检索结果-内蒙古大学图书馆

11th Joint International Information Technology and Artificial Intelligence Conference, ITAIC 2023

作者： Tian, Fuwen Xu, Meng College of Computer Science and Software Engineering Shenzhen University Shenzhen China

ISBN: (纸本)9798350333664

Column-oriented databases have emerged as effective solutions for handling massive amounts of data, and data compression plays a crucial role. Attribute columns are divided into blocks and stored in separate files, and records within the same attribute column often exhibit high similarity, providing favorable conditions for data compression. Researchers have proposed numerous compression algorithms, but these algorithms can only achieve optimal compression results when applied to data with specific features. Therefore, it is crucial to select the compression algorithm that performs best based on the features of the data. We propose a novel method for selecting compression algorithms using Stacking, a technique that effectively enhances the prediction ability of classification models in machine learning. Our experimental results demonstrate that the stacked model achieves an accuracy of 92.9%, surpassing other models in terms of F1-Score and Kappa coefficient. Furthermore, our compression algorithm selection method outperforms other existing methods in terms of compression performance. © 2023 IEEE.

关键词： classification model column-oriented database data compression Stacking

来源：评论

学校读者我要写书评

暂无评论

A Novel Automatic Prompt Tuning Method for Polyp Segmentation

A Novel Automatic Prompt Tuning Method for Polyp Segmentatio...

引用

2024 IEEE International Conference on Bioinformatics and Biomedicine, BIBM 2024

作者： Jing, Xin Zhou, Heyang Mao, Keming Zhao, Yuhai Chu, Liangyu Northeastern University Software College Shenyang China Northeastern University College of Computer Science and Engineering Shenyang China Northeastern University College of Medicine and Biological and Information Engineering Shenyang China

ISBN: (纸本)9798350386226

Accurate polyp segmentation of colonoscopy images is crucial for diagnosing and treating colorectal cancer. A new type of method for this challenging task has been introduced with the emergence of large vision models like the Segment Anything Model (SAM). However, the application of SAM to medical image segmentation falls short of clinical requirements. In efforts to tailor SAM for medical imaging tasks, researchers often work on improving adapters and prompts separately, over-looking the potential for synergy between them. Furthermore, existing methods still do not overcome the influence of prompt type, quantity, and positioning on segmentation results. SAM's dependency on professionally annotated prompts further limits its utility. To tackle these issues, we propose a fine-tuning strategy named Automatic Prompt Tuning (APT), which aims to generate prompts automatically. This approach begins with adapting SAM for polyp segmentation by incorporating lightweight adapters into the image encoder. Next, we introduce an automatic prompt generator to process features and produce relevant prompts. Finally, we ensure the mask decoder consistently delivers accurate segmentation by leveraging these automatically generated prompts. Extensive experiments on three major polyp segmentation datasets demonstrate that APT outperforms other state-of-the-art approaches, proving its effectiveness. © 2024 IEEE.

关键词： Medical imaging

来源：评论

学校读者我要写书评

暂无评论

Adaptive Lazily Aggregation Based on Error Accumulation 4

Adaptive Lazily Aggregation Based on Error Accumulation

引用

4th International Conference on Electronic Communication and Artificial Intelligence, ICECAI 2023

作者： Chen, Xiaofeng Liu, Gang Shenzhen University College of Computer Science and Software Engineering Shenzhen China

ISBN: (纸本)9798350314496

Federated Learning (FL) enables multiple clients to collaboratively train models without exposing their local data. FL is an effective approach to utilizing localized data while preserving clients' data privacy, but it also brings significant communication overhead. To reduce communication overhead of FL, this paper proposes the Adaptive Lazily Aggregation based on Error Accumulation (EA-ALA) algorithm. It uses adaptive constraints to determine whether a client can skip a communication round with the server so as to diminish communication cost. It also adopts error accumulation to improve model accuracy. The experimental results on CIFAR10 and Fashion-MNIST datasets show that compared to vanilla FL, EA-ALA consumes only 52% and 61% of communication rounds to achieve higher model accuracy. © 2023 IEEE.

关键词： Errors

来源：评论

学校读者我要写书评

暂无评论

Deep Hierarchy Quantization Compression algorithm based on Dynamic Sampling 4

Deep Hierarchy Quantization Compression algorithm based on D...

引用

4th International Conference on Electronic Communication and Artificial Intelligence, ICECAI 2023

作者： Jiang, Wan Liu, Gang Shenzhen University College of Computer Science and Software Engineering Shenzhen China

ISBN: (纸本)9798350314496

Federated learning allows multiple parties to jointly train deep learning models without the need for any participants to reveal their private data to a centralized server. However, this form of privacy-preserving collaborative learning has resulted in significant communication costs during the training period. To address this problem, we propose the Deep Hierarchical Quantization Compression (DHQC) algorithm. DHQC reduces the communication cost by using gradient sparsification and multi-bit quantization, and improves the accuracy and convergence speed of the model by using a dynamic sampling strategy. Experimental results on the CIFARIO and CIFARIOO datasets demonstrate that compared with the uncompressed baseline, DHQC can achieve a compression rate of 128x, significantly reducing the communication costs. © 2023 IEEE.

关键词： Deep learning

来源：评论

学校读者我要写书评

暂无评论

Truthful Double Auction-Based Resource Allocation Mechanisms for Latency-Sensitive Applications in Edge Clouds 18th

Truthful Double Auction-Based Resource Allocation Mechanisms...

引用

18th International Conference on Wireless Artificial Intelligent Computing Systems and Applications, WASA 2024

作者： Wu, Dongkuo Wang, Xueyi Wang, Xingwei Huang, Min Wang, Zhitong College of Computer Science and Engineering Northeastern University Liaoning Shenyang China College of Software Northeastern University Liaoning Shenyang China College of Information Science and Engineering Northeastern University Liaoning Shenyang China

ISBN: (纸本)9783031714696

With the rapid advancement of wireless networks, edge computing has emerged as a promising paradigm for providing computing services to nearby latency-sensitive applications. Toward this trend, resource trading markets among mobile users (MUs) and edge clouds are emerging which need well-designed allocation and pricing mechanisms. In this paper, we propose a double auction-based multi-market resource trading framework (DAMRT), which integrates the dynamic programming and padding-based double auction methods, aiming to achieve approximate social welfare maximization and guarantee the properties of truthfulness and budget balance in the constrained edge computing system. To be specific, for admission control among different base station (BS) central markets, a dynamic programming-based method is proposed to obtain the optimal BS-MU association set. For each market, a truthful padding-based double auction resource allocation mechanism (TPDA) is proposed, which uses a linear programming-based padding method and a binary search algorithm to obtain the near-optimal allocation solution, and leverages a critical-value and a VCG-based pricing strategy for winning MUs and service providers. Our theoretical analysis proves that TPDA achieves truthfulness, individual rationality, and budget balance. Furthermore, simulation results verify the effectiveness of DAMRT. © The Author(s), under exclusive license to Springer Nature Switzerland AG 2025.

关键词： Budget control

来源：评论

学校读者我要写书评

暂无评论

Learning Contention Window Selection in Age of Information-Oriented IEEE 802.11 Networks 11

Learning Contention Window Selection in Age of Information-O...

引用

2023 IEEE 11th International Conference on Information, Communication and Networks, ICICN 2023

作者： Li, Jinhua Jian, Feng Shenzhen University College of Computer Science and Software Engineering Shenzhen China

ISBN: (纸本)9798350314014

We study the problem of selecting the contention window (CW) for age of information (AoI)-oriented IEEE 802.11 networks using deep reinforcement learning (DRL) techniques. AoI quantifies information freshness and is defined as the time elapsed since the generation of the latest received information update. As the number of stations increases, conventional CW selection methods defined in IEEE 802.11 standards, such as binary exponential backoff (BEB), fail to achieve a low network-wide average AoI. Therefore, we leverage DRL and propose a CW control algorithm based on deep deterministic policy gradient (DDPG) that can dynamically learn appropriate CW values under different network conditions. Simulation results demonstrate that our proposed DRL-based approach reduces the average AoI of the system by about 57.6% in a congested network with 50 stations compared to the conventional BEB algorithm. Furthermore, compared to the algorithm using throughput as the DRL reward function, our AoI-reward-function approach achieves a 20.4% improvement in information freshness, indicating the key difference between AoI and throughput. © 2023 IEEE.

关键词： Reinforcement learning

来源：评论

学校读者我要写书评

暂无评论

Social NSTransformers: Low-Quality Pedestrian Trajectory Prediction

IEEE Transactions on Artificial Intelligence

引用

IEEE Transactions on Artificial Intelligence 2024年第11期5卷 5575-5588页

作者： Jiang, Zihan Ma, Yiqun Shi, Bingyu Lu, Xin Xing, Jian Goncalves, Nuno Jin, Bo Suzhou215000 China Northeast Forestry University College of Computer and Control Engineering Harbin150000 China University of Coimbra Institute of Systems and Robotics Department of Electrical and Computer Engineering Coimbra3030-290 Portugal

This article introduces a novel model for low-quality pedestrian trajectory prediction, the social nonstationary transformers (NSTransformers), that merges the strengths of NSTransformers and spatiotemporal graph transformer (STAR). The model can capture social interaction cues among pedestrians and integrate features across spatial and temporal dimensions to enhance the precision and resilience of trajectory predictions. We also propose an enhanced loss function that combines diversity loss with logarithmic root mean squared error (log-RMSE) to guarantee the reasonableness and diversity of the generated trajectories. This design adapts well to complex pedestrian interaction scenarios, thereby improving the reliability and accuracy of trajectory prediction. Furthermore, we integrate a generative adversarial network (GAN) to model the randomness inherent in pedestrian trajectories. Compared to the conventional standard Gaussian distribution, our GAN approach better simulates the intricate distribution found in pedestrian trajectories, enhancing the trajectory prediction's diversity and robustness. Experimental results reveal that our model outperforms several state-of-the-art methods. This research opens the avenue for future exploration in low-quality pedestrian trajectory prediction. © 2024 IEEE.

关键词： Stars

来源：评论

学校读者我要写书评

暂无评论

Dirty page prediction by machine learning methods based on temporal and spatial locality 26

Dirty page prediction by machine learning methods based on t...

引用

26th International Conference on computer Supported Cooperative Work in Design, CSCWD 2023

作者： Lu, Yahui Jiang, Yu Shenzhen University College of Computer Science and Software Engineering Shenzhen China

ISBN: (纸本)9798350331684

The memory dirty page prediction technology can effectively predict whether a memory page will be modified (dirty) at the next moment, and is widely used in virtual machine migration, container migration and other fields. In this paper, we propose a machine-learning based method for memory dirty page prediction. The method exploits the temporal and spatial locality principle of memory changes, collects dirty records of pages over a period of time, and uses supervised learning methods for training and predicting the dirty page. We also discuss the influence of data contradiction and data repetition in memory page dataset. The experiments with different memory change frequency dataset show that compared with the traditional time series methods, our machine-learning based method has better performance. © 2023 IEEE.

关键词： Forecasting

来源：评论

学校读者我要写书评

暂无评论

Dynamic Constrained Multi-Objective Evolutionary Optimization via Adaptive Two-Stage Archiving and Autoencoder Prediction 5

Dynamic Constrained Multi-Objective Evolutionary Optimizatio...

引用

5th International Conference on Data-Driven Optimization of Complex Systems, DOCS 2023

作者： Wang, Qianhui Ye, Yulong Zhu, Qingling Liu, Songbai Lin, Qiuzhen College of Computer Science and Software Engineering Shenzhen University Shenzhen China

ISBN: (纸本)9798350393521

Dynamic constrained multi-objective optimization problems (DCMOPs) are characterized by time-varying objectives and constraints, requiring optimization algorithms that can rapidly track the changing Pareto-Optimal Set (POS).A new dynamic constrained multi-objective evolutionary algorithm with adaptive two-stage archiving and autoencoder prediction is proposed in this paper, called ATAP to effectively solve DCMOPs. Specifically, ATAP designs a differential denoising autoencoder (DDA) prediction strategy, which applies a denoising autoencoder to thoroughly analyze change trends of historical population and predict some initial solutions in the new environment. Subsequently, to promote greater diversity within the initial population, a differential rule is implemented, effectively addressing the potential scarcity of diversity caused by constraints. Moreover, ATAP introduces an adaptive two-stage archiving (ATA) constraint handling technique, which can dynamically adjust evolutionary stages based on the state of the population. This approach can adaptively determine preferences between objectives and constraints, achieving a better balance. In this way, ATA and DDA are well cooperated to efficiently solve DCMOPs with time-varying constraints and objectives. The experimental results demonstrate that the proposed ATAP is effective and has some advantages over three competitive algorithms when solving the CEC2023 DCMOP benchmark. © 2023 IEEE.

关键词： Forecasting

来源：评论

学校读者我要写书评

暂无评论

A Neocognitron Based on Multi-objective Optimization for Few-Shot Learning 1

引用

6th EAI International Conference on Robotics and Networks, EAI ROSENET 2022

作者： Pan, Hao Guo, Guiyu Shi, Daming College of Computer Science and Software Engineering Shenzhen University Shenzhen China

ISBN: (数字)9783031338267

ISBN: (纸本)9783031338250

In recent years, with the continuous development of deep learning, more and more network models have been proposed to solve practical problems. However, most models often need a large number of labeled samples to train the network to achieve better recognition results. Because a large number of samples need to be labeled, it is labor-intensive and prone to human error labels. For this reason, a large number of researchers have created the concept of few-shot learning, which entails that the model acquires new information from a limited amount of labeled samples in order to get improved recognition outcomes. In this chapter, we learn from the structure of the visual cortex of the human brain and propose new shape templates to improve the traditional neural network. First, we extract the global characteristics of the image to form a 0-1 mask. As the network hierarchy increases and the field of perception expands, the mask is cut to form shape templates. Then evolutionary computation selects some excellent shape templates, which provide a non-linear constraint for the convolution layer. Moreover, the convolution layer introduces inhibition signals for training, and a new function is proposed to make it more in line with the biomimetic principles of the human visual cortex. The results show that the proposed method can improve the accuracy of few-shot learning. © 2023, The Author(s), under exclusive license to Springer Nature Switzerland AG.

关键词： Multiobjective optimization

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：