Filters








248,270 Hits in 6.9 sec

Harp: Collective Communication on Hadoop

Bingjing Zhang, Yang Ruan, Judy Qiu
2015 2015 IEEE International Conference on Cloud Engineering  
But if we use allgather bucket algorithm [14] , it is reduced to kdβ. In iterative algorithms, communication participated in by all the workers happens once or more per iteration.  ...  By plugging Harp into Hadoop, we convert MapReduce model to Map-Collective model and enable efficient in-memory communication between map tasks across a variety of important data analysis applications.  ...  Another model used for iterative computation is the Graph model, which abstracts data as vertices and edges.  ... 
doi:10.1109/ic2e.2015.35 dblp:conf/ic2e/ZhangRQ15 fatcat:wqkvtszvbbg3nnh27eks7h3plq

Efficient initialization of Mixtures of Experts for human pose estimation

Huazhong Ning, Yuxiao Hu, Thomas Huang
2008 2008 15th IEEE International Conference on Image Processing  
However, the EM algorithm that learns the BME model may converge to a suboptimal local maximum. And the quality of the final solution depends largely on the initial values.  ...  We first partition the training set so that each subset can be well modeled by a single expert and the total regression error is minimized.  ...  its gate from a partition subset, and use this estimation as starting values of EM iteration.  ... 
doi:10.1109/icip.2008.4712217 dblp:conf/icip/NingHH08 fatcat:tyouhyrw3bg3zg42klb5ltipdu

Scaling iterative graph computations with GraphMap

Kisung Lee, Ling Liu, Karsten Schwan, Calton Pu, Qi Zhang, Yang Zhou, Emre Yigitoglu, Pingpeng Yuan
2015 Proceedings of the International Conference for High Performance Computing, Networking, Storage and Analysis on - SC '15  
that improve computational efficiency.  ...  random access. (2) It entails a two-level graph partitioning algorithm that enables balanced workloads and locality-optimized data placement. (3) It contains a proposed suite of locality-based optimizations  ...  By combining it with efficient processing at each local node of a compute cluster, one can perform iterative graph computations more efficiently than the distributed memory-based model in many cases.  ... 
doi:10.1145/2807591.2807604 dblp:conf/sc/LeeLSPZZYY15 fatcat:sxwjv6iq6beexlz5tz2qb5od5e

GPOP: A cache- and work-efficient framework for Graph Processing Over Partitions [article]

Kartik Lakhotia, Sourav Pati, Rajgopal Kannan, Viktor Prasanna
2019 arXiv   pre-print
Its built-in analytical performance model enables it to use a hybrid of source and partition centric communication modes in a way that ensures work-efficiency each iteration while simultaneously boosting  ...  It completely abstracts away underlying programming model details from the user and provides an easy to program set of APIs with the ability to selectively continue the active vertex set across iterations  ...  The performance of a graph analytics framework depends on the cache efficiency, DRAM communication, use of synchronization primitives and theoretical efficiency of its underlying programming model.  ... 
arXiv:1806.08092v2 fatcat:gfpry5mg2nfnxhiaxco6lkmkfa

P3VI

David Wingate, Kevin D. Seppi
2004 Twenty-first international conference on Machine learning - ICML '04  
We present an examination of the stateof-the-art for using value iteration to solve large-scale discrete Markov Decision Processes.  ...  We give special attention to parallelization issues, discussing how to efficiently partition states, distribute partitions to processors, minimize message passing and ensure high scalability.  ...  Value iteration is also used as part of larger algorithms: RTDP [2] performs some value iteration off-line between executing controls, and Modified Policy Iteration [9] performs some value iteration  ... 
doi:10.1145/1015330.1015440 dblp:conf/icml/WingateS04 fatcat:k5lpm255gfey7dtygxk7teo4ty

Design optimization of stochastic complex systems via iterative density estimation [article]

Wang-Sheng Liu, Sai Hung Cheung
2020 arXiv   pre-print
Unlike traditional density estimation schemes, where the esti-mation is conducted in the entire design space, in the proposed method we iteratively partition the design space into several subspaces according  ...  In this contribution, we propose an efficient method which ap-proximates the failure probability functions (FPF) to decouple reliability.  ...  The efficiency of the proposed method can be measured by the total number of model performance evaluations.  ... 
arXiv:2003.00167v1 fatcat:raxgrtowdbgltnghhirpt6qb4m

Longitudinal partitioning based waveform relaxation algorithm for transient analysis of long delay transmission lines

Sourajeet Roy, Anestis Dounavis
2011 2011 IEEE MTT-S International Microwave Symposium  
algorithms based on longitudinal partitioning using the conventional lumped model.  ...  In this paper a waveform relaxation algorithm based on longitudinal partitioning is presented to efficiently model large distributed networks.  ...  Moreover since a delay extraction based longitudinal partitioning is used, the number of subcircuits required to model the line is much smaller than that required using the conventional lumped model.  ... 
doi:10.1109/mwsym.2011.5972822 fatcat:rd3fqqqjnrecfhrv7wtkrm4tcy

Longitudinal partitioning based waveform relaxation algorithm for transient analysis of long delay transmission lines

S. Roy, A. Dounavis
2011 2011 IEEE MTT-S International Microwave Symposium  
algorithms based on longitudinal partitioning using the conventional lumped model.  ...  In this paper a waveform relaxation algorithm based on longitudinal partitioning is presented to efficiently model large distributed networks.  ...  Moreover since a delay extraction based longitudinal partitioning is used, the number of subcircuits required to model the line is much smaller than that required using the conventional lumped model.  ... 
doi:10.1109/mwsym.2011.5973517 fatcat:rkuxszgpjbavdodscce5h574xu

Exploiting iterative-ness for parallel ML computations

Henggang Cui, Garth A. Gibson, Eric P. Xing, Alexey Tumanov, Jinliang Wei, Lianghong Xu, Wei Dai, Jesse Haber-Kucharsky, Qirong Ho, Gregory R. Ganger, Phillip B. Gibbons
2014 Proceedings of the ACM Symposium on Cloud Computing - SOCC '14  
Many large-scale machine learning (ML) applications use iterative algorithms to converge on parameter values that make the chosen model fit the input data.  ...  Often, this approach results in the same sequence of accesses to parameters repeating each iteration.  ...  Iterative ML approaches assume a particular mathematical model will describe the input data and use an algorithm to identify parameter values for that model that make it fit the input data most closely  ... 
doi:10.1145/2670979.2670984 dblp:conf/cloud/CuiTWXDHHGGGX14 fatcat:jvl7v67lynem5my4s3ppvujwje

A Multi-GPU Program for Uncertainty-Aware Drainage Basin Delineation: Scalability benchmarking with country-wide data sets

Ville Makinen, Tapani Sarjakoski, Juha Oksanen, Jan Westerholm
2016 IEEE Geoscience and Remote Sensing Magazine  
Processing high-resolution digital elevation models (DEM) can be tedious due to the large size of the data.  ...  All the computations are run on the GPUs, and the parallel processes communicate using Message Passing Interface (MPI) via the host CPUs.  ...  Each partition is extended by a region called halo zone that is used to hold copies of the values from the neighbouring partitions.  ... 
doi:10.1109/mgrs.2016.2561405 fatcat:rh7mhdhgpfcn5o6dfqlajteyfm

High-Level Programming Abstractions for Distributed Graph Processing

Vasiliki Kalavri, Vladimir Vlassov, Seif Haridi
2018 IEEE Transactions on Knowledge and Data Engineering  
Writing distributed graph applications is inherently hard and requires programming models that can cover a diverse set of problem domains, including iterative refinement algorithms, graph transformations  ...  Efficient processing of large-scale graphs in distributed environments has been an increasingly popular topic of research in recent years.  ...  For instance, the popular vertex-centric model [2] is well-suitable for iterative value propagation algorithms, while the neighborhoodcentric model [24] is designed to efficiently support operations  ... 
doi:10.1109/tkde.2017.2762294 fatcat:m5luctm3yjhapbbdrgjcernbbm

Experiments and Recommendations for Partitioning Systems of Equations

Liviu Octavian Mafteiu-Scai
2014 Annals of the West University of Timisoara: Mathematics and Computer Science  
This paper presents some criteria which leads to more efficient parallelization, that must be taken into consideration.  ...  partitioning (or one as close) and the second consist in achieving an adequate preconditioning, depending on a given/desired partitioning.  ...  To solve these shortcomings, hypergraphs are used as models for partitioning in parallel computing [6, 7, 17, 23] .  ... 
doi:10.2478/awutm-2014-0009 fatcat:2omgxs2q25c6jjxm3zbj7cr2im

Mutation-based validation of high-level microprocessor implementations

J. Campos, H. Al-Asaad
2004 Proceedings. Ninth IEEE International High-Level Design Validation and Test Workshop (IEEE Cat. No.04EX940)  
Then we present the clusteringand-partitioning technique that single-handedly makes the concurrent design error simulation of a large set of design errors efficient and allows for the acquisition of statistical  ...  In this paper we present a preliminary method of validating a high-level microprocessor implementation by generating a test sequence for a collection of abstract design error models that can be used to  ...  Once we have chosen a partitioning point per cluster, we can organize each cluster as a hash table where the value of the partitioning point is used as the hashing key.  ... 
doi:10.1109/hldvt.2004.1431242 dblp:conf/hldvt/CamposA04 fatcat:b4flgt3ypjdcpposyhrltsaigq

Full likelihood inference for max-stable data [article]

Raphaël Huser, Clément Dombry, Mathieu Ribatet, Marc G. Genton
2018 arXiv   pre-print
how to perform full likelihood inference for max-stable multivariate distributions or processes based on a stochastic Expectation-Maximisation algorithm, which combines statistical and computational efficiency  ...  The good performance of this methodology is demonstrated by simulation based on the popular logistic and Brown--Resnick models, and it is shown to provide dramatic computational time improvements with  ...  full likelihood (4) can be efficiently computed in this case using a recursive algorithm (Shi, 1995) , thus allowing us to compare θ SEM and θ in high dimensions.  ... 
arXiv:1703.08665v2 fatcat:epb5c7itffctpbnizvrjfym7fe

Accelerated needle steering using partitioned value iteration

A Asadian, M R Kermani, R V Patel
2010 Proceedings of the 2010 American Control Conference  
Unlike conventional solvers, e.g. the value iterator, which suffers from the curse of dimensionality, partitioned-based solvers show promising numerical performance.  ...  This approach exploits a nonholonomic system approach, which models tissue-needle interaction, and formulates the problem as a Markov Decision Process that is solvable using infinite horizon Dynamic Programming  ...  , the pseudocode of prioritized-partitioned value iteration (PPVI) is as follows.  ... 
doi:10.1109/acc.2010.5531362 fatcat:l6smrazg4vg7zdiqoallnupcbu
« Previous Showing results 1 — 15 out of 248,270 results