检索结果-内蒙古大学图书馆

IEEE Transactions on Artificial Intelligence 2025年第2期6卷 254-267页

作者： Akbar, Muhammad Azeem Khan, Arif Ali Liang, Peng Lappeenranta-Lahti University of Technology Software Engineering Department Lappeenranta15210 Finland University of Oulu M3S Empirical Software Engineering Research Unit Oulu90014 Finland Wuhan University School of Computer Science Wuhan430072 China

ChatGPT can improve software engineering (SE) research practices by offering efficient, accessible information analysis, and synthesis based on natural language interactions. However, ChatGPT could bring ethical challenges, encompassing plagiarism, privacy, data security, and the risk of generating biased or potentially detrimental data. This research aims to fill the given gap by elaborating on the key elements: motivators, demotivators, and ethical principles of using ChatGPT in SE research. To achieve this objective, we conducted a literature survey, identified the mentioned elements, and presented their relationships by developing a taxonomy. Furthermore, the identified literature-based elements (motivators, demotivators, and ethical principles) were empirically evaluated by conducting a comprehensive questionnaire-based survey involving SE researchers. In addition, we employed an interpretive structure modeling approach to analyze the relationships between the ethical principles of using ChatGPT in SE research and develop a level-based decision model. We further conducted a cross-impact matrix multiplication applied to classification analysis to create a cluster-based decision model. These models aim to help SE researchers devise effective strategies for ethically integrating ChatGPT into SE research by following the identified principles by adopting the motivators and addressing the demotivators. The findings of this study will establish a benchmark for incorporating ChatGPT services in SE research with an emphasis on ethical considerations. © 2020 IEEE.

关键词： software engineering

来源：评论

学校读者我要写书评

暂无评论

Enhancing Large Language Model Inference Efficiency via Lookahead Cache Filtering

Enhancing Large Language Model Inference Efficiency via Look...

引用

2025 IEEE International Conference on Acoustics, Speech, and Signal Processing, ICASSP 2025

作者： Ou, Jie Chen, Yueming Jiang, Shuaihong Tian, Wenhong School of Information and Software Engineering University of Electronic Science and Technology of China Chengdu China

ISBN: (纸本)9798350368741

The large Key-Value (KV) cache is a significant challenge in deploying Large Language Models (LLMs). Current research addressing these issues employs cache compression techniques, which we find suffer from information loss and the "lost-in-the-middle" problem. We propose the Lookahead Cache Filtering (LCF), which retains the full cache in host memory to keep complete details, using the Important Key-Value Lookahead Prediction by Approximate Sorting and the Gather-based Matrix Multiplication on CPU, to filter important KV cache to reduce the overhead of loading cache to GPU and improve the throughput of inference. Finally, we use the Multi-Scale Pyramid Information Fusion mechanism to enhance information fusion to further improve the effectiveness of inference results. Experiments demonstrate that LCF effectively maintains accuracy without introducing additional latency while reducing memory requirements. Our code will be available on GitHub. © 2025 IEEE.

关键词： Approximate Sorting Important Key-Value Lookahead Information Fusion Lost-in-the-Middle Operator Fusion on CPU

来源：评论

学校读者我要写书评

暂无评论

Evaluating the Suitability of Inception Score and Fréchet Inception Distance as Metrics for Quality and Diversity in Image Generation 24

Evaluating the Suitability of Inception Score and Fréchet I...

引用

7th International Conference on computational Intelligence and Intelligent Systems, CIIS 2024

作者： Chan, Derrick Adrian Sithungu, Siphesihle Philezwini Academy of Computer Science and Software Engineering University of Johannesburg Gauteng Johannesburg South Africa

ISBN: (纸本)9798400717437

Variational Autoencoders (VAEs) have gained popularity as one of the main approaches for generating diverse and high-quality synthetic images. This study examines the suitability of evaluation metrics, specifically Inception Score and Fréchet Inception Distance (FID), for assessing these images. Particularly, the study focuses on the generation of synthetic images based on the MNIST handwritten digits dataset. Through the use of VAE-generated MNIST image samples, the study analyses the abovementioned metrics alongside alternative methods that can be used to assess image quality and diversity. The findings made from the study reveal the strengths and limitations of each metric in evaluating image quality and diversity. This paper underscores the need for tailored metrics to enhance the evaluation of generative models, while specifically using the performance of a VAE as the domain of investigation. © 2024 Copyright held by the owner/author(s).

关键词： Fréchet Inception Distance

来源：评论

学校读者我要写书评

暂无评论

More and Larger Auxiliary Feature-Guided Spatial-Temporal Super-Resolution for Rendered Sequences 17th

More and Larger Auxiliary Feature-Guided Spatial-Temporal S...

引用

17th Asian Conference on Computer Vision, ACCV 2024

作者： Zheng, Lijie Liang, Xiao School of Computer Science and Software Engineering Southwest Petroleum University Chengdu610500 China

ISBN: (纸本)9789819609628

The post-processing rendered sequences improves the quality of the sequences and shortens the time of the rendering phase. However, most of the current post-processing methods for sequences are suitable for video. Directly transferring these methods to rendered sequences cannot obtain high-quality results. To address this problem, we propose an end-to-end spatial-temporal super-resolution network for rendered sequences, which improves rendering efficiency by simultaneously implementing frame interpolation (FI) and super-resolution. In the FI task, accurately inferring results that closer to the real motion state of the target frames is the key and difficult point to ensure the generation effect. For this issue, we design an auxiliary feature-guided interpolation (AFGI) module. By introducing the auxiliary features corresponding to the target frames, AFGI module provides the real motion state of the target frames to the network. In the part of aggregating contextual information, we propose a weighted aggregation upsampling (WAUpS) module. Aggregation is selectively performed based on the correlation between incoming information and the current frame. WAUpS module effectively prevents irrelevant information from affecting the super-resolution results, which is a problem with the direct aggregation methods used previously. At the same time, WAUpS module combines the upsampled features with the corresponding high-resolution auxiliary features. This addition provides the output with rich detail textures and other key information, further improving the overall processing quality. Experimental results show that compared with state-of-the-art (SOTA) methods, our method not only obtains high-quality rendered sequences processing results, but also effectively improves the rendering efficiency. © The Author(s), under exclusive license to Springer Nature Singapore Pte Ltd. 2025.

关键词： Rendering (computer graphics)

来源：评论

学校读者我要写书评

暂无评论

Automated Pest Control: Computer Vision for Wildlife Surveillance 24

Automated Pest Control: Computer Vision for Wildlife Surveil...

引用

7th International Conference on computational Intelligence and Intelligent Systems, CIIS 2024

作者： Crossling, Cleveland J. Vadapalli, Hima Van Der Haar, Dustin Academy of Computer Science and Software Engineering University of Johannesburg GP Johannesburg South Africa

ISBN: (纸本)9798400717437

The integration of artificial intelligence in agriculture has revolutionized farming practices, enhancing crop yields and resource efficiency. However, existing machine learning systems primarily focus on livestock, overlooking the impact of wildlife on agricultural productivity. This study addresses the gap by introducing two classification pipelines utilizing VGG16 and ResNet50 models for identifying five major classes of wildlife (foxes, rats, pigeons, raccoons, and squirrels) affecting agricultural sectors. A comprehensive dataset, comprising existing datasets and trail-camera footage, is compiled for training and validation. The results show that models trained on clear images struggle to generalize to real-world environments (the best accuracy being 48.19%). Grey-scaled models perform worse with a 2-6% decrease. Training on the training dataset and evaluating the validation set yields better average results for the weaker-performing model. This study contributes to invasive animal control in agriculture, providing a foundation for effective wildlife management with computer vision. © 2024 Copyright held by the owner/author(s). Publication rights licensed to ACM.

关键词： Livestock

来源：评论

学校读者我要写书评

暂无评论

Anomaly Detection in Retinal Imagery Using Variational Autoencoders 24

Anomaly Detection in Retinal Imagery Using Variational Autoe...

引用

7th International Conference on computational Intelligence and Intelligent Systems, CIIS 2024

作者： Lebaka, Kagiso Blessing Sithungu, Siphesihle Philezwini Academy of Computer Science and Software Engineering University of Johannesburg Gauteng Johannesburg South Africa

ISBN: (纸本)9798400717437

This paper situates itself within the broader medical imaging field, focusing on applying generative modelling techniques for anomaly detection in retinal images. Given the complexity of retinal structures, detecting subtle abnormalities that signal disease requires models that can learn intricate patterns and discern deviations. This study explores the use of a Variational Autoencoder (VAE) to reconstruct healthy retinal images and identify anomalies through reconstruction loss and latent space analysis. With the use of a dataset with an imbalance of healthy and abnormal samples, the model was trained exclusively on healthy images to capture a baseline of normal retinal features. The findings suggest that while the VAE can effectively reconstruct broader structural patterns, it needs to work on finer details, impacting its sensitivity in anomaly detection. These results highlight the challenges inherent in applying generative models to medical diagnostics and emphasise the need for advanced techniques, such as contrastive learning, that may enhance feature extraction. Thus, this paper contributes to the ongoing discourse on the suitability of generative models for medical imaging and supports continued research into more nuanced methods to meet the specific demands of retinal anomaly detection. © 2024 Copyright held by the owner/author(s).

关键词： Ophthalmology

来源：评论

学校读者我要写书评

暂无评论

Uncertainty-aware Correspondence Distillation for Deep Image Clustering

Uncertainty-aware Correspondence Distillation for Deep Image...

引用

2025 IEEE International Conference on Acoustics, Speech, and Signal Processing, ICASSP 2025

作者： Chang, Luyao Chen, Leiting Zhou, Chuan School of Information and Software Engineering University of Electronic Science and Technology of China Chengdu China

ISBN: (纸本)9798350368741

Image clustering is a challenging task in computer vision, with performance heavily dependent on the quality of feature representations due to the inherent complexity of images. However, current image clustering methods overlook the underlying semantic information during representation learning, leading to low-quality feature representations. Moreover, the absence of ground-truth labels amplifies the detrimental effects of unreliable data on semantic guidance, steering the model towards incorrect learning directions. In this work, we propose a novel deep image clustering method named Uncertainty-aware Correspondence Distillation (UCD) to address these issues. Specifically, we introduce the concept of representation correspondence to establish cross-level connections between instances and semantics, which is further employed as a distillation target to improve the network's feature learning by complementing semantic information. To mitigate unnecessary similarity penalties arising from unreliable data, we develop robust dynamic weights for semantic guidance by modeling the uncertainty of image semantics. Extensive experiments on five benchmark datasets demonstrate the superiority of the proposed method. The code is available at https://***/YL616/UCD. © 2025 IEEE.

关键词： Image clustering Knowledge distillation Representation learning Semantic guidance

来源：评论

学校读者我要写书评

暂无评论

MVDC: A Multi-view Dental Completion Model Based on Contrastive Learning

MVDC: A Multi-view Dental Completion Model Based on Contrast...

引用

2025 IEEE International Conference on Acoustics, Speech, and Signal Processing, ICASSP 2025

作者： Yang, Xunyu Deng, Qingxin Huang, Minghan Jiang, Landu Zhang, Dian College of Computer Science and Software Engineering Shenzhen University Guangdong China Guangdong China

ISBN: (纸本)9798350368741

Restoring the patient's occlusal function of broken teeth is a challenging task since tooth texture is very complex, a slight deviation may affect the patient's chewing function and temporomandibular joint function. Therefore, how to efficiently repair the complete shape and real surface of the crown is a critical problem. Traditional technologies are hard to restore complete shape of the dental crown or lack inlay surface details, due to dataset limitations and complexity of missing parts. In this paper, we propose a multi-view crown restoration framework MVDC based on contrastive learning. Specifically, MVDC contains: 1) a multi-view generator with a specially designed loss measurement by using contrastive learning;2) a multi-scale discriminator mechanism able to consider relation and consistency between teeth from different scales;3) an occlusal groove extraction network to extract the occlusal details. We conducted extensive experiments on existing public datasets. The results showcase the superior performance of MVDC. © 2025 IEEE.

关键词： Contrastive learning Multi-view Teeth completion

来源：评论

学校读者我要写书评

暂无评论

Study on the effects of combustion characteristics of pyrotechnic charges on pyrotechnic shocks

引用

Defence Technology(防务技术) 2025年第4期46卷 79-98页

作者： Jingcheng Wang Shihui Xiong Huina Mu Xiaogang Li Yuquan Wen State Key Laboratory of Explosion Science and Safety Protection Beijing Institute of TechnologyBeijing 100081China CAS Key Laboratory of Mechanical Behavior and Design of Materials Department of Modern MechanicsUniversity of Science and Technology of ChinaHefei 230027China

This study calculates the combustion characteristics of various gas-generating and micro gas pyrotechnic charges,including aluminium/potassium perchlorate,boron/potassium nitrate,carbon black/potassium nitrate,and silicon-based delay compositions,using thermodynamic software.A multiphase flowthermal-solid coupling model was established,and the combustion process of the pyrotechnic charges within a closed bomb was *** pyrotechnic shock generated by combustion was *** combustion pressures and pyrotechnic shocks were *** simulation results demonstrated good agreement with experimental ***,the mechanism of shock generation by the combustion of pyrotechnic charges in the closed bomb was *** effects of the combustion characteristics of the pyrotechnic charges on the resulting pyrotechnic shocks were systematically ***,the shock response spectrum of the gas-generating pyrotechnic charges is greater than that of the micro gas compositions at most frequencies,particularly in the mid-field pyrotechnic shocks(3-10 kHz).Furthermore,the pyrotechnic shocks increase approximately linearly with the impulse of the gas-generating pyrotechnic charges.

关键词： Pyrotechnic charges Pyroshock Closed bomb tests Combustion characteristics Multiphase flow-thermal-solid coupling

来源：评论

学校读者我要写书评

暂无评论

Adaptive Model Compression for Efficient Federated Learning in IoT Systems

引用

IEEE Internet of Things Journal 2025年

作者： Zhu, Xi Wang, Junbo Sato, Kento Zheng, Zibin Sun Yat-Sen University School of System Science and Engineering China School of Intelligent Systems Engineering China RIKEN Center for Computational Science Japan School of Software Engineering China

Federated learning (FL), as an emerging collaborative learning paradigm, offers a promising solution to tackle learning challenges in IoT environments. However, the inherent mechanism of FL often results in significant communication overhead, which poses challenges in resource-constrained IoT systems. Moreover, most existing model compression methods struggle to balance model accuracy and compression ratio effectively with low resource consumption: excessive compression degrades accuracy, while insufficient compression incurs high communication overhead. To address this issue, we propose adaMC, an adaptive model compression algorithm for FL in IoT systems. This method integrates two types of network models: main and auxiliary networks. The auxiliary networks generate optimal sparsification strategies to guide the corresponding layers in the main network to become sparse. To ensure robust performance, we develop a bias correction method that guarantees both model accuracy and convergence, accompanied by a theoretical analysis. Experimental results on widely used deep learning models and public datasets demonstrate that our adaMC method is effective, achieving competitive performance compared to FL without compression. © 2014 IEEE.

关键词： Federated learning

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：