检索结果-内蒙古大学图书馆

Unveiling factuality and injecting knowledge for LLMs via reinforcement learning and data proportion

Science China(Information Sciences) 2024年第10期67卷 389-390页

作者： Wenjun KE Ziyu SHANG Zhizhao LUO Peng WANG Yikai GUO Qi LIU Yuxuan CHEN School of Computer Science and Engineering Southeast University Key Laboratory of New Generation Artificial Intelligence Technology and Its Interdisciplinary Applications(Southeast University) Beijing Institute of Technology Zhuhai Beijing Institute of Computer Technology and Application

Large language models(LLMs) have demonstrated remarkable effectiveness across various natural language processing(NLP) tasks, as evidenced by recent studies [1, 2]. However, these models often produce responses that conflict with reality due to the unreliable distribution of facts within their training data, which is particularly critical for applications requiring high credibility and accuracy [3].

关键词：

来源：评论

学校读者我要写书评

暂无评论

Obstacle Avoidance for Unmanned Surface Vehicle by Null-Space Guidance Vector field With Deep Reinforcement Learning

引用

IEEE Internet of Things Journal 2025年第13期12卷 24518-24529页

作者： Yao, Peng Liu, Qiming Zhao, Zhiyao Ocean University of China College of Engineering Qingdao266100 China Beijing Technology and Business University School of Computer and Artificial Intelligence Beijing100048 China

This paper proposes an efficient hierarchical solution framework, null-space guidance vector field (NSGVF) with deep reinforcement learning (DRL) based coefficient adjustment, to solve the problem of obstacle avoidance for unmanned surface vehicle (USV) with 3D-LiDAR in an unstructured environment. First, a simulation platform with high fidelity based on Virtual RobotX is developed, including the USV model, ocean environment, perception module, decision making and path planning, tracking controller, etc. With the modulated velocity commands, the NSGVF method enables the USV to navigate towards the target point and avoid obstacles, by defining the modulation matrix to quantify the influence of real-time sampling point, and introducing a null-space strategy to integrate the influence of point clouds. Because the key coefficients of NSGVF need to be set in advance with poor environmental adaptability, the DRL method with curriculum learning is then developed to find the optimum for the key coefficients. The experiment results show that the hierarchical DRL+NSGVF framework can be better applied to the task of obstacle avoidance in unstructured environments. © 2025 IEEE.

关键词： Deep reinforcement learning

来源：评论

学校读者我要写书评

暂无评论

Joint fuzzy background and adaptive foreground model for moving target detection

引用

Frontiers of computer Science 2024年第2期18卷 93-106页

作者： Dawei ZHANG Peng WANG Yongfeng DONG Linhao LI Xin LI School of Artificial Intelligence Hebei University of TechnologyTianjin 300401China Lane Department of Computer Science and Electrical Engineering West Virginia UniversityMorgantownWV 26506-6109USA

Moving target detection is one of the most basic tasks in computer *** conventional wisdom,the problem is solved by iterative optimization under either Matrix Decomposition(MD)or Matrix Factorization(MF)*** utilizes foreground information to facilitate background *** uses noise-based weights to fine-tune the *** both noise and foreground information contribute to the recovery of the *** jointly exploit their advantages,inspired by two framework complementary characteristics,we propose to simultaneously exploit the advantages of these two optimizing approaches in a unified framework called Joint Matrix Decomposition and Factorization(JMDF).To improve background extraction,a fuzzy factorization is *** fuzzy membership of the background/foreground association is calculated during the factorization process to distinguish their contributions of both to background *** describe the spatio-temporal continuity of foreground more accurately,we propose to incorporate the first order temporal difference into the group sparsity constraint *** temporal constraint is adjusted *** foreground and the background are jointly estimated through an effective alternate optimization process,and the noise can be modeled with the specific probability *** experimental results of vast real videos illustrate the effectiveness of our *** with the current state-of-the-art technology,our method can usually form the clearer background and extract the more accurate ***-noise experiments show the noise robustness of our method.

关键词： matrixdecomposition matrix factorization generalized sparsity noise modeling

来源：评论

学校读者我要写书评

暂无评论

Survey on Deep Learning-based Speech Technologies in Voice Chatbot Systems

引用

KSII Transactions on Internet and Information Systems 2025年第5期19卷 1406-1440页

作者： Ma, Seunghee Oh, Junseok Kim, Minseo Kim, Ji-Hwan Dept. of Artificial Intelligence Sogang University Seoul Korea Republic of Dept. of Computer Science and Engineering Sogang University Seoul Korea Republic of Dept. of Artificial Intelligence Technology LOTTE INNOVATE Company Seoul Korea Republic of

Recent advancements in large language models (LLMs) such as ChatGPT have contributed to the development of chatbot systems. Specifically, speech has been recognized as the optimal tool for interactive dialogue, leading to increased interest in voice chatbots. Voice chatbots offer information and services through voice interactions, enhancing user experience and improving service accessibility. This survey paper introduces the latest developments in the core technologies of voice chatbot systems, including deep learning-based automatic speech recognition, speech synthesis, and speech emotion recognition. It focuses on advanced research to enhance speed and performance, which is crucial for applying speech technologies in voice chatbots. In automatic speech recognition, we introduce methodologies such as Connectionist Temporal Classification (CTC), Attention based Encoder-Decoder (AED), and Recurrent Neural Network Transducer (RNN-T), along with studies optimizing Transformer-based models for real-time automatic speech recognition. In speech emotion recognition, we explore the use of pre-trained models and the latest techniques for accurate emotion prediction. For speech synthesis, the focus extends to two-stage and End-to-End (E2E) approaches, with additional research on integrating emotional information to generate natural speech. Copyright © 2025 KSII.

关键词： Chatbots

来源：评论

学校读者我要写书评

暂无评论

LGDF-Net: Local and Global Feature-Based Dual-Branch Fusion Networks for Deepfake Detection

引用

IEEE Transactions on Circuits and Systems for Video technology 2025年第6期35卷 5489-5500页

作者： Long, Min Liu, Zhenyu Zhang, Le-Bing Peng, Fei School of Electronics and Communication Engineering Guangzhou University Guangdong Guangzhou510006 China School of Computer and Communication Engineering Changsha University of Science and Technology Changsha410114 China School of Computer and Artificial Intelligence Huaihua University Huaihua418000 China School of Artificial Intelligence Guangzhou University Guangdong Guangzhou510006 China

With the rapid development of Deepfake technology, social security is facing great challenges. Although numerous Deepfake detection algorithms based on traditional CNN frameworks perform well on specific datasets, they still suffer from overfitting due to an over-reliance on localized artifact information. This limitation leads to degraded detection performance across diverse datasets. To address this issue, this study proposes a dual-branch fusion network called LGDF-Net. LGDF-Net uses a dual-branch structure to process the local artifact features and global texture features generated by Deepfake separately, preserving their unique characteristics. Specifically, the local compression branch utilizes a specially designed local compression module (LCM) that allows the network to focus more accurately on key regions of localized artifacts in Deepfake faces. The global expansion branch enhances the analysis of the global facial context through a global expansion module (GEM), which captures image context information and subtle texture features more comprehensively. Additionally, the proposed multi-scale feature extraction module (MSFE) delves into image features at various scales, enriching the extraction of detailed information. Finally, the multi-level feature fusion strategy (MLFF) improves the integration of local and global features through multiple layers, enabling the network to learn the intrinsic connections between these two types of features. A series of experimental validations demonstrate that the proposed scheme outperforms many existing detection networks in terms of accuracy and generalization ability. © 2025 IEEE. All rights reserved, including rights for text and data mining, and training of artificial intelligence and similar technologies.

关键词： Feature extraction

来源：评论

学校读者我要写书评

暂无评论

A novel approach for image retrieval in remote sensing using vision-language-based image caption generation

引用

Multimedia Tools and Applications 2025年第6期84卷 2985-3014页

作者： Yadav, Prem Shanker Tyagi, Dinesh Kumar Vipparthi, Santosh Kumar Department of Computer Science and Engineering Malaviya National Institute of Technology Rajasthan Jaipur302017 India School of Artificial Intelligence and Data Engineering Indian Institute of Technology Ropar Punjab Rupnagar140001 India

Recent advancements in satellite technologies have resulted in the emergence of Remote Sensing (RS) images. Hence, the primary imperative research domain is designing a precise retrieval model for retrieving the most pertinent images based on the query. Present Remote Sensing Image Retrieval (RSIR) systems use visual descriptors to characterize the primitives (such as various land-cover types) that are visible in the images. However, the visual descriptors are inadequate for defining the complicated content of RS images. To solve this problem, a new model is devised for image retrieval based on image captions. The goal is to generate textual illustrations with captions to define relations amongst objects precisely. Here, image captioning is attained based on the vision-language pre-training model. The image captions are utilized for generating features like term frequency-inverse document frequency (TF-IDF), length of text, and Bag of Words. Meanwhile, query text is utilized wherein features like TF-IDF, text length, and Bag of Words are obtained. The similarity between query text features and the image captions features has been computed on the basis of a hybrid similarity measure wherein weights are tuned with the proposed Honey Badger Political Optimizer (HBPO) to retrieve the image. The proposed HBPO provided enhanced efficiency with elevated precision of 93.3%, recall of 93.7%, F1-score of 93.5%, and Recall-Oriented Understudy for Gisting Evaluation (ROUGE) of 0.441. © The Author(s), under exclusive licence to Springer Science+Business Media, LLC, part of Springer Nature 2024.

关键词： Visual languages

来源：评论

学校读者我要写书评

暂无评论

A Variational Inference-Based LSTM-Enhanced Deep Neural Model for Sequential Recommendations

引用

IEEE Access 2025年 13卷 95945-95962页

作者： Ngaffo, Armielle Noulapeu Kamgang, Inès Raïssa Djouela Maka, Ebenezer Maka Malong, Yannick University of Douala Laboratory of Computer Science Data Science and Artificial Intelligence National Higher Polytechnic School of Douala Douala Cameroon University of Buea Faculty of Engineering and Technology Department of Computer Engineering Buea Cameroon

Long Short-Term Memory (LSTM) networks are particularly useful in recommender systems since user preferences change over time. Unlike traditional recommender models which assume static user-item interactions, LSTM models capture sequential dependencies and temporal dynamics, making them ideal for personalized, session-based, and sequential recommendation tasks. However, traditional LSTM-based recommender systems do not model uncertainty since they perform deterministic predictions. Moreover, those recommender models exhibit poor generalization on small datasets, and poor-handle cold-start and data sparsity problems. To overcome those limitations, this paper presents an LSTM-based recommender model that leverages an improved deep matrix factorization to accurately address the data sparsity problem. In addition, an enhanced variational inference coupled to the Evidence Lower Bound Optimization is applied on LSTM unit layers to perform uncertainty-aware predictions. Thereafter, latent vectors obtained from LSTM networks are effectively involved in the sequential recommendation process. Experiments have been conducted on several real-world datasets to evaluate our proposal and the results show the proposed model presents significant performances compared to state-of-the-art models. © 2013 IEEE.

关键词： Deep neural networks

来源：评论

学校读者我要写书评

暂无评论

White blood cell segmentation using U-Net and its variants to improve leukemia diagnosis

引用

Neural Computing and Applications 2025年第5期37卷 3265-3286页

作者： Joshi, Vivek C. Mehta, Mayuri A. Kotecha, Ketan Ahmedabad India Department of Computer Engineering Sarvajanik College of Engineering and Technology Surat India Symbiosis Centre for Applied Artificial Intelligence Symbiosis Institute of Technology Symbiosis International University Pune411045 India

Numerous computer-aided leukemia detection methods have been introduced to overcome the limitations of clinical diagnosis procedures. The precision of computer-aided leukemia detection highly depends on the accurate segmentation of white blood cells (WBCs) from the stained whole slide image (WSI). This paper proposes WBC segmentation from WSI using U-Net and its variants. The major contributions to this paper are as follows. First, WBC segmentation is proposed using U-Net, U-Net++ and three transfer learning-based U-Net models, namely U-Net-VGG16, U-Net-VGG19, and U-Net-ResNet. Second, a comprehensive and comparative experimental analysis of the proposed WBC segmentation approaches is presented using four evaluation parameters, namely Dice coefficient, Intersection over Union (IoU), precision, and recall. Third, WBC segmentation approaches are also evaluated using four loss functions such as binary cross-entropy, focal, Dice, and IoU to identify the most effective loss function for WBC segmentation. Finally, the trained models’ ability to perform WBC segmentation is validated to assess their practical applicability. The empirical results reveal that U-Net-VGG16 and U-Net-VGG19 achieve a high Dice coefficient of 90% and IoU of 83%. Besides, U-Net++ achieves high precision and recall of 98% and 70%, respectively. Although the results reveal that transfer learning-based U-Net models perform better, deployment of the trained model shows that the U-Net segments WBC more precisely than transfer learning-based U-Net models. © The Author(s), under exclusive licence to Springer-Verlag London Ltd., part of Springer Nature 2024.

关键词： Image segmentation

来源：评论

学校读者我要写书评

暂无评论

Impact of asymptomatic infected individuals on epidemic transmission dynamics in multiplex networks with partial coupling

引用

Chinese Physics B 2024年第10期33卷 80-87页

作者： Xin Hu Jiaxing Chen Chengyi Xia School of Control Science and Engineering Tiangong UniversityTianjin 300387China School of Computer Science and Engineering Tianjin University of TechnologyTianjin 300384China School of Artificial Intelligence Tiangong UniversityTianjin 300387China

The theory of network science has attracted great interest of many researchers in the realm of biomathematics and public health,and numerous valuable epidemic models have been *** previous studies,it is common to set up a one-to-one correspondence between the nodes of a multi-layer network,ignoring the more complex situations in *** the present work,we explore this situation by setting up a partially coupled model of a two-layer network and investigating the impact of asymptomatic infected individuals on *** propose a self-discovery mechanism for asymptomatic infected individuals,taking into account situations such as nucleic acid testing in the community and individuals performing self-antigen testing during the *** these factors together,through the microscopic Markov chain approach(MMCA)and extensive Monte Carlo(MC)numerical simulations,we find that the greater the coupling between the networks,the more information dissemination is *** order to control the epidemics,more asymptomatic infected individuals should be made aware of their *** adoption of nucleic acid testing and individual adoption of antigenic self-testing can help to contain epidemic ***,the epidemic threshold of the proposed model is derived,and then miscellaneous factors affecting the epidemic threshold are also *** results are conducive to devising the prevention and control policies of pandemics.

关键词： asymptomatic infected individuals multi-layer networks partial interdependence

来源：评论

学校读者我要写书评

暂无评论

DFS-Net: StyleGAN2-based dual feature separation for face De-Morphing

引用

Multimedia Tools and Applications 2024年 1-23页

作者： Long, Ming Chen, Song Zhang, Le-Bing Peng, Fei Zhang, Dengyong School of Computer and Communication Engineering Changsha University of Science and Technology Changsha410114 China School of Computer and Artificial Intelligence Huaihua University Huaihua418000 China Institute of Artificial Intelligence and Blockchain Guangzhou University Guangzhou510006 China

Face morphing attacks pose a significant threat to face recognition systems. In order to solve this problem, several methods for detecting these attacks have been proposed. However, the restoration of the accomplice’s face image remains in its nascent developmental stage. In this paper, we introduce a novel network architecture termed DFS-Net, which leverages double feature spaces (latent code and feature tensor) and a dual-feature separation (DFS) network. DFS-Net is built upon the StyleGAN2 generator. By utilizing the feature vector and feature tensor extracted by the encoder, distinct separation networks are designed. This design facilitates effective feature separation, enabling DFS-Net to separate the identity features of accomplices. The incorporation of the feature tensor enhances the precision in extracting the identity features of the accomplice and thereby elevates the perceptual quality of the restoration image. Experimental results indicate that DFS-Net can effectively restore the accomplice’s face and outperforms previous works in terms of restoration accuracy and image visual quality. © The Author(s), under exclusive licence to Springer Science+Business Media, LLC, part of Springer Nature 2024.

关键词： Network architecture

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：