small handhelds as well as on high definition glasses-free 3D displays. This requires sophisticated video processing methods to ensure best viewing conditions on all output devices. This thesis aims at improving the way 2D and 3D video content can be adapted and artistically enhanced. The first two challenges emerge from the content-display gap. Often the resolution and aspect ratio of the output display may significantly vary from the original content. This thesis introduces a novel image domain warping (IDW) framework and shows that it can be utilized to fit content to different display dimensions without introducing too much noticeable artifacts. We present a novel efficient system that gives complete artistic control over this retargeting process. A similar content-display gap also exists for stereoscopic productions. Different display dimensions result in different depth perception. To counteract those undesired depth changes we will extend the technique of IDW to be able to change the depth perception of 3D content. Our novel method does not only allow to close the 3D-content display gap, but also gives artists a new innovative post processing tool to locally modify the depth composition. We introduce disparity mapping operators. These operators formalize insights from perception and production rules and are the basis for a general framework for stereoscopic disparity editing. Moreover, we use the same novel IDW approach to generate multi-view video
Cited by 0Open access
Farid Ghareh Mohammadi, M. Hadi Amini, Hamid R. Arabnia · Advances in intelligent systems and computing · 2020
Cited by 21