Статті в журналах з теми "Distribution shared memory"
Щоб переглянути інші типи публікацій з цієї теми, перейдіть за посиланням: Distribution shared memory.
Оформте джерело за APA, MLA, Chicago, Harvard та іншими стилями
Ознайомтеся з топ-50 статей у журналах для дослідження на тему "Distribution shared memory".
Біля кожної праці в переліку літератури доступна кнопка «Додати до бібліографії». Скористайтеся нею – і ми автоматично оформимо бібліографічне посилання на обрану працю в потрібному вам стилі цитування: APA, MLA, «Гарвард», «Чикаго», «Ванкувер» тощо.
Також ви можете завантажити повний текст наукової публікації у форматі «.pdf» та прочитати онлайн анотацію до роботи, якщо відповідні параметри наявні в метаданих.
Переглядайте статті в журналах для різних дисциплін та оформлюйте правильно вашу бібліографію.
1
Chandra, Rohit, Ding-Kai Chen, Robert Cox, Dror E. Maydan, Nenad Nedeljkovic, and Jennifer M. Anderson. "Data distribution support on distributed shared memory multiprocessors." ACM SIGPLAN Notices 32, no. 5 (May 1997): 334–45. http://dx.doi.org/10.1145/258916.258945.
Повний текст джерела
2
Marzolla, Moreno, and Gabriele D’angelo. "Parallel Data Distribution Management on Shared-memory Multiprocessors." ACM Transactions on Modeling and Computer Simulation 30, no. 1 (February 8, 2020): 1–25. http://dx.doi.org/10.1145/3369759.
Повний текст джерела
3
Axtmann, Michael, Sascha Witt, Daniel Ferizovic, and Peter Sanders. "Engineering In-place (Shared-memory) Sorting Algorithms." ACM Transactions on Parallel Computing 9, no. 1 (March 31, 2022): 1–62. http://dx.doi.org/10.1145/3505286.
Повний текст джерелаАнотація:
We present new sequential and parallel sorting algorithms that now represent the fastest known techniques for a wide range of input sizes, input distributions, data types, and machines. Somewhat surprisingly, part of the speed advantage is due to the additional feature of the algorithms to work in-place, i.e., they do not need a significant amount of space beyond the input array. Previously, the in-place feature often implied performance penalties. Our main algorithmic contribution is a blockwise approach to in-place data distribution that is provably cache-efficient. We also parallelize this approach taking dynamic load balancing and memory locality into account. Our new comparison-based algorithm In-place Parallel Super Scalar Samplesort ( IPS 4 o ) , combines this technique with branchless decision trees. By taking cases with many equal elements into account and by adapting the distribution degree dynamically, we obtain a highly robust algorithm that outperforms the best previous in-place parallel comparison-based sorting algorithms by almost a factor of three. That algorithm also outperforms the best comparison-based competitors regardless of whether we consider in-place or not in-place, parallel or sequential settings. Another surprising result is that IPS 4 o even outperforms the best (in-place or not in-place) integer sorting algorithms in a wide range of situations. In many of the remaining cases (often involving near-uniform input distributions, small keys, or a sequential setting), our new In-place Parallel Super Scalar Radix Sort ( IPS 2 Ra ) turns out to be the best algorithm. Claims to have the – in some sense – “best” sorting algorithm can be found in many papers which cannot all be true. Therefore, we base our conclusions on an extensive experimental study involving a large part of the cross product of 21 state-of-the-art sorting codes, 6 data types, 10 input distributions, 4 machines, 4 memory allocation strategies, and input sizes varying over 7 orders of magnitude. This confirms the claims made about the robust performance of our algorithms while revealing major performance problems in many competitors outside the concrete set of measurements reported in the associated publications. This is particularly true for integer sorting algorithms giving one reason to prefer comparison-based algorithms for robust general-purpose sorting.
4
O'Hearn, Kurt A., Abdullah Alperen, and Hasan Metin Aktulga. "Fast Solvers for Charge Distribution Models on Shared Memory Platforms." SIAM Journal on Scientific Computing 42, no. 1 (January 2020): C1—C22. http://dx.doi.org/10.1137/18m1224684.
Повний текст джерела
5
BYNA, SURENDRA, KIRK W. CAMERON, and XIAN-HE SUN. "ISOLATING COSTS IN SHARED MEMORY COMMUNICATION BUFFERING." Parallel Processing Letters 15, no. 04 (December 2005): 357–65. http://dx.doi.org/10.1142/s0129626405002271.
Повний текст джерелаАнотація:
Communication in parallel applications is a combination of data transfers internally at a source or destination and across the network. Previous research focused on quantifying network transfer costs has indirectly resulted in reduced overall communication cost. Optimized data transfer from source memory to the network interface has received less attention. In shared memory systems, such memory-to-memory transfers dominate communication cost. In distributed memory systems, memory-to-network interface transfers grow in significance as processor and network speeds increase at faster rates than memory latency speeds. Our objective is to minimize the cost of internal data transfers. The following examples illustrating the impact of memory transfers on communication, we present a methodology for classifying the effects of data size and data distribution on hardware, middleware, and application software performance. This cost is quantified using hardware counter event measurements on the SGI Origin 2000. For the SGI O2K, we empirically identify the cost caused by just copying data from one buffer to another and the middleware overhead. We use MPICH in our experiments, but our techniques are generally applicable to any communication implementation.
6
Addison, C., Y. Ren, and M. van Waveren. "OpenMP Issues Arising in the Development of Parallel BLAS and LAPACK Libraries." Scientific Programming 11, no. 2 (2003): 95–104. http://dx.doi.org/10.1155/2003/278167.
Повний текст джерелаАнотація:
Dense linear algebra libraries need to cope efficiently with a range of input problem sizes and shapes. Inherently this means that parallel implementations have to exploit parallelism wherever it is present. While OpenMP allows relatively fine grain parallelism to be exploited in a shared memory environment it currently lacks features to make it easy to partition computation over multiple array indices or to overlap sequential and parallel computations. The inherent flexible nature of shared memory paradigms such as OpenMP poses other difficulties when it becomes necessary to optimise performance across successive parallel library calls. Notions borrowed from distributed memory paradigms, such as explicit data distributions help address some of these problems, but the focus on data rather than work distribution appears misplaced in an SMP context.
7
Crooks, P., and R. H. Perrott. "Language Constructs for Data Partitioning and Distribution." Scientific Programming 4, no. 2 (1995): 59–85. http://dx.doi.org/10.1155/1995/656010.
Повний текст джерелаАнотація:
This article presents a survey of language features for distributed memory multiprocessor systems (DMMs), in particular, systems that provide features for data partitioning and distribution. In these systems the programmer is freed from consideration of the low-level details of the target architecture in that there is no need to program explicit processes or specify interprocess communication. Programs are written according to the shared memory programming paradigm but the programmer is required to specify, by means of directives, additional syntax or interactive methods, how the data of the program are decomposed and distributed.
8
Nikolopoulos, Dimitrios S., Ernest Artiaga, Eduard Ayguadé, and Jesús Labarta. "Scaling Non-Regular Shared-Memory Codes by Reusing Custom Loop Schedules." Scientific Programming 11, no. 2 (2003): 143–58. http://dx.doi.org/10.1155/2003/379739.
Повний текст джерелаАнотація:
In this paper we explore the idea of customizing and reusing loop schedules to improve the scalability of non-regular numerical codes in shared-memory architectures with non-uniform memory access latency. The main objective is to implicitly setup affinity links between threads and data, by devising loop schedules that achieve balanced work distribution within irregular data spaces and reusing them as much as possible along the execution of the program for better memory access locality. This transformation provides a great deal of flexibility in optimizing locality, without compromising the simplicity of the shared-memory programming paradigm. In particular, the programmer does not need to explicitly distribute data between processors. The paper presents practical examples from real applications and experiments showing the efficiency of the approach.
9
Bozkus, Zeki, Larry Meadows, Steven Nakamoto, Vincent Schuster, and Mark Young. "PGHPF – An Optimizing High Performance Fortran Compiler for Distributed Memory Machines." Scientific Programming 6, no. 1 (1997): 29–40. http://dx.doi.org/10.1155/1997/705102.
Повний текст джерелаАнотація:
High Performance Fortran (HPF) is the first widely supported, efficient, and portable parallel programming language for shared and distributed memory systems. HPF is realized through a set of directive-based extensions to Fortran 90. It enables application developers and Fortran end-users to write compact, portable, and efficient software that will compile and execute on workstations, shared memory servers, clusters, traditional supercomputers, or massively parallel processors. This article describes a production-quality HPF compiler for a set of parallel machines. Compilation techniques such as data and computation distribution, communication generation, run-time support, and optimization issues are elaborated as the basis for an HPF compiler implementation on distributed memory machines. The performance of this compiler on benchmark programs demonstrates that high efficiency can be achieved executing HPF code on parallel architectures.
10
Warren, Karen H. "PDDP, A Data Parallel Programming Model." Scientific Programming 5, no. 4 (1996): 319–27. http://dx.doi.org/10.1155/1996/857815.
Повний текст джерелаАнотація:
PDDP, the parallel data distribution preprocessor, is a data parallel programming model for distributed memory parallel computers. PDDP implements high-performance Fortran-compatible data distribution directives and parallelism expressed by the use of Fortran 90 array syntax, the FORALL statement, and the WHERE construct. Distributed data objects belong to a global name space; other data objects are treated as local and replicated on each processor. PDDP allows the user to program in a shared memory style and generates codes that are portable to a variety of parallel machines. For interprocessor communication, PDDP uses the fastest communication primitives on each platform.
11
Chapman, Barbara, Piyush Mehrotra, and Hans Zima. "Programming in Vienna Fortran." Scientific Programming 1, no. 1 (1992): 31–50. http://dx.doi.org/10.1155/1992/258136.
Повний текст джерелаАнотація:
Exploiting the full performance potential of distributed memory machines requires a careful distribution of data across the processors. Vienna Fortran is a language extension of Fortran which provides the user with a wide range of facilities for such mapping of data structures. In contrast to current programming practice, programs in Vienna Fortran are written using global data references. Thus, the user has the advantages of a shared memory programming paradigm while explicitly controlling the data distribution. In this paper, we present the language features of Vienna Fortran for FORTRAN 77, together with examples illustrating the use of these features.
12
FANTOZZI, CARLO, ANDREA PIETRACAPRINA, and GEPPINO PUCCI. "A GENERAL PRAM SIMULATION SCHEME FOR CLUSTERED MACHINES." International Journal of Foundations of Computer Science 14, no. 06 (December 2003): 1147–64. http://dx.doi.org/10.1142/s0129054103002230.
Повний текст джерелаАнотація:
We present a general deterministic scheme to implement a shared memory abstraction on any distributed-memory machine which exhibits a clustered structure. More specifically, we develop a memory distribution strategy and an access protocol for the Decomposable BSP (D-BSP), a generic machine model whose bandwidth/latency parameters can be instantiated to closely reflect the characteristics of machines that admit a hierarchical decomposition into independent clusters. Our scheme achieves provably optimal slowdown for those machines where delays due to latency dominate over those due to bandwidth limitations. For machines where this is not the case, the slowdown is a mere logarithmic factor away from the natural bandwidth-based lower bound.
13
VEE, VOON-YEE, and WEN-JING HSU. "LOCALITY-PRESERVING LOAD-BALANCING MECHANISMS FOR SYNCHRONOUS SIMULATIONS ON SHARED-MEMORY MULTIPROCESSORS." Parallel Processing Letters 10, no. 01 (March 2000): 111–32. http://dx.doi.org/10.1142/s0129626400000123.
Повний текст джерелаАнотація:
In the past decade, many synchronous algorithms have been proposed for parallel and discrete simulations. However, the actual performance of these algorithms have been far from ideal, especially when event granularity is small. Barring the case of low parallelism in the given simulation models, one of the main reasons of low speedups is in the uneven load distribution among processors. To amend for this, both static and dynamic load balancing approaches have been proposed. Nevertheless, static schemes based on partitioning of LPs are often subject to the dynamic behavior of the specific simulation models and are therefore application dependent; dynamic load balancing schemes, on the other hand, often suffer from loss of localities and hence cache misses, which could severely penalize on fine-grained event processing. In this paper, we present several new locality-preserving load balancing mechanisms for synchronous simulations on shared-memory multiprocessors. We focus on the type of synchronous simulations where the number of LPs to be processed within a cycle decreases monotonically. We show both theoretically and empirically that some of these mechanisms incur very low overhead. The mechanisms have been implemented by using MIT's Cilk and tested with a number of simulation applications. The results confirm that one of the new mechanisms is indeed more efficient and scalable than common existing approaches.
14
Wang, Hung-Wen, Chia-Wei Tsai, Jason Lin, and Chun-Wei Yang. "Authenticated Semi-Quantum Key Distribution Protocol Based on W States." Sensors 22, no. 13 (July 2, 2022): 4998. http://dx.doi.org/10.3390/s22134998.
Повний текст джерелаАнотація:
In 2019, Wen et al. proposed authenticated semi-quantum key distribution (ASQKD) for identity and message using the teleportation of W states and GHZ-like states without pre-shared keys. However, the ASQKD protocol presents a vital issue in the teleportation of W states owing to its inappropriate design. Bob recovers the teleported W states without obtaining the position of the corresponding photons and then returns the recovered photons back to Alice. Hence, the teleportation of W states in Wen et al.’s ASQKD protocol was malfunctioning. Moreover, Wen et al.’s ASQKD protocol requires quantum memory, which strongly disobeys the definition of semi-quantum proposed by Boyer et al. Therefore, in this study, we discover the flaws of Wen et al.’s ASQKD protocol and propose an authenticated semi-quantum key distribution protocol. When compared to Wen et al.’s ASQKD protocol, the proposed ASQKD protocol has the following advantages: legal semi-quantum environment (i.e., does not require quantum memory), reduced quantum hardware requirement (i.e., based only on W states), does not involve classical cryptography (i.e., the hash function), and provided 1.6 times higher qubit efficiency.
15
Liu, Zheng, Marcelo H. Ang Jr., and Winston Khoon Guan Seah. "A Searching and Tracking Framework for Multi-Robot Observation of Multiple Moving Targets." Journal of Advanced Computational Intelligence and Intelligent Informatics 8, no. 1 (January 20, 2004): 14–22. http://dx.doi.org/10.20965/jaciii.2004.p0014.
Повний текст джерелаАнотація:
The "museum problem" is a typical research topic on multi-robot observation of multiple moving targets. The objective of museum problem is to optimize the distribution of robots, such that the maximal moving targets can be observed. In this paper, we present our memory based searching and artificial potential field based tracking framework for museum problem. For searching, a memory table, either local or shared, can help shorten the searching time for targets. For tracking, our artificial potential field based motion control provides real-time tracking of moving targets with collision avoidance. Qualitative simulations demonstrate the capability of our searching and tracking framework.
16
Yang, Fan, Sandra C. A. Nielsen, Ramona A. Hoh, Katharina Röltgen, Oliver Fabian Wirz, Emily Haraguchi, Grace H. Jean, et al. "Shared B cell memory to coronaviruses and other pathogens varies in human age groups and tissues." Science 372, no. 6543 (April 12, 2021): 738–41. http://dx.doi.org/10.1126/science.abf6648.
Повний текст джерелаАнотація:
Vaccination and infection promote the formation, tissue distribution, and clonal evolution of B cells, which encode humoral immune memory. We evaluated pediatric and adult blood and deceased adult organ donor tissues to determine convergent antigen-specific antibody genes of similar sequences shared between individuals. B cell memory varied for different pathogens. Polysaccharide antigen–specific clones were not exclusive to the spleen. Adults had higher clone frequencies and greater class switching in lymphoid tissues than blood, while pediatric blood had abundant class-switched convergent clones. Consistent with reported serology, prepandemic children had class-switched convergent clones to severe acute respiratory syndrome coronavirus 2 with weak cross-reactivity to other coronaviruses, while adult blood or tissues showed few such clones. These results highlight the prominence of early childhood B cell clonal expansions and cross-reactivity for future responses to novel pathogens.
17
Ioannidis, Sotiris, Umit Rencuzogullari, Robert Stets, and Sandhya Dwarkadas. "CRAUL: Compiler and Run-Time Integration for Adaptation under Load." Scientific Programming 7, no. 3-4 (1999): 261–73. http://dx.doi.org/10.1155/1999/603478.
Повний текст джерелаАнотація:
Clusters of workstations provide a cost‐effective, high performance parallel computing environment. These environments, however, are often shared by multiple users, or may consist of heterogeneous machines. As a result, parallel applications executing in these environments must operate despite unequal computational resources. For maximum performance, applications should automatically adapt execution to maximize use of the available resources. Ideally, this adaptation should be transparent to the application programmer. In this paper, we present CRAUL (Compiler and Run‐Time Integration for Adaptation Under Load), a system that dynamically balances computational load in a parallel application. Our target run‐time is software‐based distributed shared memory (SDSM). SDSM is a good target for parallelizing compilers since it reduces compile‐time complexity by providing data caching and other support for dynamic load balancing. CRAUL combines compile‐time support to identify data access patterns with a run‐time system that uses the access information to intelligently distribute the parallel workload in loop‐based programs. The distribution is chosen according to the relative power of the processors and so as to minimize SDSM overhead and maximize locality. We have evaluated the resulting load distribution in the presence of different types of load – computational, computational and memory intensive, and network load. CRAUL performs within 5–23% of ideal in the presence of load, and is able to improve on naive compiler‐based work distribution that does not take locality into account even in the absence of load.
18
Smith, Barry F., and William D. Gropp. "The Design of Data-Structure-Neutral Libraries for the Iterative Solution of Sparse Linear Systems." Scientific Programming 5, no. 4 (1996): 329–36. http://dx.doi.org/10.1155/1996/417629.
Повний текст джерелаАнотація:
Over the past few years several proposals have been made for the standardization of sparse matrix storage formats in order to allow for the development of portable matrix libraries for the iterative solution of linear systems. We believe that this is the wrong approach. Rather than define one standard (or a small number of standards) for matrix storage, the community should define an interface (i.e., the calling sequences) for the functions that act on the data. In addition, we cannot ignore the interface to the vector operations because, in many applications, vectors may not be stored as consecutive elements in memory. With the acceptance of shared memory, distributed memory, and cluster memory parallel machines, the flexibility of the distribution of the elements of vectors is also extremely important. This issue is ignored in most proposed standards. In this article we demonstrate how such libraries may be written using data encapsulation techniques.
19
Bagnara, Davide, Margherita Squillario, David Kipling, Thierry Mora, Aleksandra Walczak, Deborah K. Dunn-Walters, Jean-Claude Weill, and Claude-Agnès Reynaud. "High-Throughput Ig Sequencing of Paired Blood and Spleen Samples Allows a Redefinition of Memory IgM Subsets in Humans." Blood 124, no. 21 (December 6, 2014): 565. http://dx.doi.org/10.1182/blood.v124.21.565.565.
Повний текст джерелаАнотація:
Abstract In humans, whether B cells with the IgM+IgD+CD27+ phenotype represent an independent lineage involved in T-independent responses, similar to mouse marginal zone B cells, or whether they are part of the germinal center-derived memory B-cell pool generated during responses to T-dependent antigens, is still a debated issue. To address this question, we performed high-throughput Ig sequencing of B-cell subsets from paired blood and spleen samples and analyzed the clonal relationships between them. We isolated and analyzed 3 different B cell subsets based on CD27 and IgD staining from both blood and spleen: IgD+CD27+ (MZ) - amplified with Cmu primers IgD-CD27+ (switched and IgM-only) with Cmu, Cgamma and Calpha primers IgD-CD27- (CD27- memory or double-negative DN) with the same three primers We obtained 95729 unique sequences that clustered in 49199 different clones: 1125 clones were shared between blood and spleen of the same B-cell subset, and 1681 clones were shared between different subsets, allowing us to trace their relationships. We analyzed these clones that share sequences from different subsets/tissues for their mutation frequency distribution, CDR3-length, and VH/JH family usage, and compared these different characteristics with the bulk of sequences from their respective subset of origin. The analysis of clones shared between blood and spleen for switched IgG/IgA and for MZ subsets suggests different recirculation dynamics. For switched cells, the blood appears to be a mixture of splenic and other lymphoid tissues B cells. For MZ B cells in contrast, the blood appear to be only composed of a subgroup of the splenic repertoire, in agreement with the observation that marginal zone B cells recirculate and are mainly generated in the spleen. Clonal relationships between the IgM clones (originating from the MZ, IgM-only and double negative compartments) show that the clones involved display the characteristics of IgM-only B cells whatever their subset of origin, even in the case of the paired MZ/double-negative sequences that were not supposed to include IgM-only sequences. We therefore conclude that the clones shared between the various IgM subsets do not represent b between them, but rather correspond to a heterogeneous phenotype of the IgM-only population that concerns both IgD and CD27 expression, leading to a partial overlap with the MZ and double-negative gates. Clones shared between the MZ and the switched IgG and IgA compartment also show, for their IgM part, the mutation and repertoire characteristics of IgM-only cells and not of MZ B cells, reinforcing the conclusion that IgM-only are true memory B cells, and constitute the only subset showing clonal relationships with switched memory B cells. In summary, we report that MZ B cells have different recirculation characteristics and do not show real clonal relationships with IgM-only and switched memory B cells, in agreement with the notion that they represent a distinct differentiation pathway. In contrast, the only precursor-product relationship between IgM memory and switched B cells appear to concern a B cell subset that has been described as "IgM-only", but appears to have a more heterogeneous expression of IgD than previously reported and therefore contribute to 3-15% of the MZ compartment. Searching for markers that would permit to discriminate between marginal zone and germinal center-derived IgM memory B cells is obviously required to further delineate their respective function. Disclosures No relevant conflicts of interest to declare.
20
achour, achouak, Xiaoping Sun, Thomas Nguyen, Annette Ko, Jeffrey Cifello, Chen Ling, Jay Sharma та ін. "Subset distinct richness reduction and clonal expansion of human CD4+ and CD8+ αβ TCR repertoires with aging". Journal of Immunology 206, № 1_Supplement (1 травня 2021): 98.16. http://dx.doi.org/10.4049/jimmunol.206.supp.98.16.
Повний текст джерелаАнотація:
Abstract A competent T-cell receptor (TCR) repertoire contains TCRs recognizing both novel and experienced antigens. It is widely believed that the human TCR repertoire declines with age, but its precise change in size and clonal distribution with age has not been fully determined. Here, we investigated how the size and content of TCR repertoire changes with age by analyzing the TCR repertoire of human CD4+ and CD8+ T cells and their naïve and memory subsets from 30 healthy adults aged from 25 to 85 at first visit and an average of 9-year follow-up as second visit by RNAseq. First, we calculated abTCR repertoire changes with age and found a profound reduction of the TCRa and TCRb richness in naïve CD8+ T cells and an increase of clonal expansion especially in memory CD8 T cells compared to CD4+ T cells. We also study the degree of overlap of TCRa/b repertoire between naïve and memory T cells in each subject at the same visit and found little overlap of TCRa/b unique sequences between naïve and memory CD4 T cells and ~40% of the total TCRs in CD8 T cells. Moreover, we showed elevated overlap of TCR sequences between two visits in both CD4 and CD8 T cells with memory CD8 T cells as the highest. Finally, we determined the degree of TCR clonotypes shares among different subjects via the analysis of the sharing of TCR sequences in the 30 healthy adults. As results, we observed increased sharing of identical TCR between CD4 and CD8 T cells suggesting that the public clonotypes may share with the common pathogens shared in the population, particularly in the old population. Our findings showed more profound reduction in the CD8+TCR repertoire with age compared to CD4+ TCR repertoire. All together provide evidence of ab TCR sequence-based age-associated changes of T cells and their subsets.
21
Nadiradze, Giorgi, Ilia Markov, Bapi Chatterjee, Vyacheslav Kungurtsev, and Dan Alistarh. "Elastic Consistency: A Practical Consistency Model for Distributed Stochastic Gradient Descent." Proceedings of the AAAI Conference on Artificial Intelligence 35, no. 10 (May 18, 2021): 9037–45. http://dx.doi.org/10.1609/aaai.v35i10.17092.
Повний текст джерелаАнотація:
One key element behind the recent progress of machine learning has been the ability to train machine learning models in large-scale distributed shared-memory and message-passing environments. Most of these models are trained employing variants of stochastic gradient descent (SGD) based optimization, but most methods involve some type of consistency relaxation relative to sequential SGD, to mitigate its large communication or synchronization costs at scale. In this paper, we introduce a general consistency condition covering communication-reduced and asynchronous distributed SGD implementations. Our framework, called elastic consistency, decouples the system-specific aspects of the implementation from the SGD convergence requirements, giving a general way to obtain convergence bounds for a wide variety of distributed SGD methods used in practice. Elastic consistency can be used to re-derive or improve several previous convergence bounds in message-passing and shared-memory settings, but also to analyze new models and distribution schemes. As a direct application, we propose and analyze a new synchronization-avoiding scheduling scheme for distributed SGD, and show that it can be used to efficiently train deep convolutional models for image classification.
22
Liu, Daizong, Xiaoye Qu, Xing Di, Yu Cheng, Zichuan Xu, and Pan Zhou. "Memory-Guided Semantic Learning Network for Temporal Sentence Grounding." Proceedings of the AAAI Conference on Artificial Intelligence 36, no. 2 (June 28, 2022): 1665–73. http://dx.doi.org/10.1609/aaai.v36i2.20058.
Повний текст джерелаАнотація:
Temporal sentence grounding (TSG) is crucial and fundamental for video understanding. Although existing methods train well-designed deep networks with large amount of data, we find that they can easily forget the rarely appeared cases during training due to the off-balance data distribution, which influences the model generalization and leads to unsatisfactory performance. To tackle this issue, we propose a memory-augmented network, called Memory-Guided Semantic Learning Network (MGSL-Net), that learns and memorizes the rarely appeared content in TSG task. Specifically, our proposed model consists of three main parts: cross-modal interaction module, memory augmentation module, and heterogeneous attention module. We first align the given video-query pair by a cross-modal graph convolutional network, and then utilize memory module to record the cross-modal shared semantic features in the domain-specific persistent memory. During training, the memory slots are dynamically associated with both common and rare cases, alleviating the forgetting issue. In testing, the rare cases can thus be enhanced by retrieving the stored memories, leading to better generalization. At last, the heterogeneous attention module is utilized to integrate the enhanced multi-modal features in both video and query domains. Experimental results on three benchmarks show the superiority of our method on both effectiveness and efficiency, which substantially improves the accuracy not only on the entire dataset but also on the rare cases.
23
Jan, Yahya, and Lech Jóźwiak. "Communication and Memory Architecture Design of Application-Specific High-End Multiprocessors." VLSI Design 2012 (March 25, 2012): 1–20. http://dx.doi.org/10.1155/2012/794753.
Повний текст джерелаАнотація:
This paper is devoted to the design of communication and memory architectures of massively parallel hardware multiprocessors necessary for the implementation of highly demanding applications. We demonstrated that for the massively parallel hardware multiprocessors the traditionally used flat communication architectures and multi-port memories do not scale well, and the memory and communication network influence on both the throughput and circuit area dominates the processors influence. To resolve the problems and ensure scalability, we proposed to design highly optimized application-specific hierarchical and/or partitioned communication and memory architectures through exploring and exploiting the regularity and hierarchy of the actual data flows of a given application. Furthermore, we proposed some data distribution and related data mapping schemes in the shared (global) partitioned memories with the aim to eliminate the memory access conflicts, as well as, to ensure that our communication design strategies will be applicable. We incorporated these architecture synthesis strategies into our quality-driven model-based multi-processor design method and related automated architecture exploration framework. Using this framework, we performed a large series of experiments that demonstrate many various important features of the synthesized memory and communication architectures. They also demonstrate that our method and related framework are able to efficiently synthesize well scalable memory and communication architectures even for the high-end multiprocessors. The gains as high as 12-times in performance and 25-times in area can be obtained when using the hierarchical communication networks instead of the flat networks. However, for the high parallelism levels only the partitioned approach ensures the scalability in performance.
24
Mei, Gang, Liangliang Xu, and Nengxiong Xu. "Accelerating adaptive inverse distance weighting interpolation algorithm on a graphics processing unit." Royal Society Open Science 4, no. 9 (September 2017): 170436. http://dx.doi.org/10.1098/rsos.170436.
Повний текст джерелаАнотація:
This paper focuses on designing and implementing parallel adaptive inverse distance weighting (AIDW) interpolation algorithms by using the graphics processing unit (GPU). The AIDW is an improved version of the standard IDW, which can adaptively determine the power parameter according to the data points’ spatial distribution pattern and achieve more accurate predictions than those predicted by IDW. In this paper, we first present two versions of the GPU-accelerated AIDW, i.e. the naive version without profiting from the shared memory and the tiled version taking advantage of the shared memory. We also implement the naive version and the tiled version using two data layouts, structure of arrays and array of aligned structures, on both single and double precision. We then evaluate the performance of parallel AIDW by comparing it with its corresponding serial algorithm on three different machines equipped with the GPUs GT730M, M5000 and K40c. The experimental results indicate that: (i) there is no significant difference in the computational efficiency when different data layouts are employed; (ii) the tiled version is always slightly faster than the naive version; and (iii) on single precision the achieved speed-up can be up to 763 (on the GPU M5000), while on double precision the obtained highest speed-up is 197 (on the GPU K40c). To benefit the community, all source code and testing data related to the presented parallel AIDW algorithm are publicly available.
25
Groves, C. J., J. Carrell, R. Grady, B. Rajan, C. A. Morehouse, R. Halpin, J. Wang, et al. "CD19-positive antibody-secreting cells provide immune memory." Blood Advances 2, no. 22 (November 26, 2018): 3163–76. http://dx.doi.org/10.1182/bloodadvances.2017015172.
Повний текст джерелаАнотація:
Abstract Long-lived antibody-secreting cells (ASCs) are critical for the maintenance of humoral immunity through the continued production of antibodies specific for previously encountered pathogen or vaccine antigens. Recent reports describing humoral immune memory have suggested the importance of long-lived CD19− bone marrow (BM) ASCs, which secrete antibodies recognizing previously encountered vaccine antigens. However, these reports do not agree upon the unique contribution of the CD19+ BM ASC subset toward humoral immunity. Here, we found both CD19+ and negative ASCs from human BM were similar in functional capacity to react to a number of vaccine antigens via ELISpot assays. The CD19+ cells were the predominant ASC population found in lymphoid tissues, and unlike the CD19− ASCs, which were found only in spleen and BM, the CD19+ ASCs were found in tonsil and blood. CD19+ ASCs from the BM, spleen, and tonsil were capable of recognizing polio vaccine antigens, indicating the CD19+ ASC cells play a novel role in long-lasting immune defense. Comparative gene expression analysis indicated CD19+ and negative BM ASCs differed significantly by only 14 distinct messenger RNAs and exhibited similar gene expression for cell cycle, autophagy, and apoptosis control necessary for long life. In addition, we show identical CDR-H3 sequences found on both BM ASC subsets, indicating a shared developmental path. Together, these results provide novel insight for the distribution, function, genetic regulation, and development of long-lived ASCs and may not only impact improved cell therapies but also enhance strategies for vaccine development.
26
Han, Hyemin. "A Novel Method to Use Coordinate Based Meta-Analysis to Determine a Prior Distribution for Voxelwise Bayesian Second-Level fMRI Analysis." Mathematics 10, no. 3 (January 24, 2022): 356. http://dx.doi.org/10.3390/math10030356.
Повний текст джерелаАнотація:
Previous research showed that employing results from meta-analyses of relevant previous fMRI studies can improve the performance of voxelwise Bayesian second-level fMRI analysis. In this process, prior distributions for Bayesian analysis can be determined by information acquired from the meta-analyses. However, only image-based meta-analysis, which is not widely accessible to fMRI researchers due to the lack of shared statistical images, was tested in the previous study, so the applicability of the prior determination method proposed by the previous study might be limited. In the present study, whether determining prior distributions based on coordinate-based meta-analysis, which is widely accessible to researchers, can also improve the performance of Bayesian analysis, was examined. Three different types of coordinate-based meta-analyses, BrainMap and Ginger ALE, and NeuroQuery, were tested as information sources for prior determination. Five different datasets addressing three task conditions, i.e., working memory, speech, and face processing, were analyzed via Bayesian analysis with a meta-analysis informed prior distribution, Bayesian analysis with a default Cauchy prior adjusted for multiple comparisons, and frequentist analysis with familywise error correction. The findings from the aforementioned analyses suggest that use of coordinate-based meta-analysis also significantly enhanced performance of Bayesian analysis as did image-based meta-analysis.
27
ZAWADZKI, PIOTR. "THE PING-PONG PROTOCOL WITH A PRIOR PRIVACY AMPLIFICATION." International Journal of Quantum Information 10, no. 03 (April 2012): 1250032. http://dx.doi.org/10.1142/s0219749912500323.
Повний текст джерелаАнотація:
A modified version of the well known ping-pong protocol is proposed and analyzed in the paper. The operation of the improved version exploits inevitable errors introduced by the eavesdropping in the message mode — a feature which has not been exploited so far. A proposed protocol improvement is in its role similar to the privacy amplification known from the quantum key distribution schemes, but no randomization of transferred information is introduced and deterministic character of communication is preserved. Messages are processed in blocks which guarantees that an eavesdropper is faced with a computationally infeasible problem as long as the system parameters are within reasonable limits. Information processing does not require quantum memory registers and confidential communication is possible without prior key agreement or some shared secret.
28
Likhoded, N. A. "Conditions for the existence of broadcast and spatial locality in computation threads." Proceedings of the National Academy of Sciences of Belarus. Physics and Mathematics Series 58, no. 3 (October 12, 2022): 292–99. http://dx.doi.org/10.29235/1561-2430-2022-58-3-292-299.
Повний текст джерелаАнотація:
Graphics Processing Units (GPUs) are considered as the target computer for implementing parallel algorithms. The set of algorithm operations to be implemented on the GPU must be split into computation threads; the threads should be grouped into computation blocks that are performed atomically on stream processors. Threads of a single block are executed on a stream processor in parts-pools called warp; warp threads are executed simultaneously. The efficiency of the parallel algorithm depends on the way the data is stored in the GPU memory. If all warp threads request the same datum when executing the current operator, then it is desirable to place it in a shared or constant GPU memory; in this case, its distribution across the cores of the multiprocessor is actually realized by means of broadcast. If warp threads request data located close to the memory, then in this case there is a spatial locality of data, which makes it advisable to place this data in the GPU’s memory. The implementation of broadcast or spatial locality by placing data in a memory of the appropriate type allows one to significantly reduce traffic when exchanging data between the memory levels of the GPU. This paper formulates and proves the necessary and sufficient conditions under which it is possible to perform a broadcast or there is a spatial locality of data. The conditions are formulated in terms of functions that determine the use of array elements at occurrences in the algorithm operators and functions that define the information dependencies of the algorithm. The results of the work can be used to optimize parallel algorithms when they are implemented on the GPU.
29
Langlois, Thomas A., Nori Jacoby, Jordan W. Suchow, and Thomas L. Griffiths. "Serial reproduction reveals the geometry of visuospatial representations." Proceedings of the National Academy of Sciences 118, no. 13 (March 26, 2021): e2012938118. http://dx.doi.org/10.1073/pnas.2012938118.
Повний текст джерелаАнотація:
An essential function of the human visual system is to locate objects in space and navigate the environment. Due to limited resources, the visual system achieves this by combining imperfect sensory information with a belief state about locations in a scene, resulting in systematic distortions and biases. These biases can be captured by a Bayesian model in which internal beliefs are expressed in a prior probability distribution over locations in a scene. We introduce a paradigm that enables us to measure these priors by iterating a simple memory task where the response of one participant becomes the stimulus for the next. This approach reveals an unprecedented richness and level of detail in these priors, suggesting a different way to think about biases in spatial memory. A prior distribution on locations in a visual scene can reflect the selective allocation of coding resources to different visual regions during encoding (“efficient encoding”). This selective allocation predicts that locations in the scene will be encoded with variable precision, in contrast to previous work that has assumed fixed encoding precision regardless of location. We demonstrate that perceptual biases covary with variations in discrimination accuracy, a finding that is aligned with simulations of our efficient encoding model but not the traditional fixed encoding view. This work demonstrates the promise of using nonparametric data-driven approaches that combine crowdsourcing with the careful curation of information transmission within social networks to reveal the hidden structure of shared visual representations.
30
Rosing, Matt, and Robert Schnabel. "Flexible Language Constructs for Large Parallel Programs." Scientific Programming 3, no. 3 (1994): 169–86. http://dx.doi.org/10.1155/1994/209864.
Повний текст джерелаАнотація:
The goal of the research described in this article is to develop flexible language constructs for writing large data parallel numerical programs for distributed memory (multiple instruction multiple data [MIMD]) multiprocessors. Previously, several models have been developed to support synchronization and communication. Models for global synchronization include single instruction multiple data (SIMD), single program multiple data (SPMD), and sequential programs annotated with data distribution statements. The two primary models for communication include implicit communication based on shared memory and explicit communication based on messages. None of these models by themselves seem sufficient to permit the natural and efficient expression of the variety of algorithms that occur in large scientific computations. In this article, we give an overview of a new language that combines many of these programming models in a clean manner. This is done in a modular fashion such that different models can be combined to support large programs. Within a module, the selection of a model depends on the algorithm and its efficiency requirements. In this article, we give an overview of the language and discuss some of the critical implementation details.
31
Markussen, Jonas, Lars Bjørlykke Kristiansen, Pål Halvorsen, Halvor Kielland-Gyrud, Håkon Kvale Stensland, and Carsten Griwodz. "SmartIO." ACM Transactions on Computer Systems 38, no. 1-2 (July 2021): 1–78. http://dx.doi.org/10.1145/3462545.
Повний текст джерелаАнотація:
The large variety of compute-heavy and data-driven applications accelerate the need for a distributed I/O solution that enables cost-effective scaling of resources between networked hosts. For example, in a cluster system, different machines may have various devices available at different times, but moving workloads to remote units over the network is often costly and introduces large overheads compared to accessing local resources. To facilitate I/O disaggregation and device sharing among hosts connected using Peripheral Component Interconnect Express (PCIe) non-transparent bridges, we present SmartIO. NVMes, GPUs, network adapters, or any other standard PCIe device may be borrowed and accessed directly, as if they were local to the remote machines. We provide capabilities beyond existing disaggregation solutions by combining traditional I/O with distributed shared-memory functionality, allowing devices to become part of the same global address space as cluster applications. Software is entirely removed from the data path, and simultaneous sharing of a device among application processes running on remote hosts is enabled. Our experimental results show that I/O devices can be shared with remote hosts, achieving native PCIe performance. Thus, compared to existing device distribution mechanisms, SmartIO provides more efficient, low-cost resource sharing, increasing the overall system performance.
32
Craston, Patrick, Brad Wyble, Srivas Chennu, and Howard Bowman. "The Attentional Blink Reveals Serial Working Memory Encoding: Evidence from Virtual and Human Event-related Potentials." Journal of Cognitive Neuroscience 21, no. 3 (March 2009): 550–66. http://dx.doi.org/10.1162/jocn.2009.21036.
Повний текст джерелаАнотація:
Observers often miss a second target (T2) if it follows an identified first target item (T1) within half a second in rapid serial visual presentation (RSVP), a finding termed the attentional blink. If two targets are presented in immediate succession, however, accuracy is excellent (Lag 1 sparing). The resource sharing hypothesis proposes a dynamic distribution of resources over a time span of up to 600 msec during the attentional blink. In contrast, the ST2 model argues that working memory encoding is serial during the attentional blink and that, due to joint consolidation, Lag 1 is the only case where resources are shared. Experiment 1 investigates the P3 ERP component evoked by targets in RSVP. The results suggest that, in this context, P3 amplitude is an indication of bottom–up strength rather than a measure of cognitive resource allocation. Experiment 2, employing a two-target paradigm, suggests that T1 consolidation is not affected by the presentation of T2 during the attentional blink. However, if targets are presented in immediate succession (Lag 1 sparing), they are jointly encoded into working memory. We use the ST2 model's neural network implementation, which replicates a range of behavioral results related to the attentional blink, to generate “virtual ERPs” by summing across activation traces. We compare virtual to human ERPs and show how the results suggest a serial nature of working memory encoding as implied by the ST2 model.
33
Kim, Hoisoo. ""Effects of Cooperative Learning, Graphic Organizers, and Shared Mental Models on Learners' Situational Model Construction, Distribution of Working Memory, and Metacognition in Web-based Hypertext"." Journal of Educational Technology 27, no. 2 (June 30, 2011): 289–315. http://dx.doi.org/10.17232/kset.27.2.289.
Повний текст джерела
34
Ma, Xinwei, Yurui Yin, Yuchuan Jin, Mingjia He, and Minqing Zhu. "Short-Term Prediction of Bike-Sharing Demand Using Multi-Source Data: A Spatial-Temporal Graph Attentional LSTM Approach." Applied Sciences 12, no. 3 (January 23, 2022): 1161. http://dx.doi.org/10.3390/app12031161.
Повний текст джерелаАнотація:
As a convenient, economical, and eco-friendly travel mode, bike-sharing greatly improved urban mobility. However, it is often very difficult to achieve a balanced utilization of shared bikes due to the asymmetric spatio-temporal user demand distribution and the insufficient numbers of shared bikes, docks, or parking areas. If we can predict the short-run bike-sharing demand, it will help operating agencies rebalance bike-sharing systems in a timely and efficient way. Compared to the statistical methods, deep learning methods can automatically learn the relationship between the inputs and outputs, requiring less assumptions and achieving higher accuracy. This study proposes a Spatial-Temporal Graph Attentional Long Short-Term Memory (STGA-LSTM) neural network framework to predict short-run bike-sharing demand at a station level using multi-source data sets. These data sets include historical bike-sharing trip data, historical weather data, users’ personal information, and land-use data. The proposed model can extract spatio-temporal information of bike-sharing systems and predict the short-term bike-sharing rental and return demand. We use a Graph Convolutional Network (GCN) to mine spatial information and adopt a Long Short-Term Memory (LSTM) network to mine temporal information. The attention mechanism is focused on both temporal and spatial dimensions to enhance the ability of learning temporal information in LSTM and spatial information in GCN. Results indicate that the proposed model is the most accurate compared with several baseline models, the attention mechanism can help improve the model performance, and models that include exogenous variables perform better than the models that only consider historical trip data. The proposed short-term prediction model can be used to help bike-sharing users better choose routes and to help operators implement dynamic redistribution strategies.
35
Jensen, Stig Rune, Jonas Jusélius, Antoine Durdek, Tor Flå, Peter Wind, and Luca Frediani. "Linear scaling Coulomb interaction in the multiwavelet basis, a parallel implementation." International Journal of Modeling, Simulation, and Scientific Computing 05, supp01 (October 2014): 1441003. http://dx.doi.org/10.1142/s1793962314410037.
Повний текст джерелаАнотація:
We present a parallel and linear scaling implementation of the calculation of the electrostatic potential arising from an arbitrary charge distribution. Our approach is making use of the multi-resolution basis of multiwavelets. The potential is obtained as the direct solution of the Poisson equation in its Green's function integral form. In the multiwavelet basis, the formally non local integral operator decays rapidly to negligible values away from the main diagonal, yielding an effectively banded structure where the bandwidth is only dictated by the requested accuracy. This sparse operator structure has been exploited to achieve linear scaling and parallel algorithms. Parallelization has been achieved both through the shared memory (OpenMP) and the message passing interface (MPI) paradigm. Our implementation has been tested by computing the electrostatic potential of the electronic density of long-chain alkanes and diamond fragments showing (sub)linear scaling with the system size and efficent parallelization.
36
Canosa-Reyes, Rewer M., Andrei Tchernykh, Jorge M. Cortés-Mendoza, Bernardo Pulido-Gaytan, Raúl Rivera-Rodriguez, Jose E. Lozano-Rizk, Eduardo R. Concepción-Morales, et al. "Dynamic performance–Energy tradeoff consolidation with contention-aware resource provisioning in containerized clouds." PLOS ONE 17, no. 1 (January 20, 2022): e0261856. http://dx.doi.org/10.1371/journal.pone.0261856.
Повний текст джерелаАнотація:
Containers have emerged as a more portable and efficient solution than virtual machines for cloud infrastructure providing both a flexible way to build and deploy applications. The quality of service, security, performance, energy consumption, among others, are essential aspects of their deployment, management, and orchestration. Inappropriate resource allocation can lead to resource contention, entailing reduced performance, poor energy efficiency, and other potentially damaging effects. In this paper, we present a set of online job allocation strategies to optimize quality of service, energy savings, and completion time, considering contention for shared on-chip resources. We consider the job allocation as the multilevel dynamic bin-packing problem that provides a lightweight runtime solution that minimizes contention and energy consumption while maximizing utilization. The proposed strategies are based on two and three levels of scheduling policies with container selection, capacity distribution, and contention-aware allocation. The energy model considers joint execution of applications of different types on shared resources generalized by the job concentration paradigm. We provide an experimental analysis of eighty-six scheduling heuristics with scientific workloads of memory and CPU-intensive jobs. The proposed techniques outperform classical solutions in terms of quality of service, energy savings, and completion time by 21.73–43.44%, 44.06–92.11%, and 16.38–24.17%, respectively, leading to a cost-efficient resource allocation for cloud infrastructures.
37
Yang, Ning, Shiaaulir Wang, and Paul Schonfeld. "Simulation-Based Scheduling of Waterway Projects Using a Parallel Genetic Algorithm." International Journal of Operations Research and Information Systems 6, no. 1 (January 2015): 49–63. http://dx.doi.org/10.4018/ijoris.2015010104.
Повний текст джерелаАнотація:
A Parallel Genetic Algorithm (PGA) is used for a simulation-based optimization of waterway project schedules. This PGA is designed to distribute a Genetic Algorithm application over multiple processors in order to speed up the solution search procedure for a very large combinational problem. The proposed PGA is based on a global parallel model, which is also called a master-slave model. A Message-Passing Interface (MPI) is used in developing the parallel computing program. A case study is presented, whose results show how the adaption of a simulation-based optimization algorithm to parallel computing can greatly reduce computation time. Additional techniques which are found to further improve the PGA performance include: (1) choosing an appropriate task distribution method, (2) distributing simulation replications instead of different solutions, (3) avoiding the simulation of duplicate solutions, (4) avoiding running multiple simulations simultaneously in shared-memory processors, and (5) avoiding using multiple processors which belong to different clusters (physical sub-networks).
38
Suarez-Ramirez, Jenny E., Karthik Chandiran, and Linda S. Cauley. "Localization of terminal memory CD8 T cells in the lungs during respiratory infection." Journal of Immunology 208, no. 1_Supplement (May 1, 2022): 182.35. http://dx.doi.org/10.4049/jimmunol.208.supp.182.35.
Повний текст джерелаАнотація:
Abstract Cytotoxic CD8 T cells (CTLs) play an important role in recovery from respiratory virus infection and provide protective immunity by targeting epitopes in well-conserved viral proteins that are shared by different strains of the virus. To provide immunity, CTLs migrate to different anatomical compartments in the lungs and destroy infected cells by injecting lytic molecules into the cytoplasm. Several adhesion molecules aid migration to the infected tissues and cytolysis. These molecules include killer lectin like receptor G1 (KLRG1), which is a membrane bound adhesion molecule that is expressed on terminally differentiated CD8 T effector (Teff) cells. Recently, several studies identified some long lived “terminal memory CD8 T cells (TTM)” that express KLRG1. The goal of our study is to determine the location of these TTM cells within the different compartments of the lungs during a respiratory infection. Our previous work showed that SMAD4 regulates several homing receptors on activated CTLs including KLRG1. We used transfer experiments to examine the distribution of WT cells that express SMAD4 and mutant CTLs that lack SMAD4 in the lungs. At 30 days after intranasal infection with Listeria monocytogenes, there were larger numbers of mutant CTLs in the airways (BAL fluid), while WT CTLs were primarily located in the lung tissue. Interestingly, the localization of the WT cells correlated with the expression of KLRG1. We are using in situ imaging to visualize these transferred CTLs during respiratory infection to better understand their role in protective immunity. Intravascular staining will be used to determine whether these CTLs remain in the blood vessels or enter the tissues of the infected recipient mice. This work was supported by NIH Grant AI123864.
39
Zhang, Fan, Mei Tu, and Jinyao Yan. "Accelerating Neural Machine Translation with Partial Word Embedding Compression." Proceedings of the AAAI Conference on Artificial Intelligence 35, no. 16 (May 18, 2021): 14356–64. http://dx.doi.org/10.1609/aaai.v35i16.17688.
Повний текст джерелаАнотація:
Large model size and high computational complexity prevent the neural machine translation (NMT) models from being deployed to low resource devices (e.g. mobile phones). Due to the large vocabulary, a large storage memory is required for the word embedding matrix in NMT models, in the meantime, high latency is introduced when constructing the word probability distribution. Based on reusing the word embedding matrix in the softmax layer, it is possible to handle the two problems brought by large vocabulary at the same time. In this paper, we propose Partial Vector Quantization (P-VQ) for NMT models, which can both compress the word embedding matrix and accelerate word probability prediction in the softmax layer. With P-VQ, the word embedding matrix is split into two low dimensional matrices, namely the shared part and the exclusive part. We compress the shared part by vector quantization and leave the exclusive part unchanged to maintain the uniqueness of each word. For acceleration, in the softmax layer, we replace most of the multiplication operations with the efficient looking-up operations based on our compression to reduce the computational complexity. Furthermore, we adopt curriculum learning and compact the word embedding matrix gradually to improve the compression quality. Experimental results on the Chinese-to-English translation task show that our method can reduce 74.35% of parameters of the word embedding and 74.42% of the FLOPs of the softmax layer. Meanwhile, the average BLEU score on the WMT test sets only drops 0.04.
40
Yang, Sa, Suoping Li, Nana Yang, and Ying Lin. "Modelling Analysis of a Novel Frameless Slotted-ALOHA Protocol Based on the Number of Detectable Conflicting Users." Future Internet 14, no. 10 (September 28, 2022): 279. http://dx.doi.org/10.3390/fi14100279.
Повний текст джерелаАнотація:
To solve the conflict when multi-user packets are transmitted in a shared wireless link, a novel frameless slotted-ALOHA protocol is proposed. Signature codes are used to help the receiver identify the set of transmitting users, and successive interference cancellation technology is employed to recover conflicting packets. Thus, the information in the conflicting slot can be reused to reduce the number of retransmissions. Taking the number of backlogged users in each slot as a system state, a Markov chain model is established to analyze the protocol, in which the state transition probabilities are obtained based on the binomial distribution of packets sent in a slot. Under the maximum number of detectable conflicting users, the best value is taken, traffic balance equations are obtained, and the expressions of throughput, average number of backlogged users, average successful transmission probability and average memory size are derived. Finally, a numerical simulation is carried out to accurately analyze the influence of the first transmission probability of the packets on various performance indexes and the effectiveness of the theoretical analysis is further verified by the simulation results.
41
Wu, Chuanfeng, Ryland D. Mortlock, Taehoon Shin, Stefan Cordes, Xing Fan, Jason Brenchley, David A. Allan, So Gun Hong, and Cynthia E. Dunbar. "Tissue-Resident Clonal Expansions of Rhesus Macaque NK Cells." Blood 138, Supplement 1 (November 5, 2021): 998. http://dx.doi.org/10.1182/blood-2021-146220.
Повний текст джерелаАнотація:
Abstract Recent phenotypic, functional and transcriptomic analyses of natural killer (NK) cells in human and animals have established the presence of tissue resident NK(trNK) cells with specific characteristics and a central role in NK memory. The lack of endogenous clonal markers on NK cells impedes understanding the clonal genesis of trNKs. Transplantation of lentivirally-barcoded autologous hematopoietic stem and progenitor cells (HSPCs) has allowed tracking of NK cells at a clonal level in rhesus macaques (RM). We reported large KIR-restricted clonal expansions of mature CD56 -CD16 +NK in the peripheral blood (PB) of RM, clonally distinct from myeloid, T, B and CD56 +16 -NK, persisting for months to years, suggesting self-renewal independent of ongoing production from HSPCs (Wu et al, Cell Stem Cell, 2014 and Science Imm, 2018). We have now used this model to investigate the clonal distribution of NK cell populations in bone marrow (BM), liver, spleen, lymph nodes (LN), jejunum, colon and bronchoalveolar lavage (BAL). Serial of tissue biopsies were obtained from 3 barcoded monkeys by endoscopy and laparotomy overtime at steady state post transplantation , as well as tissues from necropsy. We compared clonal patterns between various trNKs and PB NKs collected from barcoded RM. Tissue or PB NK were defined as CD3-CD14-CD20-NKG2+ (NKG2A+ and NKG2C+), and NK subsets were further sorted for CD16, CD56, CD49a (putative liver tissue memory-like NK marker), and CXCR3 (critical for NK cell migration into tumor or normal tissues). The same expanded CD56 -CD16 +NK clones found in the PB were also detected at high abundance within BM, LN, liver and/or spleen CD56 -CD16 +NK, but not found in tissue CD56 +16 -NK and CD56 -16 -NK subsets. The liver and spleen bulk NK clonal patterns were highly correlated, and distinct from other tissues. We also observed tissue specific barcoded NK clones in BAL, jejunum and colon samples with no or very low abundance in other tissues and PB. Strikingly, a group of markedly expanded trNK clones, distinct from the expanded CD56 -CD16 +NK clones present concurrently or previously in PB, were present and shared across all tissues examined. These clones were enriched in CD56 -16 + trNK and absent in CD56 +16 - trNK. Notably, in both tissue and PB these clones were expanded in NKG2+ CD56-CD16- NK. These common expanded trNK cells were specifically enriched in both tissue and PB CD56 -16 -CXCR3 +NK, suggesting a role for this chemokine receptor and the ability of these clones to move between tissues. In contrast, CD49a expression did not enrich for these expanded clones. Clonally-expanded and persistent mature trNK cells, shared across multiple tissues but not present within PB mature CD56 -CD16 +NK subsets, combined with prior functional data suggesting NK memory is restricted to liver or other trNK cells, suggests these clonally-expanded trNK cells may be of interest. The pattern of shared clones across tissues, together with identification of a rare PB CD56-CD16-NK subpopulation harboring the clones, suggests preferential hematogenous homing of these clones to multiple tissues. Further analyses of gene expression and clonal dynamics are ongoing and should shed light on the ontology of trNK cells, with implications for the development of NK based immunotherapies and NK memory. Disclosures No relevant conflicts of interest to declare.
42
Al-Turany, Mohammad, Alexey Rybalchenko, Dennis Klein, Matthias Kretz, Dmytro Kresan, Radoslaw Karabowicz, Andrey Lebedev, Anar Manafov, Thorsten Kollegger, and Florian Uhlig. "ALFA: A framework for building distributed applications." EPJ Web of Conferences 245 (2020): 05021. http://dx.doi.org/10.1051/epjconf/202024505021.
Повний текст джерелаАнотація:
The ALFA framework is a joint development between ALICE Online-Offline and FairRoot teams. ALFA has a distributed architecture, i.e. a collection of highly maintainable, testable, loosely coupled, independently deployable processes. ALFA allows the developer to focus on building singlefunction modules with well-defined interfaces and operations. The communication between the independent processes is handled by FairMQ transport layer. FairMQ offers multiple implementations of its abstract data transport interface, it integrates some popular data transport technologies like ZeroMQ and nanomsg. Furthermore it also provides shared memory and RDMA transport (based on libfabric) for high throughput, low latency applications. Moreover, FairMQ allows the single process to use multiple and different transports at the same time. FairMQ based processes can be controlled and orchestrated via different systems by implementing the corresponding plugin. However, ALFA delivers also the Dynamic Deployment System (DDS) as an independent set of utilities and interfaces, providing a dynamic distribution of different user processes on any Resource Management System (RMS) or a laptop. ALFA is already being tested and used by different experiments in different stages of data processing as it offers an easy integration of heterogeneous hardware and software.
43
Ravankar, Ankit, Abhijeet Ravankar, Takanori Emaru, and Yukinori Kobayashi. "Multi-Robot Mapping and Navigation Using Topological Features." Proceedings 42, no. 1 (November 14, 2019): 68. http://dx.doi.org/10.3390/ecsa-6-06580.
Повний текст джерелаАнотація:
Robot mapping and exploration is basic to many robotic applications such as search and rescue operations in disaster scenarios, warehouse management, service robotics, patrolling and autonomous driving. With recent advances in robot navigation and sensor compactness, single robot systems can accurately model the environment and perform complex autonomous navigation tasks. On the other hand, multi-robot systems can speed up mapping and exploration tasks in emergency situations, such as rescue missions, by making use of distributed sensors, thereby increasing the range of exploration tasks to an extent that is not possible with a single robot. Each robot explores and maps different areas of the same environment that are finally merged and connected to make a global map. To build a map of an unknown environment, each robot must perform SLAM, or Simultaneous Localization and Mapping. A big challenge with a multi-robot SLAM system is the transfer of shared map information between multiple robots. There is a possibility of transferring individual measurement errors to the global map, resulting in excess computation and memory required to store such maps. To overcome this problem, we propose to use topological feature map representation that can store information into nodes and edges and does not have any large memory requirements. We present a combined metric-topological mapping approach to multi-robot SLAM. This method maintains a topological pose graph with sensor information stored in nodes and edges that can be optimized globally with reduced information sharing. By combining local metric and topological maps built by individual robots, the reduced graph structure can be merged and extended to map large areas effectively. To robustly merge local maps into global one, we used visual features from each robot that are matched in a distributed system. The graph node-edge structure is used for path planning and navigation. At the same time information sharing between robots results in optimized task distribution between multi-robots.
44
Yang, Cheng, and Guanming Lu. "Deeply Recursive Low- and High-Frequency Fusing Networks for Single Image Super-Resolution." Sensors 20, no. 24 (December 18, 2020): 7268. http://dx.doi.org/10.3390/s20247268.
Повний текст джерелаАнотація:
With the development of researches on single image super-resolution (SISR) based on convolutional neural networks (CNN), the quality of recovered images has been remarkably promoted. Since then, many deep learning-based models have been proposed, which have outperformed the traditional SISR algorithms. According to the results of extensive experiments, the feature representations of the model can be enhanced by increasing the depth and width of the network, which can ultimately improve the image reconstruction quality. However, a larger network generally consumes more computational and memory resources, making it difficult to train the network and increasing the prediction time. In view of the above problems, a novel deeply-recursive low- and high-frequency fusing network (DRFFN) for SISR tasks is proposed in this paper, which adopts the structure of parallel branches to extract the low- and high-frequency information of the image, respectively. The different complexities of the branches can reflect the frequency characteristic of the diverse image information. Moreover, an effective channel-wise attention mechanism based on variance (VCA) is designed to make the information distribution of each feature map more reasonably with different variances. Owing to model structure (i.e., cascading recursive learning of recursive units), DRFFN and DRFFN-L are very compact, where the weights are shared by all convolutional recursions. Comprehensive benchmark evaluations in standard benchmark datasets well demonstrate that DRFFN outperforms the most existing models and has achieved competitive, quantitative, and visual results.
45
Schonnesen, Alexandra A., Patrik Parker, Ke-Yue Ma, Jennifer McKinney, Declan Flemming, Matthew Fox, Gail Eckhardt, and Ning Jiang. "Multi-dimensional analysis of tumor-specific and bystander CD8+ tumor infiltrating lymphocytes with TetTCR-SeqHD." Journal of Immunology 204, no. 1_Supplement (May 1, 2020): 243.8. http://dx.doi.org/10.4049/jimmunol.204.supp.243.8.
Повний текст джерелаАнотація:
Abstract Objective Antigen specific CD8+ T cells were isolated from primary tumor tissue and analyzed for on the single cell level to determine tumor-specific signatures for further study. Methods Primary tumor, metastatic lesions, adjacent healthy tissue and peripheral blood were from collected from patients at Dell Seton Medical Center at the University of Texas. Matched tumor and healthy samples underwent exome and RNA sequencing to allow prediction of tumor-specific epitopes. We then synthesized DNA-barcoded tetramers for all predicted tumor associated-epitopes and an HLA-matched library of exogenous epitopes. Tetramer stained CD8+ T cells were then sorted and analyzed by TetTCR-SeqHD, a technology that we recently developed to link TCR antigen specificity, TCR sequence, gene expression, and phenotype of single T cells. Results We analyzed ~2900 tumor infiltrating lymphocytes (TILs) and T cells from adjacent healthy tissue and observed enrichment for cytotoxic effector and tissue resident memory cells within the tumor versus the healthy tissues. We also note an enrichment of bystander cells, specifically those reactive to Epstein Barr Virus epitopes, within the cytotoxic cluster. Finally, we analyze the distribution and differential expression between tumor-specific lymphocytes and bystander T cells, noting reduced expression of CD57 and CD49d in the tumor-specific TILs, as previously described. Conclusions We validate the use of TetTCR-SeqHD for gaining insight into phenotypic and gene expression differences between bystander and tumor-specific TILs. We identified markers for future exploration and demonstrate key differences between shared clonal populations within healthy and cancerous tissues.
46
Giudice, Valentina, Xingmin Feng, Lin Zenghua, Maria del Pilar Fernandez Ibanez, Keyvan Keyvanfar, Carmine Selleri, and Neal S. Young. "Oligoclonal Expansion of Effector Memory CD8+CD57+ T Cells May Sustain Bone Marrow Destruction in Aplastic Anemia." Blood 128, no. 22 (December 2, 2016): 3898. http://dx.doi.org/10.1182/blood.v128.22.3898.3898.
Повний текст джерелаАнотація:
Abstract The character of oligoclonal expansion of CD8+CD28- lymphocytes in aplastic anemia (AA), described by Risitano et al. (Blood, 2002 and Lancet, 2004), strongly suggests an antigen driving mechanism of T cell activation leading to destruction of hematopoietic stem and progenitor cells. In this study, we focused on a subset of CD8+CD28- T lymphocytes termed effector memory cells because of their high antigen-affinity and ability to be rapid activated after antigen stimulation. We investigated the frequency and oligoclonal expansion of effector CD28-CD57+ memory cells in CD4+ and CD8+ T subsets and the T cell receptor (TCR) Vβ repertoire by flow cytometry, and next-generation sequencing (NGS). Peripheral blood mononuclear cells from 20 AA patients and 14 healthy controls were evaluated for Vβ usage by flow cytometry. At the time of sampling, none of the patients had yet received immunosuppressive therapy. CD28+CD57- and CD28-CD57+ cells in CD4+ and CD8+ populations were assessed using the IOTest Beta Mark and a LSRII Fortessa cytometer for acquisition. The mean + 3SD of each Vβ group in controls was used as a cut-off to determine Vβ skewing in patients. When the frequencies of CD28+ and CD57+ cells in CD4+ and CD8+ subsets were compared, no differences were found between patients and controls (p=0.6125). Polyclonal expansion was described in CD8+CD28+, CD4+CD28+ and CD4+CD57+ T cells in 60%, 45% and 70% of patients, respectively. Using the mean of CD8+CD57+ cell frequency in controls as a cut-off, 12 of 20 patients displayed expansion of effector memory CD8+ cells and 11 of them (55% of all patients) showed expansion in 1 to 6 Vβ families with 1 to 4 immunodominant clones, while only 4 of 14 (29%) healthy controls displayed expansion of two Vβ families in expanded CD8+CD57+ cells without polyclonal expansion of CD8+CD28+ cells. In four cases (20% of all cases) of the remaining 8 patients whose cells did not display expanded CD8+CD57+ cells, there was expansion of 1 to 3 Vβ families with 1 dominant clone. These findings suggest activation of the immune response, polyclonal expansion of effector CD28+ compartments, and oligoclonal activation of effector memory CD8+ cells, regardless of frequency in peripheral blood. The frequency of CD8+CD57+ cells modestly correlated with the mean Vβ expansion in each patient (r2=0.5831, p<0.001). Also, we analyzed VDJ combinations and complementary region 3 (CDR3) sequences in CD4+ and CD8+ cells from 8 AA patients with CD8+CD57+ expansion in the ILLumina Hiseq 2000 sequencer. For the CD4+ compartment, patients were similar to healthy control CD4+ cells: frequencies of the most abundant clones less than 6%, CDR3 size profiles with Gaussian distribution and low degree of diversity (Simpson's indexes range 0-1: 0 means infinite diversity, and 1 means no diversity). For CD8+ cells, 6 of 8 patients displayed a "skyscraper" Vβ/Jβ plot due to the presence of 1 to 3 dominant clones, with frequency greater than 10%, predominant classes in CDR3 size profile, and Simpson's index similar to the CD8+CD57+ subset (p=0.1914). Analyzing CDR3 amino acid sequences for homology, we found that all the dominant sequences from patients were present at very low frequency in a healthy donor CD8+ cell pool (range, 0.12% - 53.6% vs 0.01% - 1.64%, p=0.0037), but no matches were described for the healthy CD8+CD57+ CDR3 repertoire. Only three CDR3 sequences were shared between patients. One of these CDR3 sequences (CSARDPPVSGTRGTDTQYF) was present in all patients and controls at very low frequency (mean expression, 5.01%), and it was the dominant clone in one patient, carried by TRBV20-1/TRBJ2-3 rearrangement at a frequency of 53.6%. Shared sequences were not found in reported viral TCR repertoires, nor in T cells recognizing peptide of myelin basic protein in multiple sclerosis, or synovial T cells from rheumatoid arthritis patients, nor matches were identified in T-large granular lymphocyte leukemia, AA, cancer or melanoma-derived TCR repertoires. Our data support the hypothesis of autologous immune-mediated attack leading to bone marrow destruction, likely triggered by autoantigens. Our results highlight the role of effector memory CD8+CD57+ T cells in sustaining an aberrant immune response during active disease. CD8+CD57+ cell expansion mirrors Vβ oligoclonal expansion, and Vβ flow cytometry should be useful in diagnosis and periodic evaluation of AA patients. Disclosures Fernandez Ibanez: GSK/Novartis: Research Funding. Young:Novartis: Research Funding.
47
., Vikas, M. Sai Vivek, and Pushyami Manohar. "A load balancing model using bio inspired firefly algorithm in cloud computing." International Journal of Engineering & Technology 7, no. 1.1 (December 21, 2017): 671. http://dx.doi.org/10.14419/ijet.v7i1.1.10825.
Повний текст джерелаАнотація:
Cloud computing is a model which helps in accessing the end-users interface which is adjustable as well as flexible service on Internet. The primary aim implemented scheme is for maximizing utilization of resource. It also offers a load which is good balanced compared to the all other resources in the servers of cloud. Some fundamental characteristics like usage of memory, process time, rate of the access, a resource load model can be deliver, by considering the load index which is new, present load computed to the virtual machine cloud server shared resources. If a load index calculated to the resources then initiation of balancing load for efficient utilization of resources. After all resources load index computed, the load balancing operation started to use the resources effectively by assigned resources processing for respective node for decreasing values of the load. The assignment of the resources to the nodes which are proper will become a problem i.e. optimal distribution. For this huge number of optimised schemes like algorithm of genetic, algorithm of modified genetic will be used to balance the load. But all these are may not be effective to offer best solutions. Because, the exploration issues are not overcome in these approaches. However, we can say that procedure of efficient optimisation is good for balancing load compared to others. Therefore, we implemented a new scheme for optimisation approach which is known as firefly algorithm for balancing the load. Initially the index table updated with the help of available virtual servers as well as request sequence. After that, based on new formulae, the load index will be calculated. The bal-ance of the load can be done with the help of load index using the algorithm of firefly. The results which are expected can be gathered and the implemented algorithm is efficient to balance the load in optimised time intervals.
48
Chan, Kuen Cheong, and Li Min Zhou. "Static Behaviours of Carbon Fibre Composite Strip with Bifurcated Type Shape Memory Alloy Pins." Key Engineering Materials 334-335 (March 2007): 1153–56. http://dx.doi.org/10.4028/www.scientific.net/kem.334-335.1153.
Повний текст джерелаАнотація:
A numerical study of the static behaviours of composite strip with bifurcated type shape memory alloy pins has been conducted. The case of bifurcated type shape memory alloy pins inserted inside the composite strip around the hole to reinforce the laminate, which was subjected to the axial stress was simulated. The models for stress analysis were established by using ANSYS finite element programme. Two types of shape memory alloy pins were proposed to insert along the through thickness direction of the carbon fibre woven fabric composite strip to induce the clamping force. The pre-tensioned load was applied to the shape memory alloy pins in order to reduce occurrence of delamination in the laminate. Three-dimensional elements and contact elements were used to simulate the contact between the composite laminate and shape memory alloy pin to investigate the stress distribution around the hole in the composite strip. The effect of pre-strain of shape memory alloy on the stresses inside composite was studied. The results show that the stress characteristics of the button-shaped and bifurcated shape memory alloy pin models are similar; however, the stresses for the button-shaped pin model are lower. The tensile and compressive stresses, both in button-shaped and bifurcated pin models, are strongly dependent on the percentage of pre-strain of the shape memory alloy. It is therefore concluded that the shape memory alloy pin method was significantly reduced the stress concentration of the composite strip laminate.
49
Xu, Hai-Chuan, Wei Zhang, Xiong Xiong, and Wei-Xing Zhou. "Wealth Share Analysis with “Fundamentalist/Chartist” Heterogeneous Agents." Abstract and Applied Analysis 2014 (2014): 1–11. http://dx.doi.org/10.1155/2014/328498.
Повний текст джерелаАнотація:
We build a multiassets heterogeneous agents model with fundamentalists and chartists, who make investment decisions by maximizing the constant relative risk aversion utility function. We verify that the model can reproduce the main stylized facts in real markets, such as fat-tailed return distribution and long-term memory in volatility. Based on the calibrated model, we study the impacts of the key strategies’ parameters on investors’ wealth shares. We find that, as chartists’ exponential moving average periods increase, their wealth shares also show an increasing trend. This means that higher memory length can help to improve their wealth shares. This effect saturates when the exponential moving average periods are sufficiently long. On the other hand, the mean reversion parameter has no obvious impacts on wealth shares of either type of traders. It suggests that no matter whether fundamentalists take moderate strategy or aggressive strategy on the mistake of stock prices, it will have no different impact on their wealth shares in the long run.
50
Hotta, Seiji, Senya Kiyasu, and Sueharu Miyahara. "Arbitrary-Shaped Cluster Separation Using One-Dimensional Data Mapping and Histogram Segmentation." Journal of Advanced Computational Intelligence and Intelligent Informatics 11, no. 9 (November 20, 2007): 1136–43. http://dx.doi.org/10.20965/jaciii.2007.p1136.
Повний текст джерелаАнотація:
Of the many clustering methods proposed for separating arbitrarily shaped clusters, most had drawbacks in parameter sensitivity and high-computational cost requiring large amounts of memory. We propose one-dimensional (1D) mapping for separating arbitrarily shaped clusters using a list of neighbors. After mapping, we apply a discriminant threshold selection to the histogram of the data distribution in 1D space. We verified the feasibility of performance in experiments on synthetic toy data, image, and video segmentation.