检索结果-内蒙古大学图书馆

arXiv 2025年

作者： Zheng, Li Fei, Hao Dai, Ting Peng, Zuquan Li, Fei Ma, Huisheng Teng, Chong Ji, Donghong Key Laboratory of Aerospace Information Security and Trusted Computing Ministry of Education School of Cyber Science and Engineering Wuhan University Wuhan China National University of Singapore Singapore Singapore Laboratory for Advanced Computing and Intelligence Engineering Wuxi China North China Institute of Computing Technology Beijing China

With the continuous emergence of various social media platforms frequently used in daily life, the multimodal meme understanding (MMU) task has been garnering increasing attention. MMU aims to explore and comprehend the meanings of memes from various perspectives by performing tasks such as metaphor recognition, sentiment analysis, intention detection, and offensiveness detection. Despite making progress, limitations persist due to the loss of fine-grained metaphorical visual clue and the neglect of multimodal text-image weak correlation. To overcome these limitations, we propose a multi-granular multimodal clue fusion model (MGMCF) to advance MMU. Firstly, we design an object-level semantic mining module to extract object-level image feature clues, achieving fine-grained feature clue extraction and enhancing the model’s ability to capture metaphorical details and semantics. Secondly, we propose a brand-new global-local cross-modal interaction model to address the weak correlation between text and images. This model facilitates effective interaction between global multimodal contextual clues and local unimodal feature clues, strengthening their representations through a bidirectional cross-modal attention mechanism. Finally, we devise a dual-semantic guided training strategy to enhance the model’s understanding and alignment of multimodal representations in the semantic space. Experiments conducted on the widely-used MET-MEME bilingual dataset demonstrate significant improvements over state-of-the-art baselines. Specifically, there is an 8.14% increase in precision for offensiveness detection task, and respective accuracy enhancements of 3.53%, 3.89%, and 3.52% for metaphor recognition, sentiment analysis, and intention detection tasks. These results, underpinned by in-depth analyses, underscore the effectiveness and potential of our approach for advancing MMU. Copyright © 2025, 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

Bridging the Gap using Contrastive Learning and Semantic Consistency for Unsupervised Neural Machine Translation

Bridging the Gap using Contrastive Learning and Semantic Con...

引用

Artificial intelligence and Computer engineering (ICAICE), International Conference on

作者： Chuancai Zhang Dan Qu Liming Du Kaiyuan Yang School of Information Systems Engineering University of Information Engineering Zhengzhou China Laboratory for Advanced Computing and Intelligence Engineering Zhengzhou China School of Cyberspace Security Zhengzhou University Zhengzhou China

ISBN: (数字)9798331528911

ISBN: (纸本)9798331528928

In Unsupervised Neural Machine Translation (UNMT) tasks, the lack of extensive parallel corpora makes it challenging for the model to directly optimize the correspondence between the source and target languages. UNMT models primarily rely on learning from monolingual data, making it difficult to fully capture cross-linguistic semantic alignment and consistency during training. Therefore, effectively enhancing semantic alignment between the source and target languages under unsupervised conditions has become a key challenge in the field of UNMT. To address this issue, this study proposes a semantic alignment and consistency enhancement mechanism based on contrastive learning, aiming to improve the model's ability to capture cross-lingual semantic differences by constructing an optimization objective focused on semantic consistency. The method utilizes a pretrained cross-lingual model to generate semantic embeddings for both the source and target languages, then designs a contrastive learning-based loss function that minimizes the semantic distance between correct translations and source sentences while increasing the semantic distance between incorrect translations and source sentences, thereby enhancing the model's performance in semantic consistency. The experiments validate the effectiveness and broad applicability of the semantic consistency optimization method based on contrastive learning, demonstrating great potential and advantages, especially in multilingual tasks of UNMT. The model achieved significant improvements in BLEU scores across various language pairs.

关键词： Training Translation Computational modeling Semantics Optimization methods Neural machine translation Contrastive learning Data models Multilingual Artificial intelligence

来源：评论

学校读者我要写书评

暂无评论

A Neuroevolution-Inspired Scheme for Generating Robust Internet of Things 25

A Neuroevolution-Inspired Scheme for Generating Robust Inter...

引用

25th IEEE International Conference on Computer Supported Cooperative Work in Design, CSCWD 2022

作者： Zhang, Lidi Zhang, Songwei Chen, Ning Liu, Wenyuan Zhou, Xiaobo Qiu, Tie Tianjin University College of Intelligence and Computing Tianjin300350 China Tianjin Key Laboratory of Advanced Networking Tianjin300350 China YanShan University Key Laboratory of Software Engineering of Hebei Province Qinhuangdao066004 China

ISBN: (纸本)9781665405270

Internet of Things (IoT) is growing with various applications linked in, and node failures are becoming more common as a result of malicious strikes and other issues. The cascading collapse induced by local node failures can be mitigated by robust network topology. Existing approaches for fixed topology enhance the robustness of IoT topology by reconstructing device connections. However, using existing techniques necessitates global topology optimization when new nodes are added, which takes time. To tackle this situation, this study introduces an evolutionary algorithm based on neuroevolution that generates robust IoT topology. It provides IoT topology with inherent robustness when adding extra nodes by utilizing unique mutation and crossover operators. What's more, we establish an adaptive edge density management method to reduce the rise in energy consumption caused by redundant connections when nodes join. Experimental results indicate that the proposed scheme can effectively build robust topology than multiple existing topology optimization methods in less time for diverse network sizes. © 2022 IEEE.

关键词： Network topology

来源：评论

学校读者我要写书评

暂无评论

Frozen-DETR: enhancing DETR with image understanding from frozen foundation models 24

Frozen-DETR: enhancing DETR with image understanding from fr...

引用

Proceedings of the 38th International Conference on Neural Information Processing Systems

作者： Shenghao Fu Junkai Yan Qize Yang Xihan Wei Xiaohua Xie Wei-Shi Zheng School of Computer Science and Engineering Sun Yat-sen University China and Tongyi Lab Alibaba Group and Key Laboratory of Machine Intelligence and Advanced Computing Ministry of Education China Tongyi Lab Alibaba Group School of Computer Science and Engineering Sun Yat-sen University China and Key Laboratory of Machine Intelligence and Advanced Computing Ministry of Education China and Guangdong Province Key Laboratory of Information Security Technology China School of Computer Science and Engineering Sun Yat-sen University China and Peng Cheng Laboratory Shenzhen China and Key Laboratory of Machine Intelligence and Advanced Computing Ministry of Education China and Pazhou Laboratory (Huangpu) Guangzhou Guangdong China

ISBN: (纸本)9798331314385

Recent vision foundation models can extract universal representations and show impressive abilities in various tasks. However, their application on object detection is largely overlooked, especially without fine-tuning them. In this work, we show that frozen foundation models can be a versatile feature enhancer, even though they are not pre-trained for object detection. Specifically, we explore directly transferring the high-level image understanding of foundation models to detectors in the following two ways. First, the class token in foundation models provides an in-depth understanding of the complex scene, which facilitates decoding object queries in the detector's decoder by providing a compact context. Additionally, the patch tokens in foundation models can enrich the features in the detector's encoder by providing semantic details. Utilizing frozen foundation models as plug-and-play modules rather than the commonly used backbone can significantly enhance the detector's performance while preventing the problems caused by the architecture discrepancy between the detector's backbone and the foundation model. With such a novel paradigm, we boost the SOTA query-based detector DINO from 49.0% AP to 51.9% AP (+2.9% AP) and further to 53.8% AP (+4.8% AP) by integrating one or two foundation models respectively, on the COCO validation set after training for 12 epochs with R50 as the detector's backbone.

关键词：

来源：评论

学校读者我要写书评

暂无评论

DreamView: Injecting View-specific Text Guidance into Text-to-3D Generation

arXiv

引用

arXiv 2024年

作者： Yan, Junkai Gao, Yipeng Yang, Qize Wei, Xihan Xie, Xuansong Wu, Ancong Zheng, Wei-Shi School of Computer Science and Engineering Sun Yat-sen University China Key Laboratory of Machine Intelligence and Advanced Computing Ministry of Education China Institute for Intelligent Computing Alibaba Group China

Text-to-3D generation, which synthesizes 3D assets according to an overall text description, has significantly progressed. However, a challenge arises when the specific appearances need customizing at designated viewpoints but referring solely to the overall description for generating 3D objects. For instance, ambiguity easily occurs when producing a T-shirt with distinct patterns on its front and back using a single overall text guidance. In this work, we propose DreamView, a text-to-image approach enabling multi-view customization while maintaining overall consistency by adaptively injecting the view-specific and overall text guidance through a collaborative text guidance injection module, which can also be lifted to 3D generation via score distillation sampling. DreamView is trained with large-scale rendered multi-view images and their corresponding view-specific texts to learn to balance the separate content manipulation in each view and the global consistency of the overall object, resulting in a dual achievement of customization and consistency. Consequently, DreamView empowers artists to design 3D objects creatively, fostering the creation of more innovative and diverse 3D assets. Code and model will be released at here. Copyright © 2024, The Authors. All rights reserved.

关键词： Digital elevation model

来源：评论

学校读者我要写书评

暂无评论

iManip: Skill-Incremental Learning for Robotic Manipulation

arXiv

引用

arXiv 2025年

作者： Zheng, Zexin Cai, Jia-Feng Wu, Xiao-Ming Wei, Yi-Lin Tang, Yu-Ming Zheng, Wei-Shi School of Computer Science and Engineering Sun Yat-sen University China Pengcheng Lab China Key Laboratory of Machine Intelligence and Advanced Computing Ministry of Education China

The development of a generalist agent with adaptive multiple manipulation skills has been a long-standing goal in the robotics community. In this paper, we explore a crucial task, skill-incremental learning, in robotic manipulation, which is to endow the robots with the ability to learn new manipulation skills based on the previous learned knowledge without re-training. First, we build a skill-incremental environment based on the RLBench benchmark, and explore how traditional incremental methods perform in this setting. We find that they suffer from severe catastrophic forgetting due to the previous methods on classification overlooking the characteristics of temporality and action complexity in robotic manipulation tasks. Towards this end, we propose an incremental Manipulation framework, termed iManip, to mitigate the above issues. We firstly design a temporal replay strategy to maintain the integrity of old skills when learning new skill. Moreover, we propose the extendable PerceiverIO, consisting of an action prompt with extendable weight to adapt to new action primitives in new skill. Extensive experiments show that our framework performs well in Skill-Incremental Learning. Codes of the skill-incremental environment with our framework will be open-source. Copyright © 2025, The Authors. All rights reserved.

关键词： Robot learning

来源：评论

学校读者我要写书评

暂无评论

Audio-Visual Collaborative Learning for Weakly Supervised Video Anomaly Detection

引用

IEEE Transactions on Multimedia 2025年

作者： Meng, Jingke Tian, Huilin Lin, Ge Hu, Jian-Fang Zheng, Wei-Shi Sun Yat-sen University School of Computer Science and Engineering Guangzhou China Peng Cheng Laboratory Shenzhen518005 China Key Laboratory of Machine Intelligence and Advanced Computing Sun Yat-sen University Ministry of Education China

Weakly supervised anomaly detection is to identify the time window when an anomaly event happened based on the video-level label indicating whether the video contains anomaly event or not. Recent efforts have focused on leveraging multi-modal data, specifically combining visual and audio information, to enhance detection accuracy. While some studies have explored intra-video separation techniques, the primary emphasis remains on distinguishing potentially anomalous events scoring highest from those scoring lowest as normal events. Nevertheless, challenges persist in delineating boundaries between normal and abnormal events, particularly when visual differences are subtle. Our proposed framework, called Audio-Visual Collaborative Learning (AVCL), addresses the challenge of ambiguity in weakly supervised anomaly detection. Our core idea centers around utilizing both audio track variations and the perceptual robustness of visual information to detect and differentiate challenging cases, which composed of two essential modules: the Audio-Visual collaborative Hard cases Separation (AVHS) module and the Multi-modal Mutual Learning (MML) module. The AVHS module aims to address the challenge of discerning visually ambiguous clips in anomaly videos, differentiating between normal and abnormal events. To further improve detection accuracy, we introduce the Multi-modal Mutual Learning (MML) module, and this module enables a process of mutual learning to facilitate the exchange of knowledge and expertise between the single-modal branch and the multi-modal branch. We demonstrate that the proposed approach achieves state-of-the-art detection performance on benchmarks of XD-Violence and CCTV-Fightssub datasets. © 1999-2012 IEEE.

关键词： Self-supervised learning

来源：评论

学校读者我要写书评

暂无评论

A Novel Feature Selection Strategy Based on Salp Swarm Algorithm for Plant Disease Detection

引用

Plant Phenomics 2023年第2期5卷 209-225页

作者： Xiaojun Xie Fei Xia Yufeng Wu Shouyang Liu Ke Yan Huanliang Xu Zhiwei Ji College of Artificial Intelligence Nanjing Agricultural UniversityNanjingJiangsu 210095China Center for Data Science and Intelligent Computing Nanjing Agricultural UniversityNanjingJiangsu 210095China State Key Laboratory for Crop Genetics and Germplasm Enhancement Bioinformatics CenterAcademy for Advanced Interdisciplinary StudiesNanjing Agricultural UniversityNanjingJiangsu 210095China Academy for Advanced Interdisciplinary Studies Nanjing Agricultural UniversityNanjingJiangsu 210095China Department of the Built Environment College of Design and EngineeringNational University of Singapore4 Architecture DriveSingapore 117566Singapore

Deep learning has been widely used for plant disease recognition in smart agriculture and has proven to be a powerful tool for image classification and pattern ***,it has limited interpretability for deep *** the transfer of expert knowledge,handcrafted features provide a new way for personalized diagnosis of plant ***,irrelevant and redundant features lead to high *** this study,we proposed a swarm intelligence algorithm for feature selection[salp swarm algorithm for feature selection(SSAFS)]in image-based plant disease *** is employed to determine the ideal combination of handcrafted features to maximize classification success while minimizing the number of *** verify the effectiveness of the developed SSAFS algorithm,we conducted experimental studies using SSAFS and 5 metaheuristic *** evaluation metrics were used to evaluate and analyze the performance of these methods on 4 datasets from the UCI machine learning repository and 6 plant phenomics datasets from *** results and statistical analyses validated the outstanding performance of SSAFS compared to existing state-of-the-art algorithms,confirming the superiority of SSAFS in exploring the feature space and identifying the most valuable features for diseased plant image *** computational tool will allow us to explore an optimal combination of handcrafted features to improve plant disease recognition accuracy and processing time.

关键词： Plant image redundant

来源：评论

学校读者我要写书评

暂无评论

LinNet: Linear Network for Efficient Point Cloud Representation Learning 38

LinNet: Linear Network for Efficient Point Cloud Representat...

引用

38th Conference on Neural Information Processing Systems, NeurIPS 2024

作者： Deng, Hao Jing, Kunlei Cheng, Shengmei Liu, Cheng Ru, Jiawei Bo, Jiang Wang, Lin State-Province Joint Engineering and Research Center of Advanced Networking and Intelligent Information Services School of Information Science and Technology Northwest University China Shaanxi Key Laboratory of Higher Education Institution of Generative Artificial Intelligence and Mixed Reality China School of Software Engineering Xi'an Jiaotong University China Department of Computing The Hong Kong Polytechnic University Hong Kong

Point-based methods have made significant progress, but improving their scalability in large-scale 3D scenes is still a challenging problem. In this paper, we delve into the point-based method and develop a simpler, faster, stronger variant model, dubbed as LinNet. In particular, we first propose the disassembled set abstraction (DSA) module, which is more effective than the previous version of set abstraction. It achieves more efficient local aggregation by leveraging spatial anisotropy and channel anisotropy separately. Additionally, by mapping 3D point clouds onto 1D space-filling curves, we enable parallelization of downsampling and neighborhood queries on GPUs with linear complexity. LinNet, as a purely point-based method, outperforms most previous methods in both indoor and outdoor scenes without any extra attention, and sparse convolution but merely relying on a simple MLP. It achieves the mIoU of 73.7%, 81.4%, and 69.1% on the S3DIS Area5, NuScenes, and SemanticKITTI validation benchmarks, respectively, while speeding up almost 10x times over PointNeXt. Our work further reveals both the efficacy and efficiency potential of the vanilla point-based models in large-scale representation learning. Our code will be available at https://***/DengH293/LinNet. © 2024 Neural information processing systems foundation. All rights reserved.

关键词：

来源：评论

学校读者我要写书评

暂无评论

An Economic Framework for 6-DoF Grasp Detection

arXiv

引用

arXiv 2024年

作者： Wu, Xiao-Ming Cai, Jia-Feng Jiang, Jian-Jian Zheng, Dian Wei, Yi-Lin Zheng, Wei-Shi School of Computer Science and Engineering Sun Yat-sen University China Peng Cheng Laboratory Shenzhen China Key Laboratory of Machine Intelligence and Advanced Computing Ministry of Education China

Robotic grasping in clutters is a fundamental task in robotic manipulation. In this work, we propose an economic framework for 6-DoF grasp detection, aiming to economize the resource cost in training and meanwhile maintain effective grasp performance. To begin with, we discover that the dense supervision is the bottleneck of current SOTA methods that severely encumbers the entire training overload, meanwhile making the training difficult to converge. To solve the above problem, we first propose an economic supervision paradigm for efficient and effective grasping. This paradigm includes a well-designed supervision selection strategy, selecting key labels basically without ambiguity, and an economic pipeline to enable the training after selection. Furthermore, benefit from the economic supervision, we can focus on a specific grasp, and thus we devise a focal representation module, which comprises an interactive grasp head and a composite score estimation to generate the specific grasp more accurately. Combining all together, the EconomicGrasp framework is proposed. Our extensive experiments show that EconomicGrasp surpasses the SOTA grasp method by about 3AP on average, and with extremely low resource cost, for about 1/4 training time cost, 1/8 memory cost and 1/30 storage cost. Our code is available at https://***/iSEE-laboratory/EconomicGrasp. Copyright © 2024, The Authors. All rights reserved.

关键词： Robotics

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：