检索结果-内蒙古大学图书馆

12th International Conference on Communications and Broadband Networking, ICCBN 2024

作者： Liao, Guixiang Dong, Liwei Mo, Jinrong Zhou, Yanxia Chen, Yanli School of Big Data and Computer Science Guizhou Normal University China School of Information Science and Technology Tibet University China

ISBN: (纸本)9798400717109

Event representation in text is basic task for natural language processing. In this paper, an enhanced event representation framework using contrastive learning based on Gaussian embedding (EventGE) is proposed. To make the representation robustly for downstream tasks, Gaussian distribution for each event density embedding is constructed through specific Multi-layer Perceptron (MLP). Also we feed the Gaussian distribution for positive and negative samples to contrastive learning network to gain the accuracy representation by setting the loss function. The experimental results show the accuracy rate is improved significantly compared to the current state-of-the-art methods with some public datasets. © 2024 ACM.

关键词： Contrastive Learning

来源：评论

学校读者我要写书评

暂无评论

ReCSA:a dedicated sort accelerator using ReRAM-based content addressable memory

引用

Frontiers of computer science 2023年第2期17卷 1-13页

作者： Huize LI Hai JIN Long ZHENG Yu HUANG Xiaofei LIAO National Engineering Research Center for Big Data Technology and System Services Computing Technology and System LabClusters and Grid Computing LabSchool of Computer Science and TechnologyHuazhong University of Science and TechnologyWuhan 430074China

With the increasing amount of data,there is an urgent need for efficient sorting algorithms to process large data *** sorting algorithms have attracted much attention because they can take advantage of different hardware's *** the traditional hardware sort accelerators suffer“memory wall”problems since their multiple rounds of data transmission between the memory and the *** this paper,we utilize the in-situ processing ability of the ReRAM crossbar to design a new ReCAM array that can process the matrix-vector multiplication operation and the vector-scalar comparison in the same array *** this designed ReCAM array,we present ReCSA,which is the first dedicated ReCAM-based sort *** hardware designs,we also develop algorithms to maximize memory utilization and minimize memory exchanges to improve sorting *** sorting algorithm in ReCSA can process various data types,such as integer,float,double,and *** also present experiments to evaluate the performance and energy efficiency against the state-of-the-art sort *** experimental results show that ReCSA has 90.92×,46.13×,27.38×,84.57×,and 3.36×speedups against CPU-,GPU-,FPGA-,NDP-,and PIM-based platforms when processing numeric data *** also has 24.82×,32.94×,and 18.22×performance improvement when processing string data sets compared with CPU-,GPU-,and FPGA-based platforms.

关键词： ReCAM parallel sorting architecture design processing-in-memory

来源：评论

学校读者我要写书评

暂无评论

Optimized Automated Stock Trading using DQN and Double DQN

Optimized Automated Stock Trading using DQN and Double DQN

引用

2024 International Conference on Intelligent Algorithms for Computational Intelligence Systems, IACIS 2024

作者： Bharadwaj, Gurudutt S Pratap, David Darapaneni, Narayana Pes University Department of Computer Science Bengaluru India Great Learning Department of Data Science and Machine Learning Bengaluru India

ISBN: (纸本)9798350360660

Stock Portfolio management involves managing the buying, holding and selling decisions for the various stocks in the portfolio. There has been work where Reinforcement Learning (RL) based actor-critic methods like Deep Direct Policy Gradient (DDPG) have been used for asset allocation problems. Here an attempt has been made to use solely critic-based value function methods like Deep Q-network (DQN) and Double DQN for estimating Q-values of market actions. Then, an optimized portfolio management algorithm designed to balance trades across a basket of stocks is designed. Five stocks are chosen of different price ranges from NYSE and NASDAQ stock exchanges. The average cumulative percentage returns provided by DQN was 55% on testing data with an average Maximum Drawdown (MDD) of 2.5%. The same with Double DQN was 71% on testing data with an average MDD of 2.83%. These results were significantly better than the case when a traditional Buy and Hold strategy was to be employed. © 2024 IEEE.

关键词： Reinforcement learning

来源：评论

学校读者我要写书评

暂无评论

FCC: Fast Source Congestion Control for Ultra-Low Latency in data Center Networks 32

FCC: Fast Source Congestion Control for Ultra-Low Latency in...

引用

32nd IEEE/ACM International Symposium on Quality of Service, IWQoS 2024

作者： Chen, Haodong Zhou, Jiajia Chen, Xinyang Cai, Jiayi Zhu, Longlong Liu, Hongyan Zhang, Dong Wu, Chunming College of Computer and Data Science Fuzhou University China College of Computer Science and Technology Zhejiang University China Zhicheng College Fuzhou University Fuzhou China

data center networks will elevate link bandwidth to over 200Gbps to meet the performance demands of distributed ML and NVMe. That means more traffic will be transmitted within one RTT, making the workload more burst. The challenge posed to congestion control (CC) is that there is barely enough time to make correct decisions. Sub-RTT, a method for switches to proactively generate congestion signals, reduces feedback delays, enhancing CC to handle burst traffic. However, Sub-RTT increases the sensitivity of the sender to queuing in switch buffers, thus leading to issues such as overreactions and link oscillations. We identified that utilizing fine-grained congestion signals and refining congestion stages to adjust congestion control decisions accordingly is key to solving the overreaction problem. In this paper, we propose FCC, a CC algorithm capable of swiftly addressing congestion with Sub-RTT while preventing overreactions in burst workload, thus achieving ultra-low latency and maintaining high link throughput. FCC utilizes programmable data planes to gather fine-grained congestion information through in-band network telemetry (INT), deploys bitmap-based snapshots on switches for real-time congestion stage identification, and dynamically allocates decisions. Extensive simulation experiments indicate that, compared to HPCC, FCC reduces the peak buffer occupancy on bottleneck switches by a factor of 3.2. In large-scale cluster simulations, FCC, when compared to HPCC and DCQCN, respectively, reduces the 99-th-percentile flow completion time of short flows by 39.2% and 128.2%. © 2024 IEEE.

关键词： Time switches

来源：评论

学校读者我要写书评

暂无评论

Ensemble of deep learning models with Walrus Optimization Algorithm for accurate botnet recognition and classification

引用

Iran Journal of computer science 2025年 1-18页

作者： Chacko, Ashwathy Anda Edwin, E. Bijolin Thanka, M. Roshni Ebenezer, V. Division of Computer Science and Engineering Karunya Institute of Technology and Sciences Coimbatore India Division of Data Science and Cyber Security Karunya Institute of Technology and Sciences Coimbatore India

The rapid growth of Internet of Things (IoT) networks has introduced significant security challenges, with botnet attacks being one of the most prevalent threats. These attacks exploit vulnerabilities in IoT devices, leading to severe disruptions and damage to critical infrastructures. Detecting botnet attacks in IoT environments is challenging due to the large volume of data, the dynamic nature of traffic, and the diverse attack patterns. To address these issues, we propose a novel approach called Walrus Optimized Ensemble Deep Learning for Anomaly-Based Recognition Classifier (WOAEDL-ABRC), which leverages a combination of advanced machine learning techniques for effective botnet detection. The methodology of this research involves four key components: (1) data preprocessing through min–max normalization to scale the features appropriately, (2) feature selection using the social cooperation search algorithm (SCSA) to identify the most informative attributes, (3) an ensemble deep learning model combining convolutional autoencoder (CAE), bidirectional gated recurrent unit (BiGRU), and deep belief network (DBN) for robust anomaly detection, and (4) hyperparameter optimization using the Walrus Optimization Algorithm (WAOA), which fine-tunes the model parameters for optimal performance. This ensemble approach ensures that the model benefits from the strengths of each individual technique while mitigating the weaknesses of others. The dataset used for this research includes network traffic data from IoT environments, consisting of various botnet attack scenarios and normal traffic patterns. The data undergoes extensive preprocessing and feature selection to reduce dimensionality and enhance the model’s performance. The implementation is carried out in Python using TensorFlow for deep learning, with the WAOA applied to optimize hyperparameters. The results demonstrate the effectiveness of the WOAEDL-ABRC in detecting botnet attacks, achieving superior accuracy, precision

关键词： Botnet recognition Deep learning Feature selection Internet of things Walrus Optimization Algorithm

来源：评论

学校读者我要写书评

暂无评论

Information Flows during Constraint Satisfaction Search with Dynamic Heuristics: Information from Preprocessing 1

Information Flows during Constraint Satisfaction Search with...

引用

1st International Workshop on Artificial Intelligence for Climate Change, 12th Italian Workshop on Planning and Scheduling, 31st RCRA Workshop on Experimental Evaluation of Algorithms for Solving Problems with Combinatorial Explosion, and SPIRIT Workshop on Strategies, Prediction, Interaction, and Reasoning in Italy, AI4CC-IPS-RCRA-SPIRIT 2024

作者： Wallace, Richard J. Insight Centre for Data Analytics School of Computer Science and IT University College Cork Ireland

We still do not have an adequate understanding of heuristic methods used for solving constraint satisfaction problems (CSPs). An example of this involves the effects of preprocessing, an essential means of improving CSP search. The traditional explanation for its beneficial effect is that the resulting "problem reduction" leaves fewer possibilities to explore. Recently, however, it was shown that when dynamic variable ordering heuristics are used, other factors related to domain size reduction are much more important. This paper extends this earlier work and clarifies what is involved. The key idea is that the pattern of domain reductions produced by preprocessing constitutes information that by itself enables heuristics to make better decisions. Treating domain reduction as a code and the set of reductions as a message of length n transmitted to the search algorithm, we can distinguish two factors, message discriminability and code quality. We provide evidence that greater problem reduction enhances both. © 2024 Copyright for this paper by its authors.

关键词： Constraint satisfaction problems

来源：评论

学校读者我要写书评

暂无评论

Optimizing Supply Chain Operations in the Electronics Industry Using Machine Learning and Integer Linear Programming 3rd

Optimizing Supply Chain Operations in the Electronics Indust...

引用

3rd International Conference on Advanced Computing and Intelligent Technologies, ICACIT 2023

作者： Padhi, Spandan Marken, Gagandeep Data Science Department of Mathematics Chandigarh University Gharuan India Department of Computer Science Engineering Chandigarh University Gharuan India

ISBN: (纸本)9789819719600

Efficient operations, within the supply chain are vital for the electronics industry. In this study, we focus on optimizing the supply chain network of a hypothetical company ABC Electronics, an entity crafted for the purposes of this research paper. The company faces challenges related to production, inventory management, and transportation logistics. To address these challenges we employ machine learning (ML) and integer linear programming (ILP) techniques. Our primary objectives include enhancing demand predictions through ML, optimizing production schedules and inventory using ILP, and devising strategies for transportation logistics. Our research makes a contribution toward reducing costs and improving efficiency at ABC Electronics enabling them to stay competitive in the electronics market. Furthermore, our findings have implications, for industry standards, sustainability practices, and customer satisfaction. © The Author(s), under exclusive license to Springer Nature Singapore Pte Ltd. 2024.

关键词： Machine learning

来源：评论

学校读者我要写书评

暂无评论

Anomaly Detection in Network Traffic data Using Deep Learning 2

Anomaly Detection in Network Traffic Data Using Deep Learnin...

引用

2nd International Conference on Computing and data Analytics, ICCDA 2024

作者： Ramya, S. Selvi, S.Annal Ezhil Bishop Heber Collge Department of Data Science Thiruchirappalli India Bishop Heber College Department of Computer Science Thiruchirappalli India

ISBN: (纸本)9798331542573

In the Tech Renaissance, spotting network traffic anomalies has become a game changer for security. With the rapid growth of network traffic and the increasing frequency of cyberattacks, detecting anomalies and intrusions in real-time is critical. The growing complexity and volume of modern traffic patterns have rendered traditional security measures inadequate for identifying sophisticated cyber threats. This research aims to detect unusual behaviors in network traffic that could signal security breaches, misuse, or attacks. Current detection methods, like rule-based intrusion detection systems (IDS) and signature-based approaches, depend on known threat signatures and predefined rules. These methods are limited in detecting new or unknown threats, prone to high false positive rates, and lack the adaptability needed to respond to evolving threats. To address these limitations, this paper proposes the use of advanced machine learning and deep learning techniques, such as autoencoders, RNNs, CNNs, and GANs, which can learn from evolving traffic patterns and improve detection accuracy. By adapting to both known and unknown anomalies, these models significantly reduce false positives and enhance the overall security of network infrastructures. Organizations and businesses will benefit from improved detection rates, a reduced risk of cyberattacks, and enhanced protection of sensitive data, leading to a more secure and resilient network environment. The proposed models anticipate a notable reduction in false positives and an increase in anomaly detection accuracy, making them a more reliable solution than traditional methods. © 2024 IEEE.

关键词： Network intrusion

来源：评论

学校读者我要写书评

暂无评论

SMMGCL: A novel multi-level graph contrastive learning framework for integrating spatial multi-omics data

SMMGCL: A novel multi-level graph contrastive learning frame...

引用

2024 IEEE International Conference on Bioinformatics and Biomedicine, BIBM 2024

作者： Wang, Bo Liu, Wei Luo, Jiawei Chen, Xiangtao Kwoh, Chee Keong Hunan University College of Computer Science and Electronic Engineering China Nanyang Technological University College of Computing and Data Science Singapore

ISBN: (纸本)9798350386226

Recent advances in spatial omics technologies have allowed various omics data to be obtained from a single tissue section. To fully explore the relationships among these different types of omics data, it is urgent to develop more effective methods for spatial multi-omics data integration. In this work, we propose a novel Multi-level Graph Contrastive Learning framework, named SMMGCL, to simultaneously mine complementary information at both spot and graph levels for integrating Spatial Multi-omics data. Specifically, to adaptively fuse multi-omics modalities, we first design a multi-modality autoencoder that integrates spatial locations with spot omic expressions to extract modality-specific embeddings. These embeddings are then fused into a consensus representation using an attention mechanism to capture spot-level cross-omics representations. Next, to explore the complex inter-omic structural information, we connect corresponding spots across different omics adjacency graphs into a heterogeneous graph. We then employ a graph convolutional network (GCN) to extract spatial correlations across the omics, learning a graph-level cross-omics global representation. Finally, SMMGCL aligns feature similarity matrixes between spot-level and graph-level representations with their pseudo-label similarity matrix, ensuring multi-level clustering consistency and leading to more accurate spatial multi-omics integration. Experimental results on simulated and real datasets from across tissues show that SMMGCL consistently outperforms other state-of-the-art methods in spatial multi-omics integration performance. The code for SMMGCL is available for download from the GitHub repository at https://***/cs-wangbo/SMMGCL. © 2024 IEEE.

关键词： Contrastive Learning

来源：评论

学校读者我要写书评

暂无评论

Toward data Efficient Model Merging between Different datasets without Performance Degradation 16

Toward Data Efficient Model Merging between Different Datase...

引用

16th Asian Conference on Machine Learning, ACML 2024

作者： Yamada, Masanori Yamashita, Tomoya Yamaguchi, Shin’ya Chijiwa, Daiki NTT Social Informatics Laboratories Japan NTT Computer and Data Science Laboratories Japan

Model merging is attracting attention as a novel method for creating a new model by combining the weights of different trained models. While previous studies reported that model merging works well for models trained on a single dataset with different random seeds, model merging between different datasets remains unsolved. In this paper, we attempt to reveal the difficulty in merging such models trained on different datasets and alleviate it. Our empirical analyses show that, in contrast to the single-dataset scenarios, dataset information needs to be accessed to achieve high accuracy when merging models trained on different datasets. However, the requirement to use full datasets not only incurs significant computational costs but also becomes a major limitation when integrating models developed and shared by others. To address this, we demonstrate that dataset reduction techniques, such as coreset selection and dataset condensation, effectively reduce the data requirement for model merging. In our experiments with SPLIT-CIFAR10 model merging, the accuracy is significantly improved by 31% when using the full dataset and 24% when using the sampled subset compared with not using the dataset. Our code is available at this URL. © 2024 M. Yamada, T. Yamashita, S. Yamaguchi & D. Chijiwa.

关键词： data accuracy

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：