检索结果-内蒙古大学图书馆

Online Anomaly Detection With Bandwidth Optimized Hierarchical Kernel Density Estimators

IEEE TRANSACTIONS ON NEURAL NETWORKS AND LEARNING SYSTEMS 2021年第9期32卷 4253-4266页

作者： Kerpicci, Mine Ozkan, Huseyin Kozat, Suleyman Serdar Georgia Inst Technol Sch Elect & Comp Engn Atlanta GA 30332 USA Sabanci Univ Fac Engn & Nat Sci TR-34956 Istanbul Turkey Bilkent Univ Dept Elect & Elect Engn TR-06800 Ankara Turkey

We propose a novel unsupervised anomaly detection algorithm that can work for sequential data from any complex distribution in a truly online framework with mathematically proven strong performance guarantees. First, a partitioning tree is constructed to generate a doubly exponentially large hierarchical class of observation space partitions, and every partition region trains an online kernel density estimator (KDE) with its own unique dynamical bandwidth. At each time, the proposed algorithm optimally combines the class estimators to sequentially produce the final density estimation. We mathematically prove that the proposed algorithm learns the optimal partition with kernel bandwidths that are optimized in both region-specific and time-varying manner. The estimated density is then compared with a data-adaptive threshold to detect anomalies. Overall, the computational complexity is only linear in both the tree depth and data length. In our experiments, we observe significant improvements in anomaly detection accuracy compared with the state-of-the-art techniques.

关键词： Bandwidth Anomaly detection Kernel Estimation partitioning algorithms Data models Real-time systems Anomaly detection bandwidth selection kernel density estimation online regret analysis

来源：评论

学校读者我要写书评

暂无评论

Using Domain Knowledge to Assist Process Scenario Discoveries 46

Using Domain Knowledge to Assist Process Scenario Discoverie...

引用

46th Annual IEEE-Computer-Society International Computers, Software, and Applications Conference (COMPSAC) - Computers, Software, and Applications in an Uncertain World

作者： Zhang, Zhenyu Zhu, Junjie Hildebrant, Ryan Venkatasubramanian, Nalini Ren, Shangping San Diego State Univ Dept Comp Sci San Diego CA 92182 USA Univ Calif Irvine Dept Comp Sci Irvine CA 92697 USA Princeton Univ Dept Civil & Environm Engn Princeton NJ 08544 USA

ISBN: (纸本)9781665488105

Trace clustering techniques are often used to assist in discovering different process scenarios. Most existing trace clustering methods aim to partition the event log into different subsets of logs where event traces in the same subset most likely belong to the same scenario. Typically, the partitioning is based on certain patterns, the similarity between the traces, or done by discovering a process model for each cluster of traces. However, most algorithms achieve this by solely using an event log, without allowing the domain expert to influence the discovery in any way. The domain expert may have certain domain expertise which should be exploited to create better process scenario models. This paper presents a density-based trace clustering technique based on expert knowledge to assist process scenario discovery results. A real wastewater treatment process provided by a domain expert is used as a case study to investigate the effectiveness and validity of the approach. We also use five real-life event logs to compare the performance of the process approach for process scenario discoveries with the commonly used k-means clustering approach. To measure the validity of our results, we take the weighted average fitness, the weighted average precision, and the F1 score. The experiment data show that (1) the proposed approach is able to discover the process scenarios from event logs by incorporating domain knowledge and (2) the process models obtained with the proposed approach have higher fitness, precision, and F1 scores than the models obtained by the commonly used k-means clustering approaches.

关键词： Weight measurement Computational modeling Clustering methods Clustering algorithms Data models partitioning algorithms Wastewater treatment

来源：评论

学校读者我要写书评

暂无评论

Efficient Pipeline Planning for Expedited Distributed DNN Training 41

Efficient Pipeline Planning for Expedited Distributed DNN Tr...

引用

41st IEEE Conference on Computer Communications (IEEE INFOCOM)

作者： Luo, Ziyue Yi, Xiaodong Long, Guoping Fan, Shiqing Wu, Chuan Yang, Jun Lin, Wei Univ Hong Kong Dept Comp Sci Hong Kong Peoples R China Alibaba Hangzhou Peoples R China

ISBN: (纸本)9781665458221

To train modern large DNN models, pipeline parallelism has recently emerged, which distributes the model across GPUs and enables different devices to process different microbatches in pipeline. Earlier pipeline designs allow multiple versions of model parameters to co-exist (similar to asynchronous training), and cannot ensure the same model convergence and accuracy performance as without pipelining. Synchronous pipelining has recently been proposed which ensures model performance by enforcing a synchronization barrier between training iterations. Nonetheless, the synchronization barrier requires waiting for gradient aggregation from all microbatches and thus delays the training progress. Optimized pipeline planning is needed to minimize such wait and hence the training time, which has not been well studied in the literature. This paper designs efficient, near-optimal algorithms for expediting synchronous pipeline-parallel training of modern large DNNs over arbitrary inter-GPU connectivity. Our algorithm framework comprises two components: a pipeline partition and device mapping algorithm, and a pipeline scheduler that decides processing order of microbatches over the partitions, which together minimize the periteration training time. We conduct thorough theoretical analysis, extensive testbed experiments and trace-driven simulation, and demonstrate our scheme can accelerate training up to 157% compared with state-of-the-art designs.

关键词： Training Performance evaluation Graphics processing units Life estimation partitioning algorithms Planning Delays

来源：评论

学校读者我要写书评

暂无评论

A Fast Multi-tree Partition Algorithm Based on Spatial-temporal Correlation for VVC

A Fast Multi-tree Partition Algorithm Based on Spatial-tempo...

引用

Data Compression Conference (DCC)

作者： Liu, Zhi Qian, Houyu Zhang, Mengmeng North China Univ Technol Beijing Peoples R China

ISBN: (纸本)9781665478939

The quadtree nested multi-type trees (MTT) partitioning scheme is adopted by the latest generation of video coding standard H.266/VVC. While the performance of the encoder is improved, the complexity has increased by 2.2-5.6 times. In order to reduce the coding complexity, a fast multi-tree partition algorithm based on spatio-temporal information for inter-frame coding is proposed in this paper. The motion characteristics and texture complexity of the current coding block are utilized simultaneously to determine whether the multi-tree partition can be terminated early. For blocks that need to be further divided, the partition direction is determined in advance.

关键词： Video coding Correlation Data compression Encoding Complexity theory partitioning algorithms Standards

来源：评论

学校读者我要写书评

暂无评论

Perceptual Quality Aware Adaptive 360-Degree Video Streaming with Deep Reinforcement Learning

Perceptual Quality Aware Adaptive 360-Degree Video Streaming...

引用

IEEE International Conference on Communications (ICC)

作者： Feng, Qingxuan Yang, Peng Lyu, Feng Yu, Li Huazhong Univ Sci & Technol Sch Elect Informat & Commun Wuhan Peoples R China Cent South Univ Sch Comp Sci & Engn Changsha Peoples R China

ISBN: (数字)9781538683477

ISBN: (纸本)9781538683477

As 360-degree videos are with high data volume, it is a great challenge to deliver the content and provide a high quality of experience (QoE) for users. In this paper, we investigate the tile-level rate allocation problem with the purpose of optimizing users' QoE. Specifically, we find the nonlinearity between video quality and bitrate through extensive experiments. Thus, a QoE metric is defined to better measure the perceptual quality. Considering the sequential decision nature of video streaming, we formulate the rate decision problem as an Markov Decision Process. Then we propose a deep reinforcement learning based rate adaptive streaming approach to solve this problem. However, the solution space is large as a 360-degree video is spatially partitioned into multiple tiles. In order to address the problem of combinatorial explosion, we propose a tile classification method based on the predicted viewpoint. Experimental results based on real-world traces show that our algorithm can improve the overall QoE by 16% - 22% compared to existing algorithms.

关键词： Measurement Reinforcement learning Streaming media Prediction algorithms Classification algorithms Quality assessment partitioning algorithms

来源：评论

学校读者我要写书评

暂无评论

A New Nested Partition Algorithm for Parallel Machine Scheduling Problem with hard Q-Times and Setup Times 18

A New Nested Partition Algorithm for Parallel Machine Schedu...

引用

IEEE 18th International Conference on Automation Science and Engineering (IEEE CASE)

作者： Wang, Chaoran Shi, Leyuan Univ Wisconsin Coll Engn Dept Ind & Syst Engn Madison WI 53706 USA

ISBN: (数字)9781665490429

ISBN: (纸本)9781665490429

This paper addresses the non-preemptive unrelated parallel machine scheduling problems with machine-dependent and sequence-dependent setup times, machine eligibility restrictions and hard Q-time constraints. The objective is to minimize the makespan. All jobs are assumed to be available at time zero and all times are deterministic. To formulate this complex problem, a mixed integer linear programming (MILP) model is developed and the lower bound is derived to evaluate the proposed algorithms. Due to the complexity of the problem, the nested partition framework is introduced and two new heuristic algorithms are designed based on it. One is embedded with one-step-backtracking rule (NP-O), and the other uses root-backtracking (NP-R). The performance is evaluated by conducting both algorithms on test instances. The computational results demonstrate that both proposed algorithms are efficient and effective at solving small scale instances while NP-O outperforms NP-R for large scale problems.

关键词： Job shop scheduling Processor scheduling Heuristic algorithms Parallel machines Sampling methods Robustness partitioning algorithms

来源：评论

学校读者我要写书评

暂无评论

Typical Characteristic-Based Type-2 Fuzzy C-Means Algorithm

引用

IEEE TRANSACTIONS ON FUZZY SYSTEMS 2021年第5期29卷 1173-1187页

作者： Yang, Xiyang Yu, Fusheng Pedrycz, Witold Quanzhou Normal Univ Key Lab Intelligent Comp & Informat Proc Fujian P Quanzhou 362000 Peoples R China Beijing Normal Univ Sch Math Sci Beijing 100875 Peoples R China Univ Alberta Dept Elect & Comp Engn Edmonton AB T6R 2V4 Canada King Abdulaziz Univ Dept Elect & Comp Engn Fac Engn Jeddah 21589 Saudi Arabia Polish Acad Sci Syst Res Inst PL-01447 Warsaw Poland

Type-2 fuzzy sets provide an efficient vehicle for handling uncertainties of real-world problems, including noisy observations. Bringing type-2 fuzzy sets to clustering algorithms offers more flexibility to handle uncertainties associated with membership concepts caused by a noisy environment. However, the existing type-2 fuzzy clustering algorithms suffer from a time-consuming type-reduction process, which not only hampers the clustering performance but also increases the burden of understanding the clustering results. In order to alleviate the problem, this article introduces a set of typical characteristics of type-2 fuzzy sets and establishes a characteristic-based type-2 fuzzy clustering algorithm. Being different from the objective function used in the fuzzy C-means (FCM) algorithm that produces cluster centers and type-1 memberships, the objective function in the proposed algorithm contains additional characteristics of type-2 membership grades, namely, centers of gravity and cardinalities of the secondary fuzzy sets. The derived iterative formulas used for these parameters are much more efficient than the interval type-2 FCM algorithm. The experiments carried out in this study show that the proposed typical characteristic-based type-2 FCM algorithm has an ability of detecting noise as well as assigning suitable membership degrees to the individual data.

关键词： Clustering algorithms Fuzzy sets Noise measurement Phase change materials Uncertainty Linear programming partitioning algorithms Cardinality center of gravity (COG) characteristics fuzzy clustering type-2 fuzzy sets (T2FS)

来源：评论

学校读者我要写书评

暂无评论

An Integrated Circuit partitioning and TDM Assignment Optimization Framework for Multi-FPGA Systems

An Integrated Circuit Partitioning and TDM Assignment Optimi...

引用

Asia and South Pacific Design Automation Conference

作者： Dan Zheng Evangeline F.Y. Young The Chinese University of Hong Kong Hong Kong SAR China

In multi-FPGA systems, Time-Division Multiplexing (TDM) is a widely used method for transferring multiple signals over a common wire. The circuit performance will be significantly influenced by this inter-Fpga delay. Some inter-Fpga nets are driven by different clocks, in which case they cannot share the same wire. In this paper, to minimize the maximum delay of inter-Fpga nets, we propose a two-step framework. First, a TDM-aware partitioning algorithm is adopted to minimize the maximum cut size between an FPGA-pair. A TDM ratio assignment method is then applied to assign TDM ratio for each inter-Fpga net optimally. Experimental results show that our algorithm can reduce the maximum TDM ratio significantly within reasonable runtime.

关键词： Integrated circuits Runtime Design automation System performance Wires Time division multiplexing partitioning algorithms

来源：评论

学校读者我要写书评

暂无评论

HPG-Tree: An Enhanced B+-Tree for Spatial Indexing

引用

IEEE Access 2025年 13卷 91312-91324页

作者： Amal Saif Alaa M. Altarazi Amer Al-Badarneh Mohammad J. Abdel-Rahman Computer Science Department Princess Sumaya University for Technology Amman Jordan Data Science Department Princess Sumaya University for Technology Amman Jordan

In modern applications such as geographic information systems, game development, and image processing, the demand for efficient two-dimensional spatial indexing is rising exponentially. Traditional approaches like R-tree can suffer from excessive overlap and fragmentation, while standard $B^{+}$ -trees are limited to one-dimensional data. Motivated by these gaps, we propose the hyperbolic paraboloid grid (HPG) tree, an enhanced $B^{+}$ -tree structure specifically tailored for two-dimensional datasets. First, we employ a grid-based strategy that maps two-dimensional points into one-dimensional values without resorting to hashing. By combining grid partitioning with the $B^{+}$ -tree’s efficient, self-balancing properties, the HPG-tree simultaneously addresses overlap, storage utilization, and query performance. Unlike purely one-dimensional indexes, this method preserves spatial locality and avoids excessive rebalancing. Further, the HPG-tree integrates a hyperbolic paraboloid-like framework in its node layout, allowing it to halve the number of index nodes in many cases. This optimization is key to reducing disk or memory scans, thereby accelerating point lookups, range queries, and updates. Our experiments demonstrate that the HPG-tree outperforms comparable state-of-the-art structures like $R^{*}$ -tree, achieving up to a 50% improvement in storage utilization and significant gains in query throughput. HPG-tree opens a promising avenue for modern spatial databases that require both high scalability and sub-second response, making it a robust and flexible option for today’s evolving data-intensive environments. These advantages make HPG-tree particularly well-suited to large-scale spatial databases requiring both high scalability and sub-second responsiveness. Following a detailed discussion of the shortcomings in existing methods, the paper presents the HPG-tree methodology, including its grid partitioning, node layout, and insertion procedures, illustrating why

关键词： Indexing Spatial databases Spatial indexes Memory management Standards Nearest neighbor methods Vegetation Taxonomy partitioning algorithms Optimization

来源：评论

学校读者我要写书评

暂无评论

BaPa: A Novel Approach of Improving Load Balance in Parallel Matrix Factorization for Recommender Systems

引用

IEEE TRANSACTIONS ON COMPUTERS 2021年第5期70卷 789-802页

作者： Guo, Ruixin Zhang, Feng Wang, Lizhe Zhang, Wusheng Lei, Xinya Ranjan, Rajiv Zomaya, Albert Y. China Univ Geosci Sch Comp Sci Wuhan 430074 Peoples R China Tsinghua Univ Dept Comp Sci Beijing 100084 Peoples R China Univ Newcastle Sch Comp Newcastle Upon Tyne NE1 7RU Tyne & Wear England Univ Sydney Sch Comp Sci Camperdown NSW 2006 Australia

A simplified approach to accelerate matrix factorization of big data is to parallelize it. A commonly used method is to divide the matrix into multiple non-intersecting blocks and concurrently calculate them. This operation causes the Load balance problem, which significantly impacts parallel performance and is a big concern. A general belief is that the load balance across blocks is impossible by balancing rows and columns separately. We challenge the belief by proposing an approach of "Balanced partitioning (BaPa)". We demonstrate under what circumstance independently balancing rows and columns can lead to the balanced intersection of rows and columns, why, and how. We formally prove the feasibility of BaPa by observing the variance of rating numbers across blocks, and empirically validate its soundness by applying it to two standard parallel matrix factorization algorithms, DSGD and CCD++. Besides, we establish a mathematical model of "Imbalance Degree" to explain further why BaPa works well. BaPa is applied to synchronous parallel matrix factorization, but as a general load balance solution, it has significant application potential.

关键词： Matrix decomposition Task analysis partitioning algorithms Charge coupled devices Sparse matrices Recommender systems Computer science Matrix factorization stochastic gradient descent recommender systems parallel and distributed computing load balancing

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：