检索结果-内蒙古大学图书馆

2023 International Conference on Algorithms, Computing and data Processing, ACDP 2023

作者： Long, Junwei Yi, Xiangbo School of Computer Science and Artificial Intelligence Wuhan University of Technology Wuhan China

ISBN: (纸本)9798350326680

Inspired by the mechanism of batch processing systems, this paper proposes a multi-level and multi-swarm particle swarm optimization (MMPSO) algorithm to alleviate the performance degradation caused by imbalances in exploration and exploitation within the swarm. The algorithm constrains the learning behavior of different swarms by dividing the swarm into multi-level and multi-swarm. Swarms of different levels are grouped into various to perform different tasks. The small-scale in top-level swarm is responsible for exploiting the global optimal position. The large-scale in bottom-level swarm uses the Levy flight strategy based on the improvement rate to explore unknown areas, effectively improving the exploration ability. The middle-level swarm is divided into multi-swarm to explore and exploit local locations and design an internal and external comprehensive learning strategy. A lifting mechanism is proposed to improve the information exchangeability of swarms at different levels. Experimental results on the 30-D of the CEC2017 benchmark optimization problems show that the MMPSO algorithm exhibits the best convergence accuracy on most benchmark functions due to the balance of exploration and exploitation. Furthermore, the algorithm combine with BP Neural Networks (BPNN) shows the best convergence accuracy on the application problem of developing force-field parameters (FFPs). © 2023 IEEE.

关键词： Swarm intelligence

来源：评论

学校读者我要写书评

暂无评论

A Novel Knowledge-Based Neural Network Approach to the Small-Signal Modeling of Packaged Diodes

引用

Progress in Electromagnetics Research C 2024年 149卷 9-14页

作者： Liu, Wenyuan Yang, Ningning Yan, Shuxia Xu, Yanlin School of Electronic Information and Artificial Intelligence Shaanxi University of Science and Technology Xi’an710021 China School of Electronics and Information Engineering Tiangong University Tianjin300387 China School of Electronic Science and Technology National University of Defense Technology Changsha410073 China

This paper proposes a novel knowledge-based neural network approach that, in the absence of specific device SPICE models, can utilize the measured data of actual diode devices to map the existing diode coarse model to a more accurate packaged model through neural network mapping techniques, thereby achieving precise and efficient modeling of the small-signal characteristics of diode devices. A knowledge-based neural network model for packaged diodes is proposed, which enhances modeling accuracy by learning the discrepancies between the diode coarse model and the actual device data. A training method for rapid parameter adjustment is suggested, where the neural networks within the input and output package modules automatically learn and adjust, continuously optimizing their internal parameters to enhance modeling efficiency. Modeling experiments conducted on the measurement data of the MA4AGFCP910 diode show that the proposed packaged diode model can effectively and accurately match the small-signal characteristic data of the diode device. © 2024, Electromagnetics Academy. All rights reserved.

关键词： PIN diodes

来源：评论

学校读者我要写书评

暂无评论

PRESS: Defending Privacy in Retrieval-Augmented Generation via Embedding Space Shifting

PRESS: Defending Privacy in Retrieval-Augmented Generation v...

引用

2025 IEEE International Conference on Acoustics, Speech, and Signal Processing, ICASSP 2025

作者： He, Jiaming Liu, Cheng Hou, Guanyu Jiang, Wenbo Li, Jiachen School of Computer Science and Engineering University of Electronic Science and Technology of China China College of Computer Science and Cyber Security Oxford Brookes College Chengdu University of Technology China School of Data Science The Chinese University of Hong Kong Shenzhen China School of Computer Science and Artificial Intelligence Wuhan University of Technology China

ISBN: (纸本)9798350368741

Retrieval-augmented generation (RAG) expands the capabilities of large language models (LLMs) in various applications by integrating relevant information retrieved from external data sources. However, the RAG systems are exposed to substantial privacy risks during the information retrieval process, leading to potential data leakage of private information. In this work, we present a Privacy-preserving Retrieval-augmented generation via Embedding Space Shifting (PRESS), systematically exploring how to protect privacy in RAG systems. Specifically, we first conduct proximal policy optimization (PPO) based training on pre-trained language models to generate target training samples. Then we employ a purposive shift fine-tuning on the text embedding model with the generated samples for guiding the RAG system to map potential privacy leaking queries to safe target in embedding space. Extensive experimental results on representative models and datasets demonstrate that our protection method achieves high defense performance with high efficiency while keeping the normal functionality of the RAG system. © 2025 IEEE.

关键词：

来源：评论

学校读者我要写书评

暂无评论

Multicomponent Similarity Graphs for Cross-Network Node Classification

IEEE Transactions on Artificial Intelligence

引用

IEEE Transactions on artificial intelligence 2024年第3期5卷 1411-1424页

作者： Zhang, Yuhong Shi, Congmei Li, Xinzheng Zhang, Zan Hu, Xuegang School of Computer Science and Information Engineering Hefei University of Technology Hefei230601 China Institute of Artificial Intelligence Hefei Comprehensive National Science Center Hefei230088 China Key Laboratory of Knowledge Engineering with Big Data Hefei University of Technology Ministry of Education Hefei230009 China

Cross-network node classification aims to train a classifier for an unlabeled target network using a source network with rich labels. In applications, the degree of nodes mostly conforms to the long-tail distribution, i.e., most nodes in the network are tail nodes with sparse neighborhoods. The established methods focus on either the discrepancy cross network or the long tail in a single network. As for the cross-network node classification under long tail, the coexistence of sparsity of tail nodes and the discrepancy cross-network challenges existing methods for long tail or methods for the cross-network node classification. To this end, a multicomponent similarity graphs for cross-network node classification (MS-CNC) is proposed in this article. Specifically, in order to address the sparsity of the tail nodes, multiple component similarity graphs, including attribute and structure similarity graphs, are constructed for each network to enrich the neighborhoods of the tail nodes and alleviate the long-tail phenomenon. Then, multiple representations are learned from the multiple similarity graphs separately. Based on the multicomponent representations, a two-level adversarial model is designed to address the distribution difference across networks. One level is used to learn the invariant representations cross network in view of structure and attribute components separately, and the other level is used to learn the invariant representations in view of the fused structure and attribute graphs. Extensive experimental results show that the MS-CNC outperforms the state-of-the-art methods. Impact Statement-Node classification is an important task in graph mining. With the unavailability of labels, some researchers propose cross-network node classification, using one labeled network to assist the node classification of another unlabeled network. However, the long-tail of nodes leads to unsatisfactory performance and challenges the recent cross-network node classification m

关键词： Graphic methods

来源：评论

学校读者我要写书评

暂无评论

Prediction of Protein-Protein Interactions Based on Attention Diffusion Mechanism in Heterogeneous Information Network

Prediction of Protein-Protein Interactions Based on Attentio...

引用

2023 IEEE International Conference on Systems, Man, and Cybernetics, SMC 2023

作者： Liu, JinLing Peng, Jing School of Computer Science and Artificial Intelligence Wuhan University of Technology Wuhan China

ISBN: (纸本)9798350337020

Protein-protein interaction (PPI) prediction is a deep exploration of the mechanism of life activities, but it is costly to rely solely on experimental methods to predict PPI. To accomplish this task, many computational methods have been proposed, but existing methods fail to make full use of the connection between proteins and other molecules, and can not effectively capture the complex semantics between biological entities related to proteins, resulting the poor performance. In this paper, we propose a heterogeneous graph attention diffusion network (HGADN-PPI) to capture the complex semantics in the biological heterogeneous graph for PPI prediction. HGADN-PPI enhances heterogeneous graph learning from the intra-layer perspective and the inter-layer perspective. Concretely, in the intra-layer perspective, we constructed the heterogeneous graph attention diffusion layer (HGADL) which combines biological semantic information from different paths. To enable information passing from important nodes multiple hops away, we use graph diffusion to establish connections between nodes that are not directly connected. In the intra-layer perspective, we enlarge the receptive field by stacking multiple HGADLs. The experimental results demonstrate that the proposed model outperforms the selected representative baselines. © 2023 IEEE.

关键词： Attention Diffusion Heterogeneous Graph PPI

来源：评论

学校读者我要写书评

暂无评论

Automatically Suggesting Suitable Crop from Soil Samples Using Optimized Neural Networks 3

Automatically Suggesting Suitable Crop from Soil Samples Usi...

引用

3rd International Conference for Innovation in technology, INOCON 2024

作者： Pasha, Syed Nawaz Ramesh, D. Mohmmad, Sallauddin Deepak, Gujjula School of Computer Science and Artificial Intelligence Sr University Warangal India Kamala Institute of Technology and Science Dept. of Computer Science and Engineering Huzurabad India

ISBN: (纸本)9798350381931

A soil-centric crop recommendation system recommends suitable crops utilizing vital soil attributes as input factors. This sophisticated system incorporates parameters like Nitrogen (N), Phosphorus (P), Potassium (K) levels, temperature, humidity, pH, and rainfall to suggest ideal crops for cultivation. The dataset comprises diverse crops and their specific requirements for each input parameter. We implemented an optimized neural network to provide precise crop recommendations based on the provided input values. By analyzing this data, the system provides farmers with the information necessary to make well-informed decisions regarding their choice of crops, enhancing the potential for optimal yields. Our model performed well on training and testing data with an accuracy of 0.94. These results demonstrate its effectiveness and potential as a valuable tool for agricultural decision-making. This system offers farmers tailored recommendations based on soil properties, climatic conditions, and other environmental factors. © 2024 IEEE.

关键词： Crops

来源：评论

学校读者我要写书评

暂无评论

Oil Spill Detection on Sea Surface Based on SAR-Images by Using Global Context and Edge Perception 3

Oil Spill Detection on Sea Surface Based on SAR-Images by Us...

引用

3rd International Conference on Electronic Information Engineering and computer Communication, EIECC 2023

作者： Rong, Haitao Chen, Xianqiao Xie, Kai Wang, Qirui School of Computer Science and Artificial Intelligence Wuhan University of Technology Wuhan China

ISBN: (纸本)9798350359961

Synthetic Aperture Radar (SAR) has been widely used in marine environmental monitoring. However, there are some challenges in the detection of oil spills based on SAR images. Existing oil spill detection methods inadequately consider the characteristics of SAR images, such as high noise, complex texture, and blurry boundaries, resulting in poor edge accuracy and false positives and negatives in oil spill region detection. In response to these issues, this paper proposes a sea surface oil spill detection model based on global context and edge awareness. Firstly, the model enhances the extraction and utilization of edge information by designing an edge-Aware detection branch network. Secondly, improvements are made to the feature extraction method of the encoder. The backbone network is enhanced for capturing low-level feature information of oil spills by introducing the CA (Channel Attention) module. The use of a dense-connected spatial pyramid module expands the receptive field to enhance global context awareness, thereby improving the extraction of high-level semantic features from oil spill features. Finally, through comparative experiments, the effectiveness of the proposed detection model in this chapter is validated. © 2023 IEEE.

关键词： attention mechanism edge perception oil spill detection SAR images

来源：评论

学校读者我要写书评

暂无评论

Research and Design of a Ceramic Tile Defect Detection Edge Device Utilizing YOLOv5 and MobileNetV3 6

Research and Design of a Ceramic Tile Defect Detection Edge ...

引用

6th IEEE International Conference on Automation, Electronics and Electrical Engineering, AUTEEE 2023

作者： Chen, Yuejun Wuhan University of Technology School of Computer Science and Artificial Intelligence Wuhan China

ISBN: (纸本)9798350305623

This investigation is dedicated to developing an advanced ceramic tile defect detection system based on YOLOv5 and MobileNetV3, aimed at augmenting the efficiency and precision of quality control in industrial tile production. The research is grounded in the pressing need for automated and precise technologies in the field of industrial quality inspection. Initially, the paper outlines the potential of deep learning algorithms in detecting flaws in ceramic tiles and identifies YOLOv5 and MobileNetV3 as the pivotal models. The methodology encompasses the integration of these models and their specific optimization for tile defect detection tasks. Key findings indicate that the amalgamation of these models significantly enhances detection efficiency and accuracy, particularly in resource-constrained edge computing environments. Moreover, the study underscores the potential applications of deep learning in industrial automation, providing valuable data and insights for future research. The unique contribution of this research lies in its innovative exploration of real-time, efficient methods for detecting defects in ceramic tiles. © 2023 IEEE.

关键词： Ceramic Tile Defect Detection Deep Learning Industrial Automation MobileNetV3 YOLOv5

来源：评论

学校读者我要写书评

暂无评论

Incremental Object Detection With Image-Level Labels

IEEE Transactions on Artificial Intelligence

引用

IEEE Transactions on artificial intelligence 2024年第5期5卷 2331-2341页

作者： Yu, Qianzi Zhu, Kai Wang, Wen Cao, Yang Kang, Yu University of Science and Technology of China Department of Automation School of Information Science and Technology Hefei230026 China Hefei Comprehensive National Science Center Institute of Artificial Intelligence Hefei230039 China Zhejiang University College of Computer Science and Technology State Key Lab of CAD&CG Hangzhou310027 China University of Science and Technology of China School of Information Science and Technology Hefei230026 China

Incremental object detection (IOD) aims to achieve simultaneous prediction of old and new samples on localization and classification when new concepts are provided. It is a challenging task due to the need for a joint interpretation of semantic and spatial information. While the existing work accomplishes the detector generalization with the help of annotated sample classes and bounding boxes, this article presents one astonishing finding on the unnecessity of new bounding boxes, which will significantly reduce the annotation cost and condition constraints in IOD. To enable the incremental detection process with image-level labels, we propose a multibranch decoupling scheme in which the representation, classification, and regression branches are customized to accommodate new concepts with class labels with different semantic levels. The regression branch is first frozen to ensure generalizable localization while maintaining the stability of representation optimization. Then, the representation branch is gradient-inverted to estimate the distribution drift from the old object-level set to the new image-level one, improving the effectiveness of feature updates in the absence of homogeneous semantic and spatial supervision. Finally, the classification branch is calibrated to balance the contribution of different images to the gradient, in which the frequency ratios of old instances and new images are calculated as a postprocessing factor for the logits margin. Extensive experiments on two standard incremental detection benchmarks demonstrate the above-par recognition and localization performance while reducing the burden on new class annotation, outperforming state-of-the-art methods by 2% and 3%, respectively. © 2023 IEEE.

关键词： Feature extraction

来源：评论

学校读者我要写书评

暂无评论

Boosting Few-Shot Image Classification with Feature Map Reconstruction Network via Statistical Information 23

Boosting Few-Shot Image Classification with Feature Map Reco...

引用

15th International Conference on Machine Learning and Computing, ICMLC 2023

作者： Huo, Yongkang School of Computer Science and Artificial Intelligence Wuhan University of Technology Wuhan China

ISBN: (纸本)9781450398411

Few-shot image classification aims to recognize classes with limited labeled data. Many works have been proposed to solve this problem. Recently, the feature map reconstruction network has attracted great attention. However, due to the scarcity of data, it is difficult to find an exact solution to the reconstructed model. We try to use statistical information to draw the support set distribution and query set distribution closer, which helps the model obtain a precise solution. We consider first-order information, calculate the mean of each set and eliminate the bias between them. In addition to one-dimensional information, high-dimensional information (i.e. the covariance matrix) is used to represent the data distribution and make Euclidean distance of them as the loss. Our statistical information feature reconstruction network(SI-FRN) does not bring in any other supervised information. Compared to recent state-of-the-art few-shot image classification methods with reconstruction, our SI-FRN gets excellent performance on four benchmarks: mini-ImageNet, tiered-ImageNet, CUB and Aircraft. © 2023 ACM.

关键词： computer vision

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：