检索结果-内蒙古大学图书馆

arXiv 2024年

作者： Chen, Tingwei Wang, Yantao Chen, Hanzhi Zhao, Zijian Li, Xinhao Piovesan, Nicola Zhu, Guangxu Shi, Qingjiang Shenzhen Research Institute of Big Data Shenzhen China School of Computer Science and Engineering Sun Yat-sen University Guangzhou China Shenzhen China Huawei Technologies Boulogne-Billancourt France School of Software Engineering Tongji University Shanghai China

The introduction of fifth-generation (5G) radio technology has revolutionized communications, bringing unprecedented automation, capacity, connectivity, and ultra-fast, reliable communications. However, this technological leap comes with a substantial increase in energy consumption, presenting a significant challenge. To improve the energy efficiency of 5G networks, it is imperative to develop sophisticated models that accurately reflect the influence of base station (BS) attributes and operational conditions on energy usage. Importantly, addressing the complexity and interdependencies of these diverse features is particularly challenging, both in terms of data processing and model architecture design. This paper proposes a novel 5G base stations energy consumption modelling method by learning from a real-world dataset used in the ITU 5G Base Station Energy Consumption Modelling Challenge in which our model ranked second. Unlike existing methods that omit the Base Station Identifier (BSID) information and thus fail to capture the unique energy fingerprint in different base stations, we incorporate the BSID into the input features and encoding it with an embedding layer for precise representation. Additionally, we introduce a novel masked training method alongside an attention mechanism to further boost the model's generalization capabilities and accuracy. After evaluation, our method demonstrates significant improvements over existing models, reducing Mean Absolute Percentage Error (MAPE) from 12.75% to 4.98%, leading to a performance gain of more than 60. Copyright © 2024, The Authors. All rights reserved.

关键词： Energy utilization

来源：评论

学校读者我要写书评

暂无评论

Assessing the Robustness of LLM-based NLP software via Automated Testing

arXiv

引用

arXiv 2024年

作者： Xiao, Mingxuan Xiao, Yan Ji, Shunhui Cai, Hanbo Xue, Lei Zhang, Pengcheng Key Laboratory of Water Big Data Technology Ministry of Water Resources College of Computer Science and Software Engineering Hohai University Nanjing211100 China School of Cyber Science and Technology Shenzhen Campus of Sun Yat-sen University Shenzhen518107 China

Benefiting from the advancements in LLMs, NLP software has undergone rapid development. Such software is widely employed in various safety-critical tasks, such as financial sentiment analysis, toxic content moderation, and log generation. Unlike traditional software, LLM-based NLP software relies on prompts and examples as inputs. Given the complexity of LLMs and the unpredictability of real-world inputs, quantitatively assessing the robustness of such software is crucial. However, to the best of our knowledge, no automated robustness testing methods have been specifically designed to evaluate the overall inputs of LLM-based NLP software. To this end, this paper introduces the first AutOmated Robustness Testing frAmework, AORTA, which reconceptualizes the testing process into a combinatorial optimization problem. Existing testing methods designed for DNN-based software can be applied to LLM-based software by AORTA, but their effectiveness is limited. To address this, we propose a novel testing method for LLM-based software within AORTA called Adaptive Beam Search. ABS is tailored for the expansive feature space of LLMs and improves testing effectiveness through an adaptive beam width and the capability for backtracking. We successfully embed 18 test methods in the designed framework AORTA and compared the test validity of ABS with three datasets and five threat models. ABS facilitates a more comprehensive and accurate robustness assessment before software deployment, with an average test success rate of 86.138%. Compared to the currently best-performing baseline PWWS, ABS significantly reduces the computational overhead by up to 3441.895 seconds per successful test case and decreases the number of queries by 218.762 times on average. Furthermore, test cases generated by ABS exhibit greater naturalness and transferability. © 2024, CC BY-NC-ND.

关键词： software testing

来源：评论

学校读者我要写书评

暂无评论

Neural Architecture Search Based on Bipartite Graphs for Text Classification

引用

IEEE Transactions on Neural Networks and Learning Systems 2024年第6期36卷 10749-10763页

作者： Xueming Yan Han Huang Yaochu Jin Zilong Wang Zhifeng Hao School of Information Science and Technology Guangdong University of Foreign Studies Guangzhou China School of Software Engineering South China University of Technology Guangzhou China Key Laboratory of Big Data and Intelligent Robot (SCUT) MOE of China Guangzhou China School of Engineering Westlake University Hangzhou China College of Mathematics and Computer Science Shantou University Guangdong China

Neural architecture search (NAS) is crucial for text representation in natural language processing (NLP); however, much less work on NAS for text classification has been proposed compared with NAS for computer vision. Similar to NAS for vision tasks, most existing work rely on a manually designed search space defined by a directed acyclic graph (DAG), resulting in limited generalization capability and high computational complexity. In text classification, the topological order of the NAS operators is essential for enhancing generalization, which cannot be accurately represented by a DAG. To address this issue, we propose a bipartite graph-based NAS (BGNAS) for text classification, which converts a DAG into a dual graph and then into a bipartite graph. This transformation makes it possible to accurately capture the topological order using multi-bigraph matching. In addition, we formulate NAS as a problem of identifying the lower bound of a submodular function, theoretically ensuring that optimal architectures in a bipartite graph-based search space can be identified using fewer search operators. Reduction of the search space is achieved by eliminating ineffective associated matching rules among search operators with a pruning strategy. As a result, the bipartite graph-based search space becomes more compact and less dependent on complex contextual semantics of text data. Experimental results on public benchmark problems demonstrate that BGNAS achieves better performance than the state-of-the-art NAS algorithms and is computationally more efficient. We also demonstrate that the bipartite graph search space can more effectively capture contextual semantics, thereby enhancing the generalization capability.

关键词： computer architecture Bipartite graph Text categorization Search problems Neural networks Semantics Neural architecture search Electronic mail Optimization Training

来源：评论

学校读者我要写书评

暂无评论

Chinese Emergency Event Extraction Based on Contrastive Learning with Event Semantic Features 19

Chinese Emergency Event Extraction Based on Contrastive Lear...

引用

19th International Conference on Natural Computation, Fuzzy Systems and Knowledge Discovery, ICNC-FSKD 2023

作者： Sheng, Xinyi Gu, Jinguang Yan, Youcheng Xu, Fangfang Wuhan University of Science and Technology College of Computer Science and Technology Wuhan430065 China Institute of Big Data Science and Engineering Wuhan University of Science and Technology Wuhan430065 China The Key Lab. of Rich-Media Knowledge Org. and Serv. of Digit. Publ. Content Inst. of Sci. and Tech. Info. of China Beijing100038 China

ISBN: (纸本)9798350304398

Extracting emergency events from a large amount of unstructured information is essential for improving early warning and emergency response. Existing event extraction methods for specialist fields often rely on well-designed syntactic dependencies or external knowledge graphs, which can provide high accuracy in their specialist fields. However, these approaches may have poor generalization capabilities and are difficult to apply directly to the emergency field. In addition, most existing methods ignore the interdependencies between event triggers and event arguments, which may be crucial semantic features for event extraction. To solve these problems, we propose a Chinese emergent event extraction method based on contrastive learning with event semantic features(CNEmergEE). Firstly, the proposed approach automatically expands the pre-trained model's vocabulary on a domain corpus. Then, continuing learning on the same corpus allows the model to complete domain adaptation. Finally, we use contrast learning to enable models to learn semantic representations of events. Experimental results on the CEC (Chinese Emergency Corpus) dataset show that our approach achieves better improvement. © 2023 IEEE.

关键词： Semantics

来源：评论

学校读者我要写书评

暂无评论

Collaborative Segmentation Model for Colonoscopy Ulcers based on Fuzzy Labeling*

Collaborative Segmentation Model for Colonoscopy Ulcers base...

引用

International Conference on Advanced Robotics and Mechatronics (ICARM)

作者： Yanning Lin Jie Chen Ziqian Ding Linxu Chen Le Li Zhongsheng Zhu Zhaoxia Wang Jianqiang Li College of Computer Science and Software Engineering Shenzhen University Shenzhen China Shenzhen Children’s Hospital Shenzhen China Gastroenterology Shenzhen Children’s Hospital China National Engineering Laboratory for Big Data System Computing Technology Shenzhen University China

ISBN: (数字)9798350385724

ISBN: (纸本)9798350385731

Inflammatory Bowel Disease (IBD) is a global chronic intestinal inflammatory disease, and its incidence rate increases year by year with the progress of economic globalization. Currently, the diagnosis of IBD in children mainly relies on endoscopic examination, but scoring endoscopic images is a challenging issue, especially in distinguishing different types of ulcers. To address this issue, this article designs a mobile application to accelerate data annotation processing and may provide reference for other unlabeled datasets. In the context of image segmentation, blurring labels has become an important issue. Deep learning methods are widely used in medical image segmentation, but their accuracy depends on high-quality annotated data. However, there are low-quality noise areas in the annotated data, and obtaining accurate and high-quality annotations becomes more time-consuming with limited annotation budgets. This article proposes a collaborative training framework to improve learning of noisy pixels. This framework determines the label confidence of an image by calculating the similarity between image pixels and surrounding pixels. Then, two parallel deep networks were constructed for semantic prediction, which aimed to guide each other on pixels that may have noise. By applying consistency in dual network prediction, the semantic information of uncertain pixels is corrected as much as possible. Experimental results have shown that this framework is slightly superior to models trained with pixel level precise labels, thus more effectively utilizing existing annotated data in the case of fuzzy labels.

关键词： Training Image segmentation Accuracy Annotations Biological system modeling Semantics Noise Collaboration Colonoscopy Medical diagnostic imaging

来源：评论

学校读者我要写书评

暂无评论

$\mathrm{D}^{3}$T: Deep Denoising Dictionary Tensor for Hyperspectral Anomaly Detection

引用

IEEE Journal of Selected Topics in Applied Earth Observations and Remote Sensing 2024年 18卷 3713-3727页

作者： Qiangqiang Shen Zonglin Liu Hanzhang Wang Yanhui Xu Yongyong Chen Yongsheng Liang School of Electronics and Information Engineering Harbin Institute of Technology Shenzhen China School of Science Harbin Institute of Technology Shenzhen China College of Electronic and Optical Engineering Nanjing University of Posts and Telecommunications Nanjing China School of Computer Science and Technology Harbin Institute of Technology Shenzhen China College of Big Data and Internet Shenzhen Technology University Shenzhen Guangdong China School of Electronics and Information Engineering Harbin Institute of Technology Shenzhen Guangdong China

Hyperspectral anomaly detection (HAD) approaches with tensor low-rank representation (TLRR) have shown engaging performance, which are capable of capturing abundant spectral and spatial information and, hence, extracting sparse anomalies from the low-rank background without ruining the intrinsic geometric structures of tensor data. However, existing HAD approaches construct the low-rank representation by a handcrafted dictionary that still contains the anomalies, resulting in an inferior representation. To this end, we propose a deep denoising dictionary tensor for hyperspectral anomaly detection ( $\mathrm{D}^{3}$ T), which can balance performance and interpretability by fusing two intuitive priors, i.e., low-rankness and deep denoising, into the dictionary tensor and coefficient tensor. In particular, the proposed $\mathrm{D}^{3}$ T first designs a tensor recovery module for dividing the low-rank background and sparse anomalies, where the background is optimized as a denoising dictionary tensor by the joint low-rankness and deep denoising. Following that, we build the TLRR model based on the denoising dictionary tensor to explore the latent representation of the input hyperspectral images, and hence, each background pixel is represented by the other clear background pixels in place of the anomaly pixels. Meanwhile, the coefficient tensor in TLRR is also optimized by the two priors to fully explore the spatial and spectral correlations in the background, and hence, the anomalies can be accurately detected. Numerous experimental results from various real-world datasets validate the effectiveness of our $\mathrm{D}^{3}$ T.

关键词： Tensors Dictionaries Noise reduction Anomaly detection Hyperspectral imaging Sparse matrices Noise Principal component analysis Feature extraction Electronic mail

来源：评论

学校读者我要写书评

暂无评论

Confidence-aware contrastive learning for selective classification 24

Confidence-aware contrastive learning for selective classifi...

引用

Proceedings of the 41st International Conference on Machine Learning

作者： Yu-Chang Wu Shen-Huan Lyu Haopu Shang Xiangyu Wang Chao Qian National Key Laboratory for Novel Software Technology and School of Artificial Intelligence Nanjing University China Key Laboratory of Water Big Data Technology of Ministry of Water Resources and College of Computer Science and Software Engineering Hohai University China and National Key Laboratory for Novel Software Technology Nanjing University China

Selective classification enables models to make predictions only when they are sufficiently confident, aiming to enhance safety and reliability, which is important in high-stakes scenarios. Previous methods mainly use deep neural networks and focus on modifying the architecture of classification layers to enable the model to estimate the confidence of its prediction. This work provides a generalization bound for selective classification, disclosing that optimizing feature layers helps improve the performance of selective classification. Inspired by this theory, we propose to explicitly improve the selective classification model at the feature level for the first time, leading to a novel Confidence-aware Contrastive Learning method for Selective Classification, CCL-SC, which similarizes the features of homogeneous instances and differentiates the features of heterogeneous instances, with the strength controlled by the model's confidence. The experimental results on typical datasets, i.e., CIFAR-10, CIFAR-100, CelebA, and ImageNet, show that CCL-SC achieves significantly lower selective risk than state-of-the-art methods, across almost all coverage degrees. Moreover, it can be combined with existing methods to bring further improvement.

关键词：

来源：评论

学校读者我要写书评

暂无评论

ShiftLIC: Lightweight Learned Image Compression with Spatial-Channel Shift Operations

引用

IEEE Transactions on Circuits and Systems for Video Technology 2025年

作者： Bao, Youneng Tan, Wen Jia, Chuanmin Li, Mu Liang, Yongsheng Tian, Yonghong Harbin Institute of Technology School of Electronics and Information Engineering School of Computer Science Guangdong Shenzhen China Beijing100871 China Shenzhen Technology University College of Big Data and Internet Guangdong Shenzhen518118 China Peng Cheng Laboratory Shenzhen China

Learned Image Compression (LIC) has attracted considerable attention due to their outstanding rate-distortion (R-D) performance and flexibility. However, the substantial computational cost poses challenges for practical deployment. The issue of feature redundancy in LIC is rarely addressed. Our findings indicate that many features within the LIC backbone network exhibit similarities. This paper introduces ShiftLIC, a novel and efficient LIC framework that employs parameter-free shift operations to replace large-kernel convolutions, significantly reducing the model’s computational burden and parameter count. Specifically, we propose the Spatial Shift Block (SSB), which combines shift operations with small-kernel convolutions to replace large-kernel. This approach maintains feature extraction efficiency while reducing both computational complexity and model size. To further enhance the representation capability in the channel dimension, we propose a channel attention module based on recursive feature fusion. This module enhances feature interaction while minimizing computational overhead. Additionally, we introduce an improved entropy model integrated with the SSB module, making the entropy estimation process more lightweight and thereby comprehensively reducing computational costs. Experimental results demonstrate that ShiftLIC outperforms leading compression methods, such as VVC Intra and GMM, in terms of computational cost, parameter count, and decoding latency. Additionally, ShiftLIC sets a new SOTA benchmark with a BD-rate gain per MACs/pixel of -102.6%, showcasing its potential for practical deployment in resource-constrained environments. The code is released at https://***/baoyu2020/ShiftLIC. © 1991-2012 IEEE.

关键词： Signal distortion

来源：评论

学校读者我要写书评

暂无评论

WMMSE Beamforming for User-Centric Cell-Free Networks with Non-Coherent Joint Transmission

WMMSE Beamforming for User-Centric Cell-Free Networks with N...

引用

IEEE Conference on Global Communications (GLOBECOM)

作者： Xi Wang Xiaotong Zhao Juncheng Wang Qingjiang Shi School of Software Engineering Tongji University Shanghai China Department of Computer Science Hong Kong Baptist University Hong Kong China Shenzhen Research Institute of Big Data Shenzhen China

We consider downlink beamforming design to maximize the weighted sum rate (WSR) in a user-centric cell-free network, where distributed access points (APs) are organized in a cluster to jointly serve each user equipment (UE). This architecture ensures uniform service quality even at the cell edge, but synchronization between APs can be problematic, necessitating the utilization of non-coherent joint transmission (NCJT) that eliminates the need for strict synchronization. Most existing works on beamforming design for NCJT assume that the classic weighted minimum mean square error (WMMSE) approach is not applicable, and design their beamforming algorithms based on the successive convex approximation (SCA) method with high computational complexity. In this work, we for the first time demonstrate the applicability of the WMMSE approach for NCJT in cell-free networks. Based on the unique observations on the structures of the WSR maximization problem for NCJT, we propose an efficient WMMSE based beamforming algorithm. Our proposed algorithm is guaranteed to converge to a stationary point of the WSR maximization problem. Furthermore, our beamforming updates are in closed form with low computational complexity. Simulation results demonstrate substantial performance gain of our proposed algorithm over the current best SCA based alternatives in both computational complexity and convergence time.

关键词：

来源：评论

学校读者我要写书评

暂无评论

Adaptive Fusion Attention Network for Facial Expression Recognition 23

Adaptive Fusion Attention Network for Facial Expression Reco...

引用

Proceedings of the 2023 12th International Conference on Computing and Pattern Recognition

作者： Guodong Li Cheng Peng Kun Zou Bowen Zhang School of Computer Science and Engineering(School of Cybersecurity) University of Electronic Science and Technology of China China School of Computing University of Electronic Science and Technology of China Zhongshan Institute China College of Big Data and Internet Shenzhen Technology University China

ISBN: (纸本)9798400707988

In this paper, we propose a novel facial expression recognition network, called Adaptive Fusion Attention Network (AFAN). Facial expression recognition is different from traditional image classification tasks. There are potentially similar features between different categories, and the differences between them may be subtle and difficult to distinguish. To solve this problem, we propose AFAN, which not only can get multi-dimensional features, but also can generate an attention map with stronger representation ability. The comparative experiments on AffectNet-7 dataset verify the excellent performance of the proposed method on facial expression recognition tasks.

关键词： Adaptive convolutional blocks Facial expression recognition Fine-grained classification Multi-head attention networks

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：