检索结果-内蒙古大学图书馆

23rd IEEE International Conference on software Quality, Reliability, and Security Companion, QRS-C 2023

作者： Su, Rui Zhang, Zhiyi Zhou, Yuqian Yao, Yongming Collage of Computer Science and Technology Nanjing University of Aeronautics and Astronautics Nanjing China Collaborative Innovation Center of Novel Software Technology and Industrialization Nanjing China Nanjing University of Aeronautics and Astronautics Ministry Key Laboratory for Safety-Critical Software Development and Verification Nanjing China Army Engineering of PLA Nanjing China

ISBN: (纸本)9798350359398

Mutation testing is a crucial technique to assess the effectiveness of test suites, but it can be computationally expensive. Symbolic execution is a formal technique that generates high-coverage test cases, but its scalability is limited. In this paper, we present a novel approach to mutation testing that utilizes symbolic execution, which explores only program regions impacted by mutations, prioritizes paths containing mutant branches, and generates test cases aimed at mutations. We implemented our approach in MutSym and evaluated it on real-world programs, demonstrating its effectiveness and efficiency. Our approach enhances the quality and reliability of software systems by detecting and eliminating faults early in the development process. © 2023 IEEE.

关键词： software testing

来源：评论

学校读者我要写书评

暂无评论

On the intrinsic structures of spiking neural networks

The Journal of Machine Learning Research

引用

The Journal of Machine Learning Research 2024年第1期25卷 9206-9279页

作者： Shao-Qun Zhang Jia-Yi Chen Jin-Hui Wu Gao Zhang Huan Xiong Bin Gu Zhi-Hua Zhou National Key Laboratory for Novel Software Technology and School of Intelligent Science and Technology Nanjing University China National Key Laboratory for Novel Software Technology and School of Artificial Intelligence Nanjing University China School of Mathematical Sciences Jiangsu Second Normal University China Department of Machine Learning Mohamed bin Zayed University of Artificial Intelligence UAE

Recent years have emerged a surge of interest in spiking neural networks (SNNs). The performance of SNNs hinges not only on searching apposite architectures and connection weights, similar to conventional artificial neural networks, but also on the meticulous configuration of their intrinsic structures. However, there has been a dearth of comprehensive studies examining the impact of intrinsic structures; thus developers often feel challenging to apply a standardized configuration of SNNs across diverse datasets or tasks. This work delves deep into the intrinsic structures of SNNs. Initially, we draw two key conclusions: (1) the membrane time hyper-parameter is intimately linked to the eigenvalues of the integration operation, dictating the functional topology of spiking dynamics; (2) various hyper-parameters of the firing-reset mechanism govern the overall firing capacity of an SNN, mitigating the injection ratio or sampling density of input data. These findings elucidate why the efficacy of SNNs hinges heavily on the configuration of intrinsic structures and lead to a recommendation that enhancing the adaptability of these structures contributes to improving the overall performance and applicability of *** by this recognition, we propose two feasible approaches to enhance SNN learning, involving developing self-connection architectures and stochastic spiking neurons to augment the adaptability of the integration operation and firing-reset mechanism, respectively. We theoretically prove that (1) both methods promote the expressive property for universal approximation, (2) the incorporation of self-connection architectures fosters ample solutions and structural stability for SNNs approximating adaptive dynamical systems, (3) the stochastic spiking neurons maintain generalization bounds with an exponential reduction in Rademacher complexity. Empirical experiments conducted on various real-world datasets affirm the effectiveness of our proposed methods.

关键词： spiking neural network intrinsic structures integration operation self-connection architecture firing-reset mechanism stochastic excitation rademacher complexity

来源：评论

学校读者我要写书评

暂无评论

MDTeamGPT: A Self-Evolving LLM-based Multi-Agent Framework for Multi-Disciplinary Team Medical Consultation

arXiv

引用

arXiv 2025年

作者： Chen, Kai Li, Xinfeng Yang, Tianpei Wang, Hewei Dong, Wei Gao, Yang State Key Laboratory for Novel Software Technology Nanjing University China Nanyang Technological University Singapore School of Computer Science Carnegie Mellon University United States

Large Language Models (LLMs) have made significant progress in various fields. However, challenges remain in Multi-Disciplinary Team (MDT) medical consultations. Current research enhances reasoning through role assignment, task decomposition, and accumulation of medical experience. Multi-role collaboration in MDT consultations often results in excessively long dialogue histories. This increases the model’s cognitive burden and degrades both efficiency and accuracy. Some methods only store treatment histories. They do not extract effective experience or reflect on errors. This limits knowledge generalization and system evolution. We propose a multi-agent MDT medical consultation framework based on LLMs to address these issues. Our framework uses consensus aggregation and a residual discussion structure for multi-round consultations. It also employs a Correct Answer Knowledge Base (CorrectKB) and a Chain-of-Thought Knowledge Base (ChainKB) to accumulate consultation experience. These mechanisms enable the framework to evolve and continually improve diagnosis rationality and accuracy. Experimental results on the MedQA and PubMedQA datasets demonstrate that our framework achieves accuracies of 90.1% and 83.9%, respectively, and that the constructed knowledge bases generalize effectively across test sets from both datasets. Our code is publicly available at: https://***/MDTeamGPT. © 2025, CC BY.

关键词： Diagnosis

来源：评论

学校读者我要写书评

暂无评论

Loss decoupling for task-agnostic continual learning 23

Loss decoupling for task-agnostic continual learning

引用

Proceedings of the 37th International Conference on Neural Information Processing Systems

作者： Yan-Shuo Liang Wu-Jun Li National Key Laboratory for Novel Software Technology Department of Computer Science and Technology Nanjing University P. R. China

Continual learning requires the model to learn multiple tasks in a sequential order. To perform continual learning, the model must possess the abilities to maintain performance on old tasks (stability) and adapt itself to learn new tasks (plasticity). Task-agnostic problem in continual learning is a challenging problem, in which task identities are not available in the inference stage and hence the model must learn to distinguish all the classes in all the tasks. In task-agnostic problem, the model needs to learn two new objectives for learning a new task, including distinguishing new classes from old classes and distinguishing between different new classes. For task-agnostic problem, replay-based methods are commonly used. These methods update the model with both saved old samples and new samples for continual learning. Most existing replay-based methods mix the two objectives in task-agnostic problem together, inhibiting the models from achieving a good trade-off between stability and plasticity. In this paper, we propose a simple yet effective method, called loss decoupling (LODE), for task-agnostic continual learning. LODE separates the two objectives for the new task by decoupling the loss of the new task. As a result, LODE can assign different weights for different objectives, which provides a way to obtain a better trade-off between stability and plasticity than those methods with coupled loss. Experiments show that LODE can outperform existing state-of-the-art replay-based methods on multiple continual learning datasets.

关键词：

来源：评论

学校读者我要写书评

暂无评论

Stitching, Fine-tuning, Re-training: A SAM-enabled Framework for Semi-supervised 3D Medical Image Segmentation

引用

IEEE Transactions on Medical Imaging 2025年 PP卷 PP页

作者： Li, Shumeng Qi, Lei Yu, Qian Huo, Jing Shi, Yinghuan Gao, Yang Nanjing University State Key Laboratory for Novel Software Technology China Nanjing University National Institute of Healthcare Data Science China School of Computer Science and Engineering China Shandong Women's University School of Data and Computer Science China

Segment Anything Model (SAM) fine-tuning has shown remarkable performance in medical image segmentation in a fully supervised manner, but requires precise annotations. To reduce the annotation cost and maintain satisfactory performance, in this work, we leverage the capabilities of SAM for establishing semi-supervised medical image segmentation models. Rethinking the requirements of effectiveness, efficiency, and compatibility, we propose a three-stage framework, i.e., Stitching, Fine-tuning, and Re-training (SFR). The current fine-tuning approaches mostly involve 2D slice-wise fine-tuning that disregards the contextual information between adjacent slices. Our stitching strategy mitigates the mismatch between natural and 3D medical images. The stitched images are then used for fine-tuning SAM, providing robust initialization of pseudo-labels. Afterwards, we train a 3D semi-supervised segmentation model while maintaining the same parameter size as the conventional segmenter such as V-Net. Our SFR framework is plug-and-play, and easily compatible with various popular semi-supervised methods. We also develop an extended framework SFR+ with selective fine-tuning and re-training through confidence estimation. Extensive experiments validate that our SFR and SFR+ achieve significant improvements in both moderate annotation and scarce annotation across five datasets. In particular, SFR framework improves the Dice score of Mean Teacher from 29.68% to 74.40% with only one labeled data of LA dataset. The code is available at https://***/ShumengLI/SFR. © 1982-2012 IEEE.

关键词： Image segmentation

来源：评论

学校读者我要写书评

暂无评论

Contrastive Noise-Guided Invertible Network for Image Steganography

引用

IEEE Transactions on Consumer Electronics 2024年

作者： Zhang, Le Lu, Yao Lu, Guangming Harbin Institute of Technology at Shenzhen Department of Computer Science and Technology Shenzhen518057 China Guangdong Provincial Key Laboratory of Novel Security Intelligence Technologies Harbin Institute of Technology at Shenzhen Department of Computer Science and Technology Shenzhen518057 China

Image steganography aims to produce stego images through hiding secret images in the cover images to achieve covert communication. To simultaneously improve the invisibility and revealing quality of covert communication, this paper proposes a Contrastive Noise-Guided Invertible Network (CNGI-Net). Specifically, the Noised-Guided Invertible (NGI) mechanism is first proposed to actively and gradually generate adaptive noise in the forward hiding process using the interaction mechanism. The generated adaptive noise effectively guides and adjusts the cover-secret fusion process, as well as blurs the secret information in the stego images, which can significantly improve the transmission security. Besides, benefiting from the reversible property of NGI, high-quality revealed secret images can also be progressively decoupled from the stego images along the backward flow of NGI. To further ensure the invisibility and security of communication, Contrastive Hiding Learning (CHL) is proposed to improve the cover-stego similarity using the contrastive information. Extensive experiments demonstrate that the proposed CNGINet significantly promotes concealment security of transmission process and the quality of revealing for covert communication. Especially, the quality of stego and revealed secret images on 1, 2, 3 image hiding has been respectively promoted by 1.53, 0.43, 0.38 dB and 4.26, 3.45, 4.98 dB on ImageNet dataset, compared to SOTA steganography methods. © 1975-2011 IEEE.

关键词： contrastive hiding learning (CHL) contrastive noise-guided invertible network (CNGI-Net) covert communication Image steganography noisedguided invertible (NGI)

来源：评论

学校读者我要写书评

暂无评论

Git Merge Conflict Resolution Leveraging Strategy Classification and LLM

Git Merge Conflict Resolution Leveraging Strategy Classifica...

引用

IEEE International Conference on software Quality, Reliability and Security (QRS)

作者： Chaochao Shen Wenhua Yang Minxue Pan Yu Zhou College of Computer Science and Technology Nanjing University of Aeronautics and Astronautics China State Key Laboratory for Novel Software Technology Nanjing University China

In the realm of collaborative software development, version control systems (VCS) like Git play an indispensable role, enabling concurrent development and facilitating seamless integration of disparate code contributions. Despite these benefits, merge conflicts resulting from simultaneous changes to identical code lines often pose significant challenges to the integration process. Addressing this challenge, our paper introduces a novel two-stage approach, termed as CHATMERGE, for resolving Git merge conflicts. CHATMERGE pioneers a unique strategy that employs machine learning to initially predict resolution strategies, and subsequently leverages a large language model, ChatGPT, to create resolutions for conflicts that necessitate complex resolution strategies. A series of comprehensive experiments validate CHATMERGE’s efficacy, demonstrating its impressive alignment with historical manual resolutions and its superior performance relative to existing, publicly accessible tools. The paper further explores the influence of various classification algorithms and the prompt construction process for ChatGPT, providing further insights into the merge conflict resolution process. Moreover, to foster continued advancements in this area, CHATMERGE, along with its associated training and testing datasets, is made publicly available, offering a valuable resource for both developers and researchers. This work, therefore, provides both an innovative solution to merge conflict resolution and a strong foundation for future explorations in this domain.

关键词：

来源：评论

学校读者我要写书评

暂无评论

An efficient split learning framework for recurrent neural network in mobile edge environment 22

An efficient split learning framework for recurrent neural n...

引用

2022 Conference on Research in Adaptive and Convergent Systems, RACS 2022

作者： Zhou, Wenxuan Qu, Zhihao Zhao, Yanchao Tang, Bin Ye, Baoliu Key Laboratory of Water Big Data Technology of Ministry of Water Resources and School of Computer and Information Hohai University Jiangsu Nanjing China State Key Laboratory for Novel Software Technology Nanjing University Jiangsu Nanjing China College of Computer Science and Technology Nanjing University of Aeronautics and Astronautics Jiangsu Nanjing China

ISBN: (纸本)9781450393980

Split learning is an emerging distributed machine learning paradigm that exhibits great advantages in privacy protection and training efficiency. Most previous studies pay attention to client-server based split learning on Convolutional Neural networks (CNNs), and little effort has been spent on investigating the Recurrent Neural Networks (RNNs) oriented split learning. To fill in this gap, we propose a multi-hop split learning algorithm for RNN in mobile edge computing, namely M-SLRNN. The core idea is to efficiently orchestrate resource-constrained mobile devices to collaboratively train a large-size model. We formulate the optimization problem with the objective of minimizing the total training time and theoretically analyze its hardness. Then, we propose an efficient algorithm that conditionally has an approximation ratio of 2. Extensive experiments on both image and text classification tasks demonstrate the effectiveness and the efficiency of the proposed algorithm. © 2022 ACM.

关键词： Recurrent neural networks

来源：评论

学校读者我要写书评

暂无评论

Deep User and Item Inter-matching Network for CTR Prediction 28th

Deep User and Item Inter-matching Network for CTR Predicti...

引用

28th International Conference on Database Systems for Advanced Applications, DASFAA 2023

作者： Yuan, Zhiyang Xiao, Yingyuan Yang, Peilin Hao, Qingbo Wang, Hongya Engineering Research Center of Learning-Based Intelligent System Ministry of Education Tianjin University of Technology Tianjin300384 China Tianjin Key Laboratory of Intelligence Computing and Novel Software Technology Tianjin University of Technology Tianjin300384 China College of Computer Science and Engineering Donghua University Shanghai China

ISBN: (纸本)9783031306716

CTR prediction plays an important role in increasing company revenue and user experience, and many efforts start with historical behavior to uncover user interest. There are two main problems with previous works: (1) When most previous works mined interests from users’ historical behaviors, they only focus on implicit or explicit interests. (2) When most previous works mined user interests through the relationship between target users and similar users, the representation of target users was not rich and accurate enough, resulting in less accurate item recommendations. Therefore the Deep User and Item Inter-Matching Network (DUIIN) is proposed in this paper to solve the above problems. First, we design Item-to-Item Network (IIN), using two different sub-networks Evolving Interest Network (EIN) and Feature Interaction Network (FIN) to mine the hidden interests and explicit interests shown by users’ historical behaviors, respectively. Then the User-to-User Network (UUN) is designed to mine user interests through the relationship between target users and similar users after representing the target users more accurately and richly. The experimental results show that the DUIIN model proposed in this paper performs better than other state-of-the-art models. © 2023, The Author(s), under exclusive license to Springer Nature Switzerland AG.

关键词： Forecasting

来源：评论

学校读者我要写书评

暂无评论

TCB Intrusion Detection Method Based on Data Enhancement 10th

TCB Intrusion Detection Method Based on Data Enhancement

引用

10th International Conference of Pioneering computer Scientists, Engineers and Educators, ICPCSEE 2024

作者： Ye, Yunlong Zhou, Lin Hu, Xiaobo Wang, Yong Department of Computer Science and Technology Shanghai University of Electric Power Shanghai China State Grid Key Laboratory of Power Industrial Chip Design and Analysis Technology Beijing Smart-Chip Microelectronics Technology Co. Ltd. Beijing China

ISBN: (纸本)9789819787456

Within the online sphere, intrusion detection technology plays an increasingly vital role in safeguarding our information security. The current research on intrusion detection mainly focuses on two aspects: machine learning and deep learning, but it cannot be ignored that data balance also affects the efficiency of intrusion detection. The vast majority of current datasets are imbalanced, and during the training process, the model ignores a few class samples, which affects the efficiency of intrusion detection. Therefore, in response to this issue, this article introduces a novel intrusion detection model that integrates Transformer and CNN-BiLSTM, leveraging WGAN-enhanced data. This model first uses Transformer for global capture, establishes global connections, then uses CNN-BiLTM to capture spatial and temporal features, and finally uses Softmax classifier for classification. The dataset used in this article is NSL-KDD. The experimental results show that after data balance processing, the accuracy of two classification tasks of TCB model reaches 92.19%, which is at least 2.24% higher than other models, and the accuracy of multi classification tasks reaches 82.94%, which is at least 1.03% higher than other models. © The Author(s), under exclusive license to Springer Nature Singapore Pte Ltd. 2024.

关键词： Network intrusion

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：