检索结果-内蒙古大学图书馆

Mariana: Tencent Deep Learning Platform and its Applications

PROCEEDINGS OF THE VLDB ENDOWMENT 2014年第13期7卷 1772-1777页

作者： Zou, Yongqiang Jin, Xing Li, Yi Guo, Zhimao Wang, Eryu Xiao, Bin Tencent Inc Shenzhen Peoples R China

Deep learning gains lots of attentions in recent years and is more and more important for mining values in big data. However, to make deep learning practical for a wide range of applications in Tencent Inc., three requirements must be considered: 1) Lots of computational power are required to train a practical model with tens of millions of parameters and billions of samples for products such as automatic speech recognition (ASR), and the number of parameters and training data is still growing. 2) The capability of training larger model is necessary for better model quality. 3) Easy to use frameworks are valuable to do many experiments to perform model selection, such as finding an appropriate optimization algorithm and tuning optimal hyper-parameters. To accelerate training, support large models, and make experiments easier, we built Mariana, the Tencent deep learning platform, which utilizes GPU and CPU cluster to train models parallelly with three frameworks: 1) a multi-GPU data parallelism framework for deep neural networks (DNNs). 2) a multi-GPU model parallelism and data parallelism framework for deep convolutional neural networks (CNNs). 3) a CPU cluster framework for large scale DNNs. Mariana also provides built-in algorithms and features to facilitate experiments. Mariana is in production usage for more than one year, achieves state-of-the-art acceleration performance, and plays a key role in training models and improving quality for automatic speech recognition and image recognition in Tencent WeChat, a mobile social platform, and for Ad click-through rate prediction (pCTR) in Tencent QQ, an instant messaging platform, and Tencent Qzone, a social networking service.

关键词： Speech recognition Big data Data mining Deep neural networks E learning Image enhancement Image recognition Neural networks Acceleration performance Automatic speech recognition Click through rate Cluster framework Computational power Deep convolutional neural networks optimization algorithms Social networking services

来源：评论

学校读者我要写书评

暂无评论

Parameter inversion estimation in photosynthetic models: Impact of different simulation methods

引用

PHOTOSYNTHETICA 2014年第2期52卷 233-246页

作者： Wang, H. B. Ma, M. G. Xie, Y. M. Wang, X. F. Wang, J. Chinese Acad Sci CAREERI Cold & Arid Reg Remote Sensing Observat Syst Expt Lanzhou 730000 Peoples R China China Meteorol Adm Lanzhou Inst Arid Meteorol Lanzhou 730000 Peoples R China

When we apply ecological models in environmental management, we must assess the accuracy of parameter estimation and its impact on model predictions. Parameters estimated by conventional techniques tend to be nonrobust and require excessive computational resources. However, optimization algorithms are highly robust and generally exhibit convergence of parameter estimation by inversion with nonlinear models. They can simultaneously generate a large number of parameter estimates using an entire data set. In this study, we tested four inversion algorithms (simulated annealing, shuffled complex evolution, particle swarm optimization, and the genetic algorithm) to optimize parameters in photosynthetic models depending on different temperatures. We investigated if parameter boundary values and control variables influenced the accuracy and efficiency of the various algorithms and models. We obtained optimal solutions with all of the inversion algorithms tested if the parameter bounds and control variables were constrained properly. However, the efficiency of processing time use varied with the control variables obtained. In addition, we investigated if temperature dependence formalization impacted optimally the parameter estimation process. We found that the model with a peaked temperature response provided the best fit to the data.

关键词： optimization algorithms P-N/C-i curve parameter estimation

来源：评论

学校读者我要写书评

暂无评论

Proposal and optimization of a fuzzy logic model for automobile injection diagnosis using evolutionary algorithms

Proposal and optimization of a fuzzy logic model for automob...

引用

IEEE International Conference on Mechatronics, Electronics and Automotive Engineering (ICMEAE)

作者： Gonzalez Pinzon, Cesar Leonardo Espitia Cuchango, Helbert Eduardo Ponce Corral, Carlos Romero Gonzalez, Jaime Univ Autonoma Ciudad Juarez Ind Engn & Mfg Ciudad Juarez Mexico Univ Distrital Francisco Jose Caldas Syst Engn Program Bogota Colombia Univ Autonoma Ciudad Juarez Ciudad Juarez Mexico

ISBN: (纸本)9781479922536;9781479922529

This article presents results for the design of a fuel injection diagnostic strategy for motor vehicles in Ciudad Juarez, Chihuahua. The first part of this research develops a diagnosis model for the injector MD162524, implementing Fuzzy Logic. Characterization of the diagnostic model took five injector operation tests (spray pattern, injector opening cycles at 3 revolution levels, electrical resistance measurements in the injector, injector flow at a constant pressure and activation of two levels of pressure in the injector) and 3 possible outcomes of diagnostic status (good shape, repairable, useless). In the final part of the investigation, two evolutionary algorithms are used as a tool to optimize the response of the model designed, generating data sets with decreasing measurement error.

关键词： fuzzy logic automotive diagnostic electronic injection optimization algorithms

来源：评论

学校读者我要写书评

暂无评论

Discussion of "Parameter Estimation of the Nonlinear Muskingum Flood-Routing Model Using a Hybrid Harmony Search Algorithm" by Halil Karahan, Gurhan Gurarslan, and Zong Woo Geem

引用

JOURNAL OF HYDROLOGIC ENGINEERING 2014年第4期19卷 845-847页

作者： Hamedi, F. Bozorg Haddad, Omid Orouji, H. Fallah-Mehdipour, E. Marino, M. A. Univ Tehran Coll Agr & Nat Resources Fac Agr Engn & Technol Dept Irrigat & Reclamat Engn Tehran *** Iran Univ Calif Davis Dept Land Air & Water Resources Davis CA 95616 USA Univ Calif Davis Dept Civil & Environm Engn Davis CA 95616 USA Univ Calif Davis Dept Biol & Agr Engn Davis CA 95616 USA

The discusser appreciates the proposed hybrid optimization procedure for a problem of enormous interest to the field of the hydro-logic engineering (Das 2004, 2007; Geem 2006, 2011; Chu 2009; Chu and Chang 2009; Luo and Xie 2010; Barati 2011; Xu et al. 2012; Orouji et al. 2013; Barati 2013a). The hybrid algorithm was planned to improve the parameter calibration of the nonlinear Muskingum model by using a heuristic algorithm [i.e., harmony search (HS)] for global search and a gradient-based technique [i.e., Broyden-Fletcher-Goldfarb- Shanno (BFGS)] for local search in a cascade. Previously, both HS and BFGS algorithms were separately used for the same problem (Kim et al. 2001; Geem 2006). Furthermore, an indirect penalty condition was proposed to prevent the negativity of outflows and storages in the calibration procedure. The developed algorithm was examined by using the benchmark and real-life examples. The discusser globally agrees with the steps followed by the authors. However, the following points and question are useful and important in the flood routing problem: 1. The classification of flood routing procedures.2. Nonlinear storage equations of Muskingum model.3. Does a hybrid optimization algorithm only find the optimal solution efficiently?

关键词： optimization Procedure Hydrogen Sulfide (H2S3) Flood routing storages Heuristic algorithms outflow Parameter estimation BENCHMARKS aluminium optimization algorithms nonlinearity

来源：评论

学校读者我要写书评

暂无评论

Best-arm Identification algorithms for Multi-Armed Bandits in the Fixed Confidence Setting 48th

Best-arm Identification Algorithms for Multi-Armed Bandits i...

引用

48th Annual Conference on Information Sciences and Systems (CISS)

作者： Jamieson, Kevin Nowak, Robert Univ Wisconsin Dept Elect & Comp Engn Madison WI 53706 USA

ISBN: (纸本)9781479930012

This paper is concerned with identifying the arm with the highest mean in a multi-armed bandit problem using as few independent samples from the arms as possible. While the so-called "best arm problem" dates back to the 1950s, only recently were two qualitatively different algorithms proposed that achieve the optimal sample complexity for the problem. This paper reviews these recent advances and shows that most best-arm algorithms can be described as variants of the two recent optimal algorithms. For each algorithm type we consider a specific instance to analyze both theoretically and empirically thereby exposing the core components of the theoretical analysis of these algorithms and intuition about how the algorithms work in practice. The derived sample complexity bounds are novel, and in certain cases improve upon previous bounds. In addition, we compare a variety of state-of-the-art algorithms empirically through simulations for the best-arm-problem.

关键词： Theoretical analysis MULTI-ARM optimization algorithms Intuition Arms algorithms Arm Core Assembly variants spiral arm sample complexity Bound

来源：评论

学校读者我要写书评

暂无评论

A Method to Enforce Map Constraints in a Particle Filter's Position Estimate

A Method to Enforce Map Constraints in a Particle Filter's P...

引用

11th Workshop on Positioning, Navigation and Communication (WPNC)

作者： Piche, Robert Koivisto, Mike Tampere Univ Technol Dept Automat Sci & Engn FIN-33101 Tampere Finland

ISBN: (纸本)9781479946716

When particle filters are used to estimate indoor position with floor plan constraints, it can happen that, even when all the particles lie in the corridor, the particles' mean is not in the corridor. Such a position estimate is perceived by the user as a mistake in the algorithm. Projecting the particles' mean to the nearest corridor location is an obvious ad-hoc solution, but it is not optimal and the trajectory may be discontinuous in time. Another solution is to use a maximum a-posteriori estimate for the particle cloud where the particles in an inaccessible region are eliminated. However, this optimal solution might also have discontinuous trajectory and so it is not ideal for the real time positioning. In this work, the following principled approach is taken. Given a particle cloud representation of a posterior distribution for position, the position estimate is defined as the solution of a least squares problem with linear inequality constraints. This problem can be solved efficiently and reliably using standard numerical optimization algorithms and codes. Results are presented for simulated data and real-world data.

关键词： position estimation Particle filters Corridors particles Discontinuous Posterior distribution trajectory optimization algorithms simulated data enforcement

来源：评论

学校读者我要写书评

暂无评论

Visual Servoing of a Robotic Manipulator Using an Optimized Trajectory Planning Technique 27

Visual Servoing of a Robotic Manipulator Using an Optimized ...

引用

IEEE 27th Canadian Conference on Electrical and Computer Engineering (CCECE)

作者： Keshmiri, Mohammad Xie, Wen-Fang Concordia Univ Mechcan & Ind Engn Dept Montreal PQ Canada

ISBN: (纸本)9781479930999

In this paper, a novel image based visual servoing (IBVS) using an optimized trajectory planning technique is proposed. In this method the camera's velocity screw is parameterized using polynomial profiles. T he p arameters o f the velocity profile a re t hen d etermined t o g uide t he r obot t o its desired position, by minimizing the error in an optimization algorithm. In addition, this technique keeps the robot in its physical limits and keeps the features in the field of view (FOV). This algorithm is tested on a 4 DOFs Denso robot in an eye-in-hand configuration. Simulation results demonstrate t hat the proposed method solves some of the IBVS problems such as reaching the robot's or cameras' limits and instability of the system for pure 180 degrees rotation about its center.

关键词： Visual servoing Robot Manipulator Robot trajectory planning optimization algorithms Cameras VELOCITY PROFILING

来源：评论

学校读者我要写书评

暂无评论

A Novel Multi-Agent System for Complex Scheduling Problems

A Novel Multi-Agent System for Complex Scheduling Problems

引用

Winter Simulation Conference

作者： Hillmann, Peter Uhlig, Tobias Rodosek, Gabi Dreo Rose, Oliver Univ Bundeswehr Munchen Dept Comp Sci D-85577 Neubiberg Germany

ISBN: (纸本)9781479974863

Complex scheduling problems require a large amount computation power and innovative solution methods. The objective of this paper is the conception and implementation of a multi-agent system that is applicable in various problem domains. Independent specialized agents handle small tasks, to reach a superordinate target. Effective coordination is therefore required to achieve productive cooperation. Role models and distributed artificial intelligence are employed to tackle the resulting challenges. We simulate a NP-hard scheduling problem to demonstrate the validity of our approach. In addition to the general agent based framework we propose new simulation-based optimization heuristics to given scheduling problems. Two of the described optimization algorithms are implemented using agents. This paper highlights the advantages of the agent-based approach, like the reduction in layout complexity, improved control of complicated systems, and extendability.

关键词： computational complexity multi-agent systems optimisation scheduling simulation NP-hard scheduling problem agent based framework agent-based approach complex scheduling problems computation power distributed artificial intelligence innovative solution methods layout complexity multiagent system optimization algorithms productive cooperation role models simulation-based optimization heuristics Abstracts Computer architecture Multi-agent systems optimization Processor scheduling Schedules Servers

来源：评论

学校读者我要写书评

暂无评论

Optimal Design of a Transfemoral Prosthesis with Energy Storage and Regeneration

Optimal Design of a Transfemoral Prosthesis with Energy Stor...

引用

American Control Conference

作者： Rarick, Rick Richter, Hanz van den Bogert, Antonie Simon, Dan Warner, Holly Barto, Taylor Cleveland State Univ Cleveland OH 44115 USA

ISBN: (纸本)9781479932740

We describe the preliminary optimal design of an electromechanical above-knee active prosthesis with energy storage and regeneration. A DC motor-generator applies a positive or negative torque to the knee. The control system regulates the exchange of energy between the motor-generator and a supercapacitor. The central idea of the design is motivated by the mechanics, energy management, and sensor-based control that constitute human movement. We use biogeography-based optimization, which is an evolutionary algorithm, to optimize the system parameters, and we evaluate its performance with Simulink (R) models. We optimize three alternative prosthesis designs. Simulation results indicate that the prosthesis can be optimized to achieve knee angle tracking with an RMS error on the order of 0.2 degrees.

关键词： Biologically-inspired methods Modeling and simulation optimization algorithms

来源：评论

学校读者我要写书评

暂无评论

Determining the optimal redistribution for a given data partition 13

Determining the optimal redistribution for a given data part...

引用

13th IEEE International Symposium on Parallel and Distributed Computing (ISPDC)

作者： Herault, Thomas Herrmann, Julien Marchal, Loris Robert, Yves Univ Tennessee Knoxville TN 37996 USA Ecole Normale Super Lyon CNRS INRIA Lyon France Univ Lyon Lyon France

ISBN: (纸本)9781479959198

The classical redistribution problem aims at optimally scheduling communications when moving from an initial data distribution to a target distribution where each processor will host a subset of data items. However, modern computing platforms are equipped with a powerful interconnection switch, and the cost of a given communication is (almost) independent of the location of its sender and receiver. This leads to generalizing the redistribution problem as follows: find the optimal one-toone mapping of the subsets of data items onto the processors for which the cost of the redistribution is minimal. This paper studies the complexity of this generalized problem. We provide optimal algorithms and evaluate their gain over classical redistribution through simulations. We also show the NP-hardness of the problem to find the optimal data partition and processor permutation (defined by new subsets) that minimize the cost of redistribution followed by a simple computation kernel.

关键词： Bipartite graph Complexity theory Distributed databases Kernel Measurement optimization Partitioning algorithms Partitioning algorithms distributed databases Bipartite graph Complexity theory PROCESSOR Kernel Seeds optimization algorithms Subset

来源：评论

学校读者我要写书评

暂无评论

建议与咨询留下您的常用邮箱和电话号码，以便我们向您反馈解决方案和替代方法

时间限定

文献类型

馆藏选择

核心期刊

语言

文献类型

帮助

文字说明：

检索规则说明：

检索范例：

分类表

所选分类

限定检索结果

文献类型

馆藏范围

日期分布

学科分类号

主题

机构

作者

语言

请选择保存的检索档案：

请选择收藏分类：

通借通还

建议与咨询 留下您的常用邮箱和电话号码，以便我们向您反馈解决方案和替代方法

时间限定

文献类型

馆藏选择

核心期刊

语言

文献类型

帮助

文字说明：

检索规则说明：

检索范例：

分类表

所选分类

限定检索结果

文献类型

馆藏范围

日期分布

学科分类号

主题

机构

作者

语言

请选择保存的检索档案： 新增检索档案 确定 取消

请选择收藏分类： 新增自定义分类 确定 取消

通借通还

建议与咨询留下您的常用邮箱和电话号码，以便我们向您反馈解决方案和替代方法

请选择保存的检索档案：

请选择收藏分类：