Cyclops is a new vertex-oriented graph-parallel framework for writing distributed graph analytics. Unlike existing distributed graph computation models, Cyclops retains simplicity and computation-efficiency by synchro...
详细信息
The proceedings contain 29 papers. The topics discussed include: distributed cooperative Q-learning for mobility-sensitive handover optimization in LTE SON;topology selection criteria for a virtual topology controller...
ISBN:
(纸本)9781479942770
The proceedings contain 29 papers. The topics discussed include: distributed cooperative Q-learning for mobility-sensitive handover optimization in LTE SON;topology selection criteria for a virtual topology controller based on neural memories;GP-m: mobile middleware infrastructure for ambient assisted living;formal modeling and checking of an enhanced variant of the IEEE 802.11 CSMAICA protocol;measuring the Internet's threat level: a global-local approach;decomposition of memory consumption footprints to identify problematic threads;monitoring applications and services to improve the cloud foundry PaaS;an efficient MAC-signature scheme for authentication in XOR network coding;near-clouds: bringing public clouds to users' doorsteps;programmable mobile core network;a slot assignment for wireless body area networks;modeling, optimization and performance prediction of parallel algorithms;and automating the Hadoop configuration for easy setup in resilient cloud systems.
An index in a Multi-Version DBMS (MV-DBMS) has to reflect different tuple versions of a single data item. Existing approaches follow the paradigm of logically separating the tuple version data from the data item, e.g....
详细信息
ISBN:
(纸本)9781450326278
An index in a Multi-Version DBMS (MV-DBMS) has to reflect different tuple versions of a single data item. Existing approaches follow the paradigm of logically separating the tuple version data from the data item, e.g. an index is only allowed to return at most one version of a single data item (while it may return multiple data items that match a search criteria). Hence to determine the valid (and therefore visible) tuple version of a data item, the MV-DBMS first fetches all tuple versions that match the search criteria and subsequently filters visible versions using visibility checks. This involves I/O storage accesses to tuple versions that do not have to be fetched. In this vision paper we present the Multi-Version Index (MV-IDX) approach that allows index-only visibility checks which significantly reduce the amount of I/O storage accesses as well as the index maintenance overhead. The MV-IDX achieves significantly lower response times and higher transactional throughput on OLTP workloads.
The proceedings contain 20 papers. The topics discussed include: certification for configurable program analysis;an improvement of the piggyback algorithm for parallel model checking;incremental bounded software model...
ISBN:
(纸本)9781450324526
The proceedings contain 20 papers. The topics discussed include: certification for configurable program analysis;an improvement of the piggyback algorithm for parallel model checking;incremental bounded software model checking;approximating happens-before order: interplay between static analysis and state space traversal;exploiting synchronization in the analysis of shared-memory asynchronous programs;local state space construction for compositional verification of concurrent systems;satisfiability modulo abstraction for separation logic with linked lists;CTL+FO verification as constraint solving;quantifying information leaks using reliability analysis;generic and efficient attacker models in spin;automatic handling of native methods in Java Pathfinder;is there a best Büchi automaton for explicit model checking?;towards a GPGPU-parallel spin model checker;toward parameterized verification of synchronous distributed applications;and towards a test automation framework for alloy.
Simulation is a critical tool for evaluating processor and program performance and behavior in newly proposed computer architectures. When modeling target machines with hundreds or thousands of cores, parallel simulat...
详细信息
The proceedings contain 42 papers. The topics discussed include: regularizing soft decision trees;a simple yet fast algorithm for the closest-pair problem using sorted projections on multi-dimensions;distributed selfi...
ISBN:
(纸本)9783319016030
The proceedings contain 42 papers. The topics discussed include: regularizing soft decision trees;a simple yet fast algorithm for the closest-pair problem using sorted projections on multi-dimensions;distributed selfish algorithms for the max-cut game;distributed binary consensus in dynamic networks;analyzing and predicting patient arrival times;optimal behavior of smart wireless users;hyper-heuristics for performance optimization of simultaneous multithreaded processors;a model of speculative parallel scheduling in networks of unreliable sensors;energy-aware admission control for wired networks;transfer learning using twitter data for improving sentiment classification of Turkish political news;a fully semantic approach to large scale text categorization;a comparative study to determine the effective window size of Turkish word sense disambiguation systems;eyes detection combined feature extraction and mouth information;depth from moving apertures;and score level fusion for face-iris multimodal biometric system.
The ever-growing gap between the computation and I/O is one of the fundamental challenges for future computing systems. This computation-I/O gap is even larger for modern large scale high-performance systems due to th...
详细信息
The proceedings contain 18 papers. The topics discussed include: OSM: an evolutionary system of systems framework for modeling and simulation;assessing instructional strategies for training robot-aided ISR tasks in si...
ISBN:
(纸本)9781632662132
The proceedings contain 18 papers. The topics discussed include: OSM: an evolutionary system of systems framework for modeling and simulation;assessing instructional strategies for training robot-aided ISR tasks in simulated environments;biodiesel sim: crowdsourcing simulations for complex model analysis;modeling and simulation of electricity generated by renewable energy sources for complex energy systems;evaluating pricing options at a museum by simulation;development of a marked structure for workload traces of parallel and distributedsystems;on the generalization of continuous-time stochastic processes simulation for industrial production modeling;solving instability problem in soft body dynamics;a simulation-based framework for the generation and evaluation of traffic management strategies;and human behavior simulation for complex scenarios based on intelligent agents.
As a new area of machine learning research, the deep learning algorithm has attracted a lot of attention from the research community. It may bring human beings to a higher cognitive level of data. Its unsupervised pre...
详细信息
ISBN:
(纸本)9781479941162
As a new area of machine learning research, the deep learning algorithm has attracted a lot of attention from the research community. It may bring human beings to a higher cognitive level of data. Its unsupervised pre-training step allows us to find high-dimensional representations or abstract features which work much better than the principal component analysis (PCA) method. However, it will face problems when being applied to deal with large scale data due to its intensive computation from many levels of training process against large scale data. The sequential deep learning algorithms usually can not finish the computation in an acceptable time. In this paper, we propose a many-core algorithm which is based on a parallel method and is used in the Intel Xeon Phi many-core systems to speed up the unsupervised training process of Sparse Autoencoder and Restricted Boltzmann Machine (RBM). Using the sequential training algorithm as a baseline to compare, we adopted several optimization methods to parallelize the algorithm. The experimental results show that our fully-optimized algorithm gains more than 300-fold speedup on parallelized Sparse Autoencoder compared with the original sequential algorithm on the Intel Xeon Phi coprocessor. Also, we ran the fully-optimized code on both the Intel Xeon Phi coprocessor and an expensive Intel Xeon CPU. Our method on the Intel Xeon Phi coprocessor is 7 to 10 times faster than the Intel Xeon CPU for this application. In addition to this, we compared our fully-optimized code on the Intel Xeon Phi with a Matlab code running on single Intel Xeon CPU. Our method on the Intel Xeon Phi runs 16 times faster than the Matlab implementation. The result also suggests that the Intel Xeon Phi can offer an efficient but more general-purposed way to parallelize the deep learning algorithm compared to GPU. It also achieves faster speed with better parallelism than the Intel Xeon CPU.
Densely-deployed embedded sensor networks are susceptible to constraints associated with contention across a shared transport medium. To improve channel reliability, as well as average power consumption across the sys...
详细信息
ISBN:
(纸本)9781479968510
Densely-deployed embedded sensor networks are susceptible to constraints associated with contention across a shared transport medium. To improve channel reliability, as well as average power consumption across the system, densely-deployed embedded sensor networks often leverage node-based neighbourhood data aggregation strategies. The tradeoff is that individual sensor nodes will have increased memory capacity and access requirements;where access requirements are determined by the memory transport bandwidth, the nature and frequency of the memory accesses, and the latencies associated with the memory storage mechanism. Individual sensor nodes consume power both directly based on the number/nature of memory operations, and indirectly through leakage current through latent circuitry. This paper considers the impact of different memory archetypes on performance of aggregation-related algorithms by individual nodes - specifically the scalability of number of required bus transactions and memory-related latencies with data-set size. The archetypes under consideration were: linear-addressing (RAM), content-based addressing (ternary CAM), and multi-dimensional addressing (Parks'). VHDL-specified MicroBlaze-based nodes, a 32 bit data-bus, and archetypical memories were implemented on a Virtex-5 development board. Operations central to aggregation algorithms (min, sum, count) were run using each type of memory on data-sets of 8 different sizes between 8 and 1024 data-points. Results suggest that appropriate selection of local-node memory architecture, can offer performance benefits in densely deployed sensor networks.
暂无评论