检索结果-内蒙古大学图书馆

arXiv 2023年

作者： Li, Huantong Wu, Xiangmiao Lv, Fanbing Liao, Daihai Li, Thomas H. Zhang, Yonggang Han, Bo Tan, Mingkui South China University of Technology China Changsha Hisense Intelligent System Research Institute Co. Ltd China Hong Kong Baptist University Hong Kong Key Laboratory of Big Data and Intelligent Robot Ministry of Education China PengCheng Laboratory China Information Technology R&D Innovation Center of Peking University China School of Electronic and Computer Engineering Peking University Shenzhen Graduate School Shenzhen China

Zero-shot quantization (ZSQ) is promising for compressing and accelerating deep neural networks when the data for training full-precision models are inaccessible. In ZSQ, network quantization is performed using synthetic samples, thus, the performance of quantized models depends heavily on the quality of synthetic samples. Nonetheless, we find that the synthetic samples constructed in existing ZSQ methods can be easily fitted by models. Accordingly, quantized models obtained by these methods suffer from significant performance degradation on hard samples. To address this issue, we propose HArd sample Synthesizing and Training (HAST). Specifically, HAST pays more attention to hard samples when synthesizing samples and makes synthetic samples hard to fit when training quantized models. HAST aligns features extracted by full-precision and quantized models to ensure the similarity between features extracted by these two models. Extensive experiments show that HAST significantly outperforms existing ZSQ methods, achieving performance comparable to models that are quantized with real data. Copyright © 2023, The Authors. All rights reserved.

关键词： Deep neural networks

来源：评论

学校读者我要写书评

暂无评论

Coordinate Attention Based 3D-CNN Using Ghost Multi-Scale for Diagnosing Alzheimer’s Disease

Coordinate Attention Based 3D-CNN Using Ghost Multi-Scale fo...

引用

International Joint Conference on Neural Networks (IJCNN)

作者： Xiaolu Lin Pinya Lu Jie Pan Hongqin Yang Xuemei Ding Fujian Provincial Engineering Research Center for Public Service Big Data Mining and Application Fujian Provincial University Engineering Research Center for Big Data Analysis and Application Fujian Normal University Fuzhou China Xiamen Jingyi Zhikang Technology Co. Ltd Xiamen China Fujian Provincial Key Laboratory for Photonics Technology Key Laboratory of OptoElectronic Science and Technology for Medicine of Ministry of Education Fujian Normal University Fuzhou China School of Computing Engineering and Intelligent Systems Ulster University Derry-Londonderry NI UK

ISBN: (数字)9798350359312

ISBN: (纸本)9798350359329

Alzheimer's disease (AD) is a neurodegenerative disease and mild cognitive impairment (MCI) is the early stage of AD. Previous studies have predominantly focused on binary classification using 3 dimensional - convolutional neural network (3D-CNN) for AD diagnosis, with limited progress in multi-classification. Moreover, the current 3D-CNNs often adopt a single-scale architecture with massive parameters growth. Additionally, obtaining precise location information of brain imaging data is crucial for improving the classification accuracy with 3D-CNN. Hence, we propose a multi-scale 3D-CNN based on coordinate attention mechanism to marvelously capture and integrate 3D features with fewer parameters, improving the accuracy of AD diagnosis. A total of 447 cognitively normal (CN), 512 MCI, and 358 AD sMRI images from the Alzheimer's Disease Neuroimaging Initiative datasets are used for multi-class classification task, yielding a classification accuracy of 92.8%. The model merely involves 2.41 M parameters and achieves the best classification results with the least number of parameters when compared to other representative CNN architectures including ResNet 18, ResNet 34, ConvNeXt tiny, and VGG 11. Through the ablation experiment, the addition of attention mechanism and the multi-scale classification enhances the classification performance by 4.5% and 1.5%, respectively. Furthermore, our model outperforms the other six existing studies in terms of accuracy for classifying AD vs. MCI vs. CN. Overall, this study underscores the efficacy of our approach for AD diagnosis, showcasing its utility in diagnosing AD patients and providing novel insights for diagnosing other neurological disorder diseases.

关键词： Neuroimaging Accuracy Attention mechanisms Three-dimensional displays Resonance Convolutional neural networks Alzheimer's disease

来源：评论

学校读者我要写书评

暂无评论

Multimodal Stacked Cross Attention Network for Fine-Grained Fake News Detection

Multimodal Stacked Cross Attention Network for Fine-Grained ...

引用

IEEE International Conference on Multimedia and Expo (ICME)

作者： Zhongqiang Huang Yuxue Hu Zhi Zeng Xiang Li Ying Sha College of Informatics Huazhong Agricultural University Wuhan China Key Laboratory of Smart Farming for Agricultural Animals Wuhan China Hubei Engineering Technology Research Center of Agricultural Big Data Wuhan China Engineering Research Center of Intelligent Technology for Agriculture Ministry of Education Wuhan China

Fake news is usually disseminated in a multimodal form, which incorporates natural language, visual language, and so on. Therefore, many deep learning approaches are proposed to detect multimodal fake news. However, a drawback of existing methods is that they simply fuse unimodal features and ignore the latent semantic alignment of image and text modalities. In this paper, we propose a novel Multimodal Stacked Cross Attention Network (MSCA) to better align and fuse multimodal token-level textual and visual features for fake news detection. Experiments conducted on two publicly available datasets show that our method can significantly improve performance compared with other models. Furthermore, experimental analysis shows that MSCA can effectively align and fuse token-level features of multiple modalities.

关键词：

来源：评论

学校读者我要写书评

暂无评论

Context Decoupling Augmentation for Weakly Supervised Semantic Segmentation

Context Decoupling Augmentation for Weakly Supervised Semant...

引用

International Conference on Computer Vision (ICCV)

作者： Yukun Su Ruizhou Sun Guosheng Lin Qingyao Wu School of Software and Engineering South China University of Technology Key Laboratory of Big Data and Intelligent Robot Ministry of Education School of Computer Science and Engineering Nanyang Technological University

ISBN: (纸本)9781665428132

data augmentation is vital for deep learning neural networks. By providing massive training samples, it helps to improve the generalization ability of the model. Weakly supervised semantic segmentation (WSSS) is a challenging problem that has been deeply studied in recent years, conventional data augmentation approaches for WSSS usually employ geometrical transformations, random cropping and color jittering. However, merely increasing the same contextual semantic data does not bring much gain to the networks to distinguish the objects, e.g., the correct image-level classification of "aeroplane" may be not only due to the recognition of the object itself, but also its co-occurrence context like "sky", which will cause the model to focus less on the object features. To this end, we present a Context Decoupling Augmentation (CDA) method, to change the inherent context in which the objects appear and thus drive the network to remove the dependence between object instances and contextual information. To validate the effectiveness of the proposed method, extensive experiments on PASCAL VOC 2012 and COCO datasets with several alternative network architectures demonstrate that CDA can boost various popular WSSS methods to the new state-of-the-art by a large margin. Code is available at https://***/suyukun666/CDA

关键词： Training Deep learning Image segmentation Image recognition Image color analysis Semantics Neural networks

来源：评论

学校读者我要写书评

暂无评论

Generalized Category Discovery with Large Language Models in the Loop

arXiv

引用

arXiv 2023年

作者： An, Wenbin Shi, Wenkai Tian, Feng Lin, Haonan Wang, QianYing Wu, Yaqiang Cai, Mingxiang Wang, Luyan Chen, Yan Zhu, Haiping Chen, Ping School of Automation Science and Engineering Xi’an Jiaotong University China School of Computer Science and Technology Xi’an Jiaotong University China Ministry of Education Key Laboratory of Intelligent Networks and Network Security China Shaanxi Province Key Laboratory of Big Data Knowledge Engineering China Lenovo Research China University of Massachusetts Boston United States

Generalized Category Discovery (GCD) is a crucial task that aims to recognize both known and novel categories from a set of unlabeled data by utilizing a few labeled data with only known categories. Due to the lack of supervision and category information, current methods usually perform poorly on novel categories and struggle to reveal semantic meanings of the discovered clusters, which limits their applications in the real world. To mitigate the above issues, we propose Loop, an end-to-end active-learning framework that introduces Large Language Models (LLMs) 1 into the training loop, which can boost model performance and generate category names without relying on any human efforts. Specifically, we first propose Local Inconsistent Sampling (LIS) to select samples that have a higher probability of falling to wrong clusters, based on neighborhood prediction consistency and entropy of cluster assignment probabilities. Then we propose a Scalable Query strategy to allow LLMs to choose true neighbors of the selected samples from multiple candidate samples. Based on the feedback from LLMs, we perform Refined Neighborhood Contrastive Learning (RNCL) to pull samples and their neighbors closer to learn clustering-friendly representations. Finally, we select representative samples from clusters corresponding to novel categories to allow LLMs to generate category names for them. Extensive experiments on three benchmark datasets show that Loop outperforms SOTA models by a large margin and generates accurate category names for the discovered clusters. Code and data are available at https://***/Lackel/LOOP. Copyright © 2023, The Authors. All rights reserved.

关键词： Semantics

来源：评论

学校读者我要写书评

暂无评论

Learning defense transformers for counterattacking adversarial examples

arXiv

引用

arXiv 2021年

作者： Li, Jincheng Cao, Jiezhang Zhang, Yifan Chen, Jian Tan, Mingkui South China University of Technology China ETH Zürich Switzerland National University of Singapore Singapore Key Laboratory of Big Data and Intelligent Robot Ministry of Education China

Deep neural networks (DNNs) are vulnerable to adversarial examples with small perturbations. Adversarial defense thus has been an important means which improves the robustness of DNNs by defending against adversarial examples. Existing defense methods focus on some specific types of adversarial examples and may fail to defend well in real-world applications. In practice, we may face many types of attacks where the exact type of adversarial examples in real-world applications can be even unknown. In this paper, motivated by that adversarial examples are more likely to appear near the classification boundary, we study adversarial examples from a new perspective that whether we can defend against adversarial examples by pulling them back to the original clean distribution. We theoretically and empirically verify the existence of defense affine transformations that restore adversarial examples. Relying on this, we learn a defense transformer to counterattack the adversarial examples by parameterizing the affine transformations and exploiting the boundary information of DNNs. Extensive experiments on both toy and real-world datasets demonstrate the effectiveness and generalization of our defense transformer. Copyright © 2021, The Authors. All rights reserved.

关键词： Network security

来源：评论

学校读者我要写书评

暂无评论

Self-supervised 3D Skeleton Action Representation Learning with Motion Consistency and Continuity

Self-supervised 3D Skeleton Action Representation Learning w...

引用

International Conference on Computer Vision (ICCV)

作者： Yukun Su Guosheng Lin Qingyao Wu School of Software and Engineering South China University of Technology Key Laboratory of Big Data and Intelligent Robot Ministry of Education School of Computer Science and Engineering Nanyang Technological University

ISBN: (纸本)9781665428132

Recently, self-supervised learning (SSL) has been proved very effective and it can help boost the performance in learning representations from unlabeled data in the image domain. Yet, very little is explored about its usefulness in 3D skeleton-based action recognition understanding. Directly applying existing SSL techniques for 3D skeleton learning, however, suffers from trivial solutions and imprecise representations. To tackle these drawbacks, we consider perceiving the consistency and continuity of motion at different playback speeds are two critical issues. To this end, we propose a novel SSL method to learn the 3D skeleton representation in an efficacious way. Specifically, by constructing a positive clip (speed-changed) and a negative clip (motion-broken) of the sampled action sequence, we encourage the positive pairs closer while pushing the negative pairs to force the network to learn the intrinsic dynamic motion consistency information. Moreover, to enhance the learning features, skeleton interpolation is further exploited to model the continuity of human skeleton data. To validate the effectiveness of the proposed method, extensive experiments are conducted on Kinetics, NTU60, NTU120, and PKUMMD datasets with several alternative network architectures. Experimental evaluations demonstrate the superiority of our approach and through which, we can gain significant performance improvement without using extra labeled data.

关键词： Representation learning Interpolation Three-dimensional displays Dynamics Transfer learning Force Network architecture

来源：评论

学校读者我要写书评

暂无评论

Vehicle Segmentation with Coarse Distance Estimation Based on Monocular Vision

Vehicle Segmentation with Coarse Distance Estimation Based o...

引用

IEEE International Conference on e-Business Engineering (ICEBE)

作者： Fangyuan Xie Pengshuai Yin Zehui Ke Ruizhou Sun Ke Ding Jian Chen Qingyao Wu School of Software Engineering South China University of Technology Guangzhou China Key Laboratory of Big Data and Intelligent Robot Ministry of Education China Pazhou Lab Guangzhou China

ISBN: (纸本)9781665444194

With the increasing population of drivers and the popularity of intelligent cars, traffic safety is becoming more concerned. Aiming to reduce the probability of road accidents, many works have explored how to provide an early warning when the distances between the current car (or the in-car camera) and vehicles ahead are dangerous. In this paper, benefiting from the robust pixel-level understanding of cutting-edge deep networks, we introduce a vehicle segmentation method based on monocular vision, detecting vehicles in front of the current one and generating a segmentation mask for each vehicle instance detected. Instead of concentrating our efforts on measuring distance precisely, we categorize distances into three classes (“near”, “medium”, and “far”), which are pretty enough for an early collision warning. Our model is trained and evaluated on our hand-made dataset, after which we transfer and deploy it onto an Android TV virtual machine to test its performance. All the results we got show that our method does a good job in vehicle segmentation with coarse distance estimation, being applicable to aided driving.

关键词： TV Sociology Estimation Cameras Virtual machining Safety Automobiles

来源：评论

学校读者我要写书评

暂无评论

Multi-Granularity Open Intent Classification via Adaptive Granular-Ball Decision Boundary 39

Multi-Granularity Open Intent Classification via Adaptive Gr...

引用

39th Annual AAAI Conference on Artificial Intelligence, AAAI 2025

作者： Li, Yanhua Ouyang, Xiaocao Pan, Chaofan Zhang, Jie Zhao, Sen Xia, Shuyin Yang, Xin Wang, Guoyin Li, Tianrui School of Computing and Artificial Intelligence Southwestern University of Finance and Economics Chengdu 611130 China Engineering Research Center of Intelligent Finance Ministry of Education Chengdu 611130 China Key Laboratory of Big Data Intelligent Computing Chongqing Key Laboratory of Computational Intelligence Chongqing University of Posts and Telecommunications Chongqing 400065 China National Center for Applied Mathematics in Chongqing Chongqing Normal University Chongqing 401331 China School of Computing and Artificial Intelligence Southwest Jiaotong University Chengdu 611756 China

ISBN: (纸本)157735897X

Open intent classification is critical for the development of dialogue systems, aiming to accurately classify known intents into their corresponding classes while identifying unknown intents. Prior boundary-based methods assumed known intents fit within compact spherical regions, focusing on coarse-grained representation and precise spherical decision boundaries. However, these assumptions are often violated in practical scenarios, making it difficult to distinguish known intent classes from unknowns using a single spherical boundary. To tackle these issues, we propose a Multi-granularity Open intent classification method via adaptive Granular-Ball decision boundary (MOGB). Our MOGB method consists of two modules: representation learning and decision boundary acquiring. To effectively represent the intent distribution, we design a hierarchical representation learning method. This involves iteratively alternating between adaptive granular-ball clustering and nearest sub-centroid classification to capture fine-grained semantic structures within known intent classes. Furthermore, multi-granularity decision boundaries are constructed for open intent classification by employing granular-balls with varying centroids and radii. Extensive experiments conducted on three public datasets demonstrate the effectiveness of our proposed method. Copyright © 2025, Association for the Advancement of Artificial Intelligence (***). All rights reserved.

关键词：

来源：评论

学校读者我要写书评

暂无评论

How to train your agent to read and write

arXiv

引用

arXiv 2021年

作者： Liu, Li He, Mengge Xu, Guanghui Tan, Mingkui Wu, Qi School of Software Engineering South China University of Technology China Pazhou Laboratory China University of Adelaide Australia Key Laboratory of Big Data and Intelligent Robot Ministry of Education United Kingdom

Reading and writing research papers is one of the most privileged abilities that a qualified researcher should master. However, it is difficult for new researchers (e.g., students) to fully grasp this ability. It would be fascinating if we could train an intelligent agent to help people read and summarize papers, and perhaps even discover and exploit the potential knowledge clues to write novel papers. Although there have been existing works focusing on summarizing (i.e., reading) the knowledge in a given text or generating (i.e., writing) a text based on the given knowledge, the ability of simultaneously reading and writing is still under development. Typically, this requires an agent to fully understand the knowledge from the given text materials and generate correct and fluent novel paragraphs, which is very challenging in practice. In this paper, we propose a Deep ReAder-Writer (DRAW) network, which consists of a Reader that can extract knowledge graphs (KGs) from input paragraphs and discover potential knowledge, a graph-to-text Writer that generates a novel paragraph, and a Reviewer that reviews the generated paragraph from three different aspects. Extensive experiments show that our DRAW network outperforms considered baselines and several state-of-the-art methods on AGENDA and M-AGENDA datasets. Our code and supplementary are released at https://***/menggehe/DRAW. Copyright © 2021, The Authors. All rights reserved.

关键词： Knowledge graph

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：