检索结果-内蒙古大学图书馆

International Joint Conference on Neural Networks

作者： Yaozheng WANG Dawei FENG Dongsheng LI Xinyuan CHEN Yunxiang ZHAO Xin NIU National Lab for Parallel and Distributed Processing School of Computer Science National University of Defense Technology

ISBN: (纸本)9781509006212

Real-life behaviors shown by the mobile users typically exhibit plenty noises, making it hard to construct an effective recommendation engine. In this paper, we present a fused model based on the LR algorithm and the GBDT algorithm to recommend vertical industry commodities in a mobile setting. A set of specifically designed methods are proposed to deal with the data preprocessing and feature extraction problem for the mobile recommendation scenario. The proposed method is evaluated on a large scale real-world dataset provided by the Alibaba mobile shopping department. Result on the F1 score has seen an improvement of 2%-36% compared with the baseline.

关键词： Mobile communication Feature extraction Predictive models Prediction algorithms Logistics Data models Computational modeling

来源：评论

学校读者我要写书评

暂无评论

Temporal Inconsistency-Based Intrinsic Reward for Multi-Agent Reinforcement Learning

Temporal Inconsistency-Based Intrinsic Reward for Multi-Agen...

引用

International Joint Conference on Neural Networks (IJCNN)

作者： Shaoqi Sun Kele Xu National Key Lab of Parallel and Distributed Processing National University of Defense Technology Changsha China

Multi-agent reinforcement learning (MARL) has shown promising results in many challenging sequential decision-making tasks. Recently, deep neural networks have dominated this field. However, the policy networks of agent's may fall into local optimum during the training phase, which severely constrains the performance of exploration. To address this issue, we propose a novel MARL learning framework named PSAM, which contains a new temporal inconsistency-based intrinsic reward and a diversity control strategy. Specifically, we save the parameters of the deep models along the optimization path of the agent's policy network, which can be denoted as snapshots. Through measuring the difference between snapshots, we can employ the difference as an intrinsic reward. Moreover, we propose a diversity control strategy to improve the performance further. Finally, to verify the effectiveness of the proposed method, we conduct extensive experiments in several widely used MARL environments. The results show that in many environments, PSAM can not only achieve state-of-the-art performance and prevent the policy network from getting stuck in local minima but also accelerate the agent's learning of the policy. It is worth noting that the proposed regularizer can be used using a plug-and-play manner without introducing any additional hyper-parameters and training costs.

关键词：

来源：评论

学校读者我要写书评

暂无评论

Graph pooling via coarsened graph infomax

arXiv

引用

arXiv 2021年

作者： Pang, Yunsheng Zhao, Yunxiang Li, Dongsheng The University of Melbourne National Lab for Parallel and Distributed Processing College of Computer National University of Defense Technology

Graph pooling that summaries the information in a large graph into a compact form is essential in hierarchical graph representation learning. Existing graph pooling methods either suffer from high computational complexity or cannot capture the global dependencies between graphs before and after pooling. To address the problems of existing graph pooling methods, we propose Coarsened Graph Infomax Pooling (CGIPool) that maximizes the mutual information between the input and the coarsened graph of each pooling layer to preserve graph-level dependencies. To achieve mutual information neural maximization, we apply contrastive learning and propose a self-attention-based algorithm for learning positive and negative samples. Extensive experimental results on seven datasets illustrate the superiority of CGIPool comparing to the state-of-the-art methods. Copyright © 2021, The Authors. All rights reserved.

关键词： Graph neural networks

来源：评论

学校读者我要写书评

暂无评论

An Improved Independent Set Ordering Algorithm for Solving Large-scale Sparse Linear Systems

An Improved Independent Set Ordering Algorithm for Solving L...

引用

2010 Second International Conference on Intelligent Human-Machine Systems and Cybernetics(第二届智能人机系统与控制论国际学术会议 IHMSC 2010)

作者： Lu Yao Wei Cao Zongzhe Li Yongxian Wang Zhenghua Wang National Key Lab for Parallel and Distributed Processing National Univ.of Defense TechnologyChangshaChina

The independent set ordering algorithm is a heuristic algorithm based on finding maximal independent sets of vertices in the matrix adjacency graph, which is commonly used for parallel matrix factorization. However, Disadvantages appear when it is applied to large-scale sparse linear systems. In this paper, we propose an improved algorithm by finding an optimal size of independent set in each elimination step rather than find a maximal independent set, which is proved to be effective by both theoretical analysis and parallel implementation.

关键词： independent set parallel LU decomposition large-scale sparse linear systems matrix ordering

来源：评论

学校读者我要写书评

暂无评论

Shuffle-Efficient distributed Locality Sensitive Hashing on Spark

Shuffle-Efficient Distributed Locality Sensitive Hashing on ...

引用

IEEE Conference on Computer Communications Workshops

作者： Wanxin Zhang Dongsheng Li Ying Xu Yiming Zhang National Lab for Parallel and Distributed Processing College of Computer National University of Defense Technology Changsha Hunan China

ISBN: (纸本)9781467399562

Locality Sensitive Hashing (LSH) is an important indexing technique for approximate similarity search in high-dimensional spaces. An obvious limitation of LSH approaches is the lack of capability and scalability to deal with massive data. This paper proposes a distributed variant of LSH called Spark-LSH, which is implemented on Apache Spark, a well-known distributed computing framework. We design a shuffle-efficient indexing scheme for the Spark-LSH, which can reduce the data shuffle and improve the network efficiency when constructing the hash table indices. Furthermore, we propose a location-aware querying scheme to improve the query performance. Experiments show that the Spark-LSH scheme can reduce the network shuffle overhead remarkably and accelerate the query significantly.

关键词： Locality Sensitive Hashing Spark Shuffle

来源：评论

学校读者我要写书评

暂无评论

Messages from the Program Co-chairs: NAS 2009

Proceedings - 2009 IEEE International Conference on Networki...

引用

Proceedings - 2009 IEEE International Conference on Networking, Architecture, and Storage, NAS 2009 2009年 xii页

作者： Zhu, Yifeng Fang, Liang Hu, Weiwu University of Maine United States National Lab of Parallel and Distributed Processing China Chinese Academy of Science China

来源：评论

学校读者我要写书评

暂无评论

A Comparative Study of parallel Prefix Adders in FPGA Implementation of EAC

A Comparative Study of Parallel Prefix Adders in FPGA Implem...

引用

Euromicro Symposium on Digital System Design

作者： Feng Liu Fariborz Fereydouni Forouzandeh Otmane Ait Mohamed Gang Chen Xiaoyu Song Qingping Tan Parallel Distributed Processing National Laboratory Hunan China National Lab of Parallel Distributed processing Hunan China ECE department Concordia University Montreal Quebec Canada Lingcore Lab Portland OR USA ECE department Portland State University Portland OR USA

Several regular parallel trees have been proposed over the years to optimize logic depth, area, fan-out and interconnect count for logic circuits. In this paper, we propose a comparative study of different parallel prefix trees used in the design of a new end-around carry (EAC) adder targeting FPGA technology. This new adder is based on the fast 128-bit binary floating-point EAC adder which has been implemented in the IBM POWER6 microprocessor's fused multiply-add unit. The parallel prefix tree implemented on the IBM's EAC adder is a Kogge-Stone tree which has been chosen for its high performance and its low power consumption. Our comparative study highlights the main performance differences among fourteen different architecture configurations when targeting an FPGA EAC adder design. We focus on the area requirements and the critical path delay of these designs. Our experimental results show that there is one architecture configuration with the lower area requirement and the higher performance.

关键词： Field programmable gate arrays Adders Microprocessors distributed processing Logic circuits Delay Testing Digital systems Design methodology Design optimization

来源：评论

学校读者我要写书评

暂无评论

Modelling with relational calculus of object and component systems - RCOS

Modelling with relational calculus of object and component s...

引用

作者： Chen, Zhenbang Hannousse, Abdel Hakim Van Hung, Dang Knoll, Istvan Li, Xiaoshan Liu, Zhiming Liu, Yang Nan, Qu Okika, Joseph C. Ravn, Anders P. Stolz, Volker Yang, Lu Zhan, Naijun International Institute for Software Technology United Nations University China Faculty of Science and Technology University of Macau Department of Computer Science Aalborg University Denmark Lab. of Computer Science Institute of Software CAS China National Laboratory for Parallel and Distributed Processing China

ISBN: (纸本)3540852883

This chapter presents a formalization of functional and behavioural requirements, and a refinement of requirements to a design for CoCoME using the Relational Calculus of Object and Component Systems (rCOS). We give a model of requirements based on an abstraction of the use cases described in Chapter 3.2. Then the refinement calculus of rCOS is used to derive design models corresponding to the top level designs of Chapter 3.4. We demonstrate how rCOS supports modelling different views and their relationships of the system and the separation of concerns in the development. © 2008 Springer-Verlag Berlin Heidelberg.

关键词： Calculations

来源：评论

学校读者我要写书评

暂无评论

FINT: FIELD-AWARE INTERACTION NEURAL NETWORK FOR CTR PREDICTION

arXiv

引用

arXiv 2021年

作者： Zhao, Zhishan Yang, Sen Liu, Guohui Feng, Dawei Xu, Kele iQIYI Inc. China National Key Lab of Parallel and Distributed Processing Changsha China National University of Defense Technology Changsha China

As a critical component for online advertising and marketing, click-through rate (CTR) prediction has drawn lots of attention from both industry and academia. Recently, deep learning has become the mainstream methodological choice for CTR. Despite sustainable efforts have been made, existing approaches still pose several challenges. On the one hand, high-order interaction between the features is under-explored. On the other hand, high-order interactions may neglect the semantic information from the low-order fields. In this paper, we proposed a novel prediction method, named FINT, that employs the Field-aware INTeraction layer which explicitly captures high-order feature interactions while retaining the low-order field information. To empirically investigate the effectiveness and robustness of the FINT, we perform extensive experiments on the three realistic databases: KDD2012, Criteo and Avazu. The obtained results demonstrate that the FINT can significantly improve the performance compared to the existing methods, without increasing the amount of computation required. Moreover, the proposed method brought about 2.72% increase to the advertising revenue of iQIYI, a big online video app through A/B testing. To better promote the research in CTR field, we released our code as well as reference implementation at: https://***/zhishan01/FINT. © 2021, CC BY.

关键词： Deep learning

来源：评论

学校读者我要写书评

暂无评论

Low-latency last-level cache structure based on grouped cores in Chip Multi-Processor

Low-latency last-level cache structure based on grouped core...

引用

IEEE International Conference on Performance, Computing and Communications (IPCCC)

作者： Jinbo Xu Weixia Xu Kefei Wang Zhengbin Pang School of Computer National University of Defense Technology Changsha China National Lab. for Parallel and Distributed Processing National University of Defense Technology Changsha China

ISBN: (纸本)9781479975761

Last-Level Cache (LLC) plays an important role in Chip Multi-Processor (CMP). The objective of this work is to optimize the structure and management strategy of LLC. Based on 8-core CMP, a LLC structure based on grouped cores is proposed, where 8 cores are divided into 4 groups. All LLC resources are classified into three types, which are fixed private cache, dynamic private cache and dynamic shared cache. The layout of the LLC structure and the corresponding dynamic partitioning strategy are designed to achieve low access latency and high efficiency. Experimental results on full-system simulator suggest that the proposed structure and method are able to reduce the access latency by 2% to 12% compared with previous works, such as tiled structure, cache-centered structure and core-centered structure. Consequently, performance measured by IPC is improved up to 7%. The contribution of this paper is useful for CMP performance, and applies to not only 8-core CMP but also all small-scale CMPs.

关键词： Dynamic scheduling Multicore processing Educational institutions Organizations Delays Semiconductor device measurement Research and development

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：