Article
Computer Science, Theory & Methods
Phillip Allen Lane, Joshua Dennis Booth
Summary: This paper introduces a heterogeneous format called CSR-k based on CSR, which achieves high-performance SpMV execution on different devices by reordering and grouping rows into hierarchical structures. It outperforms Intel MKL, NVIDIA cuSPARSE, and Sandia National Laboratories' KokkosKernels for regular sparse matrices.
PARALLEL COMPUTING
(2023)
Article
Computer Science, Theory & Methods
Elmira Karimi, Nicolas Bohm Agostini, Shi Dong, David Kaeli
Summary: This paper introduces a novel memory-aware format called VCSR, which out-performs previous formats on a GPU. VCSR achieves high thread-level parallelism and memory utilization by exploiting knowledge of GPU memory microarchitecture, reducing the number of global memory transactions, and providing a reordering mechanism. Experimental results demonstrate significant performance improvements of VCSR on different GPUs.
IEEE TRANSACTIONS ON PARALLEL AND DISTRIBUTED SYSTEMS
(2022)
Article
Chemistry, Multidisciplinary
Tadeusz Tomczak
Summary: The study analyzed the performance of an implementation based on a data-oriented language, showcasing the performance of a D2Q9 lattice solver on a GPU. While promising results were achieved, there were also issues with high and sometimes unpredictable overheads, particularly with sparse data structures.
APPLIED SCIENCES-BASEL
(2021)
Article
Computer Science, Theory & Methods
Jiaquan Gao, Yifei Xia, Renjie Yin, Guixia He
Summary: An adaptive sparse matrix-vector multiplication (SpMV) for diagonal sparse matrices on GPU, named DIA-Adaptive, is presented to automatically choose the ideal storage format and kernel, achieving high performance.
JOURNAL OF PARALLEL AND DISTRIBUTED COMPUTING
(2021)
Article
Computer Science, Information Systems
Alireza Akoushideh, Asadollah Shahbahrami, Abdorreza Joe Afshany
Summary: Automatic license plate recognition in intelligent transportation systems involves three main steps, including license plate detection, segmentation, and character recognition. The accuracy of the other steps relies on the success of the license plate detection step, which can be challenging due to various factors. A proposed algorithm in this paper uses mean filtering to remove noise from input images, computes the differences between the filtered image and the input image, and applies edge detection and morphological operations to detect license plate candidates. Experimental results on a real Iranian dataset show a 96.16% localization success rate.
MULTIMEDIA TOOLS AND APPLICATIONS
(2023)
Article
Computer Science, Theory & Methods
Jiaquan Gao, Qi Chen, Guixia He
Summary: This study introduces an efficient sparse approximate inverse preconditioning algorithm, GSPAI-Adaptive, on multiple GPUs. It presents a thread-adaptive allocation strategy for constructing the preconditioner and computes each component of the preconditioner in parallel inside a thread group of GPU, showing advantages over popular preconditioning algorithms and a latest parallel sparse approximate inverse preconditioning algorithm in experimental results.
PARALLEL COMPUTING
(2021)
Article
Computer Science, Hardware & Architecture
Wenchao Wu, Xuanhua Shi, Ligang He, Hai Jin
Summary: In this paper, a combination of optimization techniques for accelerating the performance of sampling-based GNN training process is proposed. The techniques include adaptive shared memory-based sampling, degree-guided thread block scheduling, and asynchronous pipeline-based scheduling. The experimental results show that the proposed methods can achieve up to 5.6X performance speedup compared to existing work.
IEEE TRANSACTIONS ON COMPUTERS
(2023)
Article
Computer Science, Theory & Methods
Diogo Nunes, Daniel Castro, Paolo Romano
Summary: This paper introduces CSMV, a multi-versioned Software TM for GPUs with an innovative client-server design. CSMV provides the benefits of utilizing fast on chip memory and implementing efficient collaborative commit procedures, leading to significant speed-ups compared to state of the art STMs for GPUs and CPUs.
JOURNAL OF PARALLEL AND DISTRIBUTED COMPUTING
(2023)
Article
Engineering, Electrical & Electronic
Carlos de Cea-Dominguez, Juan C. Moure, Joan Bartrina-Rapesta, Francesc Auli-Llinas
Summary: This paper presents a video codec that can process 16K video in real time with high throughput, and can trade off throughput for coding performance based on user's needs. Experimental results show that this method can double the throughput achieved by CPU implementations of High-Throughput JPEG2000 and HEVC on a GPU.
SIGNAL PROCESSING-IMAGE COMMUNICATION
(2021)
Article
Computer Science, Software Engineering
Guixia He, Qi Chen, Jiaquan Gao
Summary: This paper introduces a new diagonal storage format RBDCS and proposes an efficient SpMV kernel for handling multidiagonal sparse matrices. Experimental results demonstrate that the RBDCS kernel outperforms popular diagonal SpMV kernels.
CONCURRENCY AND COMPUTATION-PRACTICE & EXPERIENCE
(2021)
Article
Computer Science, Theory & Methods
Pieter Ghysels, Ryan Synk
Summary: The numerical factorization and triangular solve phases of the sparse direct solver STRUMPACK have been ported to GPU, achieving high performance through various optimizations.
PARALLEL COMPUTING
(2022)
Article
Mathematics, Applied
Siham Boukhris, Artem Napov, Yvan Notay
Summary: We propose a new sparse matrix format for discretized partial differential equations with piecewise-constant coefficients. This format saves memory and is suitable for parallel computing on GPUs. It is well suited for algebraic multigrid methods and has been shown to outperform other solvers in terms of both run time and memory usage.
SIAM JOURNAL ON SCIENTIFIC COMPUTING
(2023)
Article
Computer Science, Theory & Methods
Jianhua Gao, Weixing Ji, Zhaonian Tan, Yizhuo Wang, Feng Shi
Summary: This article presents a new approach for compressed binary sparse matrix-vector multiplication, which reduces data transfer and improves computational performance through partitioning and encoding.
IEEE TRANSACTIONS ON PARALLEL AND DISTRIBUTED SYSTEMS
(2022)
Article
Computer Science, Interdisciplinary Applications
E. Coronado-Barrientos, M. Antonioletti, A. Garcia-Loureiro
Summary: The paper introduces a new sparse matrix storage format AXT, which improves SpMV performance on vector capability platforms. By optimizing different subvariants of AXT and comparing performance on Intel and NVIDIA platforms, it is shown that AXT outperforms AXC and CSR significantly.
ADVANCES IN ENGINEERING SOFTWARE
(2021)
Article
Computer Science, Theory & Methods
Anil Gaihre, Xiaoye Sherry Li, Hang Liu
Summary: LU decomposition of a matrix is a crucial operation in numerical linear algebra. This article introduces GSOFA, the first GPU-based symbolic factorization design, which achieves significant speedup through optimized algorithms and reduced space consumption.
IEEE TRANSACTIONS ON PARALLEL AND DISTRIBUTED SYSTEMS
(2022)
Article
Chemistry, Multidisciplinary
Permatasari Silitonga, Alhadi Bustamam, Hengki Muradi, Wibowo Mangunwardoyo, Beti E. Dewi
Summary: The study developed models using Artificial Neural Network (ANN) and Discriminant Analysis (DA) to predict the severity level of dengue based on laboratory test results, achieving high accuracy of 90.91%, sensitivity of 91.11%, and specificity of 95.51%. The proposed model can assist physicians in timely predicting and treating dengue patients to prevent fatal cases.
APPLIED SCIENCES-BASEL
(2021)
Article
Computer Science, Artificial Intelligence
Simon M. Thomas, James G. Lefevre, Glenn Baxter, Nicholas A. Hamilton
Summary: This study applies interpretable deep learning methods to analyze the most common skin cancers in a histological setting, demonstrating the potential for automatic machine analysis of dermatopathology work. By characterizing tissue into meaningful dermatological classes, the research aims to pave the way for future computer aided diagnosis systems with human interpretable outcomes.
MEDICAL IMAGE ANALYSIS
(2021)
Article
Biochemical Research Methods
James G. Lefevre, Yvette W. H. Koh, Adam A. Wall, Nicholas D. Condon, Jennifer L. Stow, Nicholas A. Hamilton
Summary: This study introduces LLAMA, a platform for systematic analysis of terabyte-scale 4D microscopy datasets. The system utilizes machine learning for semantic segmentation and object analysis and tracking algorithms, running on high-performance computing to achieve high throughput. LLAMA provides detailed numerical and visual outputs for effective statistical analysis and has the capacity to screen large datasets for specific structural configurations.
BMC BIOINFORMATICS
(2021)
Article
Biology
Oky Hermansyah, Alhadi Bustamam, Arry Yanuar
Summary: In this study, a virtual screening workflow was developed using quantitative structure-activity relationship (QSAR) strategy based on artificial intelligence to identify DPP-4-inhibitor hit compounds selective against DPP-8 and DPP-9. The study utilized regression and classification machine learning algorithms to build the virtual screening workflows, resulting in the identification of potential hit compounds with high inhibitory potential against DPP-4 and low inhibitory potential against DPP-8 and DPP-9. This technique showed effectiveness in identifying DPP-4-inhibitor hit compounds and has potential applications for discovering hit compounds of other targets.
COMPUTATIONAL BIOLOGY AND CHEMISTRY
(2021)
Article
Biology
Alexander P. Browning, Jesse A. Sharp, Ryan J. Murphy, Gency Gunasingh, Brodie Lawson, Kevin Burrage, Nikolas K. Haass, Matthew Simpson
Summary: Tumour spheroids are common experimental models in vitro, closely mimicking avascular tumour growth. Research suggests that spheroids have a limiting structure and the mature structure is independent of seeding density. Additionally, comparing spheroid structure rather than size produces more accurate results.
Article
Computer Science, Information Systems
Prasnurzaki Anki, Alhadi Bustamam, Rinaldi Anwar Buyung
Summary: This study focuses on the application of sentence classification using the News Aggregator Dataset to create a chatbot program. Results show that the 1D CNN Transpose model achieves the highest accuracy in testing. Through testing four models via multimodal implementation, accurate sentence prediction and detection are expected for both types of chatbot.
Article
Computer Science, Information Systems
Radifa Hilya Paradisa, Alhadi Bustamam, Wibowo Mangunwardoyo, Andi Arus Victor, Anggun Rama Yudantha, Prasnurzaki Anki
Summary: Fundus image is crucial for disease detection, especially for early diagnosis of diabetic retinopathy. The demand for ophthalmologists who can read fundus images is increasing, leading to a need for an automated diagnostic system. This study proposes a deep learning approach using a concatenate model for fundus image classification, yielding improved accuracy and F1-score compared to a single model.
Article
Multidisciplinary Sciences
Simon M. Thomas, James G. Lefevre, Glenn Baxter, Nicholas A. Hamilton
Summary: The dataset includes 290 hand-annotated histopathology tissue sections of the three most common skin cancers, each with a segmentation mask dividing the tissue into 12 types. It also provides cancer margin measurements for automated assessment, allowing researchers to build upon recent work in skin cancer image analysis.
Article
Mathematics, Applied
Kevin Burrage, Pamela M. Burrage, Grant Lythe
Summary: This paper presents an algorithm for homogeneous diffusive motion on a sphere by considering the equivalent process of a randomly rotating spin vector. By introducing appropriate sets of random variables, families of methods are constructed that effectively preserve the spin modulus for every realization, achieved by exponentiating an antisymmetric matrix.
NUMERICAL ALGORITHMS
(2022)
Article
Multidisciplinary Sciences
Gloria M. Monsalve-Bravo, Brodie A. J. Lawson, Christopher Drovandi, Kevin Burrage, Kevin S. Brown, Christopher M. Baker, Sarah A. Vollert, Kerrie Mengersen, Eve McDonald-Madden, Matthew P. Adams
Summary: This work introduces a comprehensive approach to assess the sensitivity of model outputs to changes in parameter values, constrained by the combination of prior beliefs and data. It identifies stiff parameter combinations affecting the model-data fit, and reveals which of these combinations are primarily influenced by the data or the priors. The technique is beneficial in contexts where data is limited compared to the number of model parameters, and has applications in biochemistry, ecology, and cardiac electrophysiology. It also helps uncover controlling mechanisms and guide parameter prioritization for improved parameter inference.
Article
Mathematics, Applied
Brody H. H. Foy, Kevin Burrage, Ian Turner
Summary: This study proposes a meshfree numerical scheme based on strong-form finite volume style formulations. The technique uses radial basis functions to interpolate the problem domain and approximate fluxes in a disjoint finite volume scheme, eliminating the reliance on a mesh structure. The method shows potential for applications in porous media modeling and computational fluid dynamics.
NUMERICAL METHODS FOR PARTIAL DIFFERENTIAL EQUATIONS
(2023)
Article
Mathematics, Applied
Shamika Kekulthotuwage Don, Kevin Burrage, Kate J. Helmstedt, Pamela M. Burrage
Summary: We investigated the dynamical properties of discrete systems under two settings: discrete and continuous. By discretizing time, we obtained stability conditions that maintain the characteristics of continuous models and their numerical approximations. We found that small changes in model parameters can alter system dynamics unless an appropriate time discretization is chosen. We also observed similar dynamical properties in Ricker-type predator-prey systems under certain conditions. Our results highlight the importance of preliminary analysis in determining agreement or disagreement between the dynamical properties of approximated discrete systems and their continuous counterparts.
Article
Multidisciplinary Sciences
Philipp Henning, Till Koester, Fiete Haack, Kevin Burrage, Adelinde M. Uhrmacher
Summary: Studying membrane dynamics is crucial for understanding cellular response to environmental stimuli. The plasma membrane's compartmental structure, created by actin-based membrane-skeleton and anchored transmembrane proteins, plays an important role in this process. Particle-based reaction-diffusion simulation offers a suitable approach for analyzing the membrane's stochastic and spatially heterogeneous dynamics. However, different methods for modeling the compartmental structure have their own constraints and impact on simulation results and performance.
ROYAL SOCIETY OPEN SCIENCE
(2023)
Review
Multidisciplinary Sciences
Jesse A. Sharp, Kevin Burrage, Matthew J. Simpson
Summary: This review discusses the application of Pontryagin's maximum principle (PMP) in optimal control and the implementation of the forward-backward sweep method (FBSM). By conceptualizing FBSM as a fixed point iteration process and adapting existing acceleration techniques, the rate of convergence can be improved without costly tuning. Moreover, these methods can induce convergence in cases where the FBSM fails to converge.
JOURNAL OF THE ROYAL SOCIETY INTERFACE
(2021)
Article
Computer Science, Theory & Methods
Alhadi Bustamam, Haris Hamzah, Nadya A. Husna, Sarah Syarofina, Nalendra Dwimantara, Arry Yanuar, Devvi Sarwinda
Summary: This study aims to develop new DPP-4 inhibitors for the treatment of type 2 diabetes with low adverse effects using QSAR models built with Rotation Forest and Deep Neural Network. K-modes clustering and CatBoost are utilized for molecule selection and feature selection, resulting in QSAR models with high performance metrics. The study concludes that feature selection using CatBoost before building QSAR models is essential for accurate predictions.
JOURNAL OF BIG DATA
(2021)