检索结果-内蒙古大学图书馆

Enhancing text understanding of decoder-based model by leveraging parameter-efficient fine-tuning method

Neural Computing and Applications 2025年第9期37卷 6899-6913页

作者： Feroze, Wasif Cheng, Shaohuan Jimale, Elias Lemuye Jakhro, Abdul Naveed Qu, Hong School of Computer Science and Engineering University of Electronic Science and Technology of China Chengdu China Department of Information Technology Shaheed Benazir Bhutto University Naushahro Feroze Campus Naushahro Feroze Pakistan

Machine reading comprehension (MRC) is a fundamental natural language understanding task in natural language processing, which aims to comprehend the text of a given passage and answer questions based on it. Understanding implicit information, deducing the logical structure of information, and connecting context from different pieces of information make the MRC task difficult. Most current state-of-the-art approaches for MRC are using encoder-based models. However, no earlier research proposed a decoder-only model for MRC question-answering datasets, although language models based on this category achieved unprecedented performance in different generative tasks. In this paper, we propose a parameter-efficient fine-tuning framework that effectively increases MRC capabilities on decoder-only large language models. This framework designs the process for MRC and introduces the low-rank adaptation (LoRA) method to effectively fine-tune the large model with many parameters, even with lower hardware resource requirements than the previous methods. In addition, we also integrate a quantized model inference strategy for the fine-tuned model to improve practicability further. We conducted experiments on four types of MRC datasets. After extensive experiments, our results show that our model achieved a significant performance boost over baselines and outperformed other strong models for MRC. © The Author(s), under exclusive licence to Springer-Verlag London Ltd., part of Springer Nature 2025.

关键词： Decoding

来源：评论

学校读者我要写书评

暂无评论

How graph convolutions amplify popularity bias for recommendation?

引用

Frontiers of computer Science 2024年第5期18卷 121-132页

作者： Jiajia CHEN Jiancan WU Jiawei CHEN Xin XIN Yong LI Xiangnan HE School of Information Science and Technology University of Science and Technology of ChinaHefei 230026China School of Computer Science and Technology Zhejiang UniversityHangzhou 310058China School of Computer Science and Technology Shandong UniversityQingdao 250100China Department of Electronic Engineering Tsinghua UniversityBeijing 100084China

Graphconvolutional networks(GCNs)have become prevalent in recommender system(RS)due to their superiority in modeling collaborative *** improving the overall accuracy,GCNs unfortunately amplify popularity bias-tail items are less likely to be *** effect prevents the GCN-based RS from making precise and fair recommendations,decreasing the effectiveness of recommender systems in the long *** this paper,we investigate how graph convolutions amplify the popularity bias in *** theoretical analyses,we identify two fundamental factors:(1)with graph convolution(i.e.,neighborhood aggregation),popular items exert larger influence than tail items on neighbor users,making the users move towards popular items in the representation space;(2)after multiple times of graph convolution,popular items would affect more high-order neighbors and become more *** two points make popular items get closer to almost users and thus being recommended more *** rectify this,we propose to estimate the amplified effect of popular nodes on each node's representation,and intervene the effect after each graph ***,we adopt clustering to discover highly-influential nodes and estimate the amplification effect of each node,then remove the effect from the node embeddings at each graph convolution *** method is simple and generic-it can be used in the inference stage to correct existing models rather than training a new model from scratch,and can be applied to various GCN *** demonstrate our method on two representative GCN backbones LightGCN and UltraGCN,verifying its ability in improving the recommendations of tail items without sacrificing the performance of popular *** are open-sourced^(1)).

关键词： recommendation graph convolution networks popularity bias

来源：评论

学校读者我要写书评

暂无评论

Breast-NET: a lightweight DCNN model for breast cancer detection and grading using histological samples

引用

Neural Computing and Applications 2024年第32期36卷 20067-20087页

作者： Saha, Mousumi Chakraborty, Mainak Maiti, Suchismita Das, Deepanwita Department of Computer Science and Engineering National Institute of Technology West Bengal Durgapur713209 India School of Innovation Design and Engineering Division of Intelligent Future Technologies Malardalen University Västerås Sweden Department of Information Technology Narula Institute of Technology West Bengal Kolkata700109 India Department of Computer Science and Engineering Narula Institute of Technology West Bengal Kolkata700109 India

Breast cancer is a prevalent and highly lethal cancer affecting women globally. While non-invasive techniques like ultrasound and mammogram are used for diagnosis, histological examination after biopsy is considered the gold standard. However, manual examination of tissues for abnormality is labor-intensive, expensive, and requires prior domain knowledge. Early detection, awareness, and access to specialized medical infrastructure in resource-constrained and remote areas are significant challenges but crucial for saving lives. In recent years, deep learning-based approaches have shown promising results in breast cancer detection, facilitated by advancements in GPU memory, computation power, and the availability of digital data. Motivated by these observations, we propose the Breast-NET deep convolutional neural network model for breast cancer detection and grading using histological images. Our model’s performance is evaluated on the BreakHis dataset, and we demonstrate its generalization ability on the Invasive Ductal Carcinoma (IDC) grading and IDC datasets. Extensive experimental and statistical performance analysis, along with an ablation study, validates the efficiency of our proposed model. Furthermore, we demonstrate the effectiveness of transfer learning with seven pre-trained convolutional neural networks for breast cancer detection and grading. Experimental results show that our framework outperforms state-of-the-art approaches in terms of accuracy, space, and computational complexity for the BreakHis, IDC grading, and IDC datasets. © The Author(s), under exclusive licence to Springer-Verlag London Ltd., part of Springer Nature 2024.

关键词： Grading

来源：评论

学校读者我要写书评

暂无评论

A Collaborative Approach for Secured Routing in Mobile Ad-Hoc Network

引用

Intelligent Automation & Soft Computing 2023年第2期35卷 1337-1351页

作者： W.Gracy Theresa A.Gayathri P.Rama Department of Computer Science and Engineering Panimalar Institute of TechnologyChennai600123Tamil NaduIndia Department of Information Technology Vellore Institute of TechnologyVellore632014Tamil NaduIndia Department of Computer Science and Engineering Loyola Institute of TechnologyChennai600123Tamil NaduIndia

Mobile computing is the most powerful application for network com-munication and connectivity,given recent breakthroughs in theﬁeld of wireless networks or Mobile Ad-hoc networks(MANETs).There are several obstacles that effective networks confront and the networks must be able to transport data from one system to another with adequate *** most applications,a frame-work must ensure that the retrieved data reﬂects the transmitted *** driv-ing to other nodes,if the frame between the two nodes is deformed in the data-link layer,it must be *** link-layer protocols immediately disregard the frame and enable the high-layer protocols to transmit it *** other words,because of asset information must be secured from threats,information is a valu-able *** MANETs,some applications necessitate the use of a network method for detecting and blocking these *** a secure intrusion detection system in the network,which provides security to the nodes and route paths in the network,is a major difﬁculty in *** on the network can jeopardize security issues discovered by the intrusion detection system engine,which are then blocked by the network’s intrusion prevention *** bringing the Secure Intrusion Detection System(S-IDS)into the network,a new technique for implementing security goals and preventing attacks will be *** Secure Energy Routing(SER)protocol for MANETs is introduced in this *** protocol addresses the issue of network security by detecting and preventing attacks in the *** data transmission in the MANET is forwarded using Elliptical Curve Cryptography(ECC)with an objective to improve the level of *** Simulator–2 is used to simulate the network and experiments are compared with existing methods.

关键词： Mobile ad-hoc network(MANET) intrusion detection system secure energy routing(SER) elliptical curve cryptography(ECC) security

来源：评论

学校读者我要写书评

暂无评论

VivID: A Visually Improved GIF Encoding Network Design

引用

IEEE Transactions on Circuits and Systems for Video technology 2025年第6期35卷 6101-6113页

作者： Wang, Yifei Liu, Gaozhi Zhu, Zhiying Zhang, Xinpeng Qian, Zhenxing Fudan University Department of School of Computer Science Shanghai200433 China East China University of Science and Technology Department of China School of Information Science and Engineering Shanghai200237 China

Graphics Interchange Format (GIF) encoding is the art of reproducing an image with limited colors. Existing GIF encoding schemes often introduce unpleasant visual artifacts such as banding artifact, dotted-pattern noise and color shift, especially when the palette size is small. To address the issues above, we propose VivID, a Visually Improved GIF Encoding Network Design, which is compatible with exiting GIF decoders. VivID consists of three modules and two of them provide the functionality within the GIF encoding pipeline. Firstly, in order to reduce the color shift introduced by color quantization, we design the multi-palette extractor to create a GIF image with minimal distortion by extracting a near-optimal palette. This module can significantly improve the image fidelity and gains adaptability to multiple palette sizes after only one-time training. Furthermore, to reduce banding artifact and the dotted-pattern noise caused by dithering process, we propose banding remover which can randomize quantization error to neighbourhood by utilizing a learnable dithering pattern. Moreover, to further eliminate the banding artifacts, we design the banding scorer module, which is a novel metric for evaluating banding artifact and it correlates well with subjective perception. We adopt it as a customized loss for training dithering module. Extensive experiments across various aspects demonstrate that VivID produces visually pleasing results even when the palette size is extremely small, outperforming both traditional and existing learning based GIF encoding methods. © 1991-2012 IEEE.

关键词： Encoding (symbols)

来源：评论

学校读者我要写书评

暂无评论

Face image authentication scheme based on MTCNN and SLT

引用

Multimedia Tools and Applications 2025年 1-43页

作者： Thabit, Rasha Al-Askari, Mohanad A. Mohammed, Dunya Zeki Anaam, Elham Abdulwahab Mahmood, Zainab H. Jabbar, Dina Jamal Salih, Zahraa Aqeel Information System Department College of Computer Sciences and Information Technology University of Al-Anbar Al-Anbar Iraq Computer Engineering Department College of Engineering Al-Iraqia University Baghdad Iraq Electronic and Communications Engineering Department College of Engineering Gilgamesh University Baghdad Iraq Faculty of Computing and Informatics Multimedia University Selangor Cyberjaya Malaysia Department of Cybersecurity Engineering College of Information Engineering Al-Nahrain University Baghdad Iraq Electrical Engineering Technical Collage Middle Technical University Baghdad Iraq Department of Business Administration Alamaal University College Baghdad Iraq

DeepFakes and face image manipulation methods have been widely distributed in the last few years and several techniques have been presented to check the authenticity of the face image and detect manipulation if exists. Most of the available manipulation detection techniques have been successfully applied to reveal one type of manipulation under specific conditions, however, many limitations and challenges can be encountered in this field. To overcome some limitations and challenges, this paper presents a new face image authentication (FIA) scheme based on Multi-Task Cascaded Conventional Neural Networks (MTCNN) and watermarking in Slantlet transform (SLT) domain. The proposed FIA scheme has three main algorithms that are face detection and selection, embedding, and extraction algorithms. Different block sizes have been used to divide the image into non-overlapping blocks followed by classifying them into two groups that are blocks from face area (FA) and blocks from the remaining area (RA) of the image. In the embedding algorithms, the authentication information is generated from FA blocks and embedded in the RA blocks. In the extraction algorithms, the embedded information is extracted from RA blocks and compared with the calculated data from FA blocks to reveal manipulations and localize the manipulated blocks if exist. Extensive experiments have been conducted to evaluate the performance of the proposed FIA scheme for different face images. The experimental work included tests for payload, capacity, visual quality, time complexity, and localization of manipulations. The results proved the efficiency of the proposed scheme in detecting and localizing different face image manipulations such as attributes attacks, retouching attacks, expression swap, face swap, and morphing attacks. The proposed scheme overcomes many limitations and it is 100% accurate in localizing the tampered blocks which makes it a better candidate for practical applications. © The Author(s), un

关键词： Authentication

来源：评论

学校读者我要写书评

暂无评论

Modeling and Predictive Analytics of Breast Cancer Using Ensemble Learning Techniques:An Explainable Artificial Intelligence Approach

引用

computers, Materials & Continua 2024年第12期81卷 4033-4048页

作者： Avi Deb Raha Fatema Jannat Dihan Mrityunjoy Gain Saydul Akbar Murad Apurba Adhikary Md.Bipul Hossain Md.Mehedi Hassan Taher Al-Shehari Nasser A.Alsadhan Mohammed Kadrie Anupam Kumar Bairagi Computer Science and Engineering Discipline Khulna UniversityKhulna9208Bangladesh Information and Communication Engineering Noakhali Science and Technology UniversityNoakhali3814Bangladesh School of Computing Sciences and Computer Engineering University of Southern MississippiHattiesburgMS 39401USA Computer Skills Department of Self-Development SkillCommon First Year DeanshipKing Saud UniversityRiyadh11362Saudi Arabia Computer Science Department College of Computer and Information SciencesKing Saud UniversityRiyadh12372Saudi Arabia

Breast cancer stands as one of the world’s most perilous and formidable diseases,having recently surpassed lung cancer as the most prevalent cancer *** disease arises when cells in the breast undergo unregulated proliferation,resulting in the formation of a tumor that has the capacity to invade surrounding *** is not confined to a specific gender;both men and women can be diagnosed with breast cancer,although it is more frequently observed in *** detection is pivotal in mitigating its mortality *** key to curbing its mortality lies in early ***,it is crucial to explain the black-box machine learning algorithms in this field to gain the trust of medical professionals and *** this study,we experimented with various machine learning models to predict breast cancer using the Wisconsin Breast Cancer Dataset(WBCD)*** applied Random Forest,XGBoost,Support Vector Machine(SVM),Multi-Layer Perceptron(MLP),and Gradient Boost classifiers,with the Random Forest model outperforming the others.A comparison analysis between the two methods was done after performing hyperparameter tuning on each *** analysis showed that the random forest performs better and yields the highest result with 99.46%*** performance evaluation,two Explainable Artificial Intelligence(XAI)methods,SHapley Additive exPlanations(SHAP)and Local Interpretable Model-Agnostic Explanations(LIME),have been utilized to explain the random forest machine learning model.

关键词： Breast cancer prediction machine learning models explainable artificial intelligence random forest hyperparameter tuning

来源：评论

学校读者我要写书评

暂无评论

FUSE: a federated learning and U-shape split learning-based electricity theft detection framework

引用

Science China(information Sciences) 2024年第4期67卷 339-340页

作者： Xuan LI Naiyu WANG Liehuang ZHU Shuai YUAN Zhitao GUAN School of Control and Computer Engineering North China Electric Power University School of Cyberspace Science and Technology Beijing Institute of Technology Department of Finance Operationsand Information Systems(FOIS)Brock University

In recent years, the data-driven electricity theft detection methods integrated with edge cloud computing [1, 2] have not only demonstrated superior detection accuracy but also improved efficiency, making them viable alternatives to indoor inspections. Energy service providers(ESPs) typically manage regions by dividing them into various transformer districts(TDs). The detection of electricity theft in a particular region is performed by the associated TD,

关键词： Sensitive data

来源：评论

学校读者我要写书评

暂无评论

MAUN:Memory-Augmented Deep Unfolding Network for Hyperspectral Image Reconstruction

引用

IEEE/CAA Journal of Automatica Sinica 2024年第5期11卷 1139-1150页

作者： Qian Hu Jiayi Ma Yuan Gao Junjun Jiang Yixuan Yuan Electronic Information School Wuhan UniversityWuhan 430072China School of Computer Science and Technology Harbin Institute of TechnologyHarbin 150001China Department of Electronic Engineering Chinese University of Hong KongHong Kong 999077China

Spectral compressive imaging has emerged as a powerful technique to collect the 3D spectral information as 2D *** algorithm for restoring the original 3D hyperspectral images(HSIs)from compressive measurements is pivotal in the imaging *** approaches painstakingly designed networks to directly map compressive measurements to HSIs,resulting in the lack of interpretability without exploiting the imaging *** some recent works have introduced the deep unfolding framework for explainable reconstruction,the performance of these methods is still limited by the weak information transmission between iterative *** this paper,we propose a Memory-Augmented deep Unfolding Network,termed MAUN,for explainable and accurate HSI ***,MAUN implements a novel CNN scheme to facilitate a better extrapolation step of the fast iterative shrinkage-thresholding algorithm,introducing an extra momentum incorporation step for each iteration to alleviate the information ***,to exploit the high correlation of intermediate images from neighboring iterations,we customize a cross-stage transformer(CSFormer)as the deep denoiser to simultaneously capture self-similarity from both in-stage and cross-stage features,which is the first attempt to model the long-distance dependencies between iteration *** experiments demonstrate that the proposed MAUN is superior to other state-of-the-art methods both visually and *** code is publicly available at https://***/HuQ1an/MAUN.

关键词： Deep folding iteration

来源：评论

学校读者我要写书评

暂无评论

Downsampling consistency correction-based quality enhancement for CNN-based light field image super-resolution

引用

Multimedia Tools and Applications 2024年第34期83卷 80919-80939页

作者： Chung, Kuo-Liang Hsieh, Tsung-Lun Department of Computer Science and Information Engineering National Taiwan University of Science and Technology No. 43 Section 4 Keelung Road Taipei10672 Taiwan

In recent years, numerous CNN-based light field (LF) image super-resolution (SR) methods have been developed. However, due to the downsampling inconsistency between low-resolution (LR) testing LF images and LR training LF images, they may suffer from quality degradation. To address this quality degradation issue, this paper proposes a downsampling consistency correction-based (DCC-based) quality enhancement method. Firstly, a quality-based voting strategy is introduced to identify the downsampling scheme used in the training step. Next, a cascaded Swin Transformer-based recognizer is proposed to identify the downsampled position and downsampling scheme used in the LR testing LF image, after which the proposed DCC-based method is employed to significantly improve the quality of the upsampled LF image. Comprehensive experiments have been conducted on typical LF image datasets to demonstrate the significant quality improvement achieved by our method in comparison to state-of-the-art LF SR methods. © The Author(s), under exclusive licence to Springer Science+Business Media, LLC, part of Springer Nature 2024.

关键词： Image enhancement

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：