检索结果-内蒙古大学图书馆

2024 International Applied Computational Electromagnetics Society Symposium, ACES-China 2024

作者： Hu, Wangchao Xie, Guoda Dai, Jinpeng Hou, Guilin Li, Xiang Huang, Zhixiang Anhui University The Key Laboratory of Intelligent Computing and Signal Processing Hefei China Anhui Province Key Laboratory of Target Recognition and Feature Extraction Education of Anhui Province Luan237000 China Ministry of Education The Key Laboratory of Intelligent Computing and Signal Processing Hefei China Anhui Province Anhui University Key Laboratory of Electromagnetic Environmental Sensing Hefei230601 China Anhui University Hefei China

ISBN: (纸本)9798350355581

In this paper, a deep learning optimization method combining U-net model and cycle generative adversarial network (Cycle-GAN) is proposed to efficiently solve the electromagnetic inverse scattering (EMIS) problems. Firstly, a U-net model is used to obtain a rough preliminary reconstruction. Through down-sampling and up-sampling processes, the basic features of the object are extracted and reconstructed from the measured scattered field, providing a coarse dielectric parameter distribution. Then, a Cycle-GAN model further refines the output of the U-net. The Cycle-GAN model, through its adversarial training mechanism involving a generator and a discriminator, enhances the details of the preliminary reconstruction, resulting in a finely detailed dielectric parameter distribution his method improves the problem of rough results obtained by using U-net method alone, and can effectively solve and optimize EMIS problems. © 2024 IEEE.

关键词： Generative adversarial networks

来源：评论

学校读者我要写书评

暂无评论

A Gradient-Guided Evolutionary Approach to Training Deep Neural Networks

引用

IEEE TRANSACTIONS ON NEURAL NETWORKS AND LEARNING SYSTEMS 2022年第9期33卷 4861-4875页

作者： Yang, Shangshang Tian, Ye He, Cheng Zhang, Xingyi Tan, Kay Chen Jin, Yaochu Key Laboratory of Intelligent Computing and Signal Processing of Ministry of Education School of Computer Science and Technology Anhui University Hefei China Key Laboratory of Intelligent Computing and Signal Processing of Ministry of Education Institutes of Physical Science and Information Technology Anhui University Hefei China Department of Computer Science and Engineering Guangdong Provincial Key Laboratory of Brain-Inspired Intelligent Computation Southern University of Science and Technology Shenzhen China Department of Computing The Hong Kong Polytechnic University Hong Kong SAR Department of Computer Science University of Surrey Guildford U.K

It has been widely recognized that the efficient training of neural networks (NNs) is crucial to classification performance. While a series of gradient-based approaches have been extensively developed, they are criticized for the ease of trapping into local optima and sensitivity to hyperparameters. Due to the high robustness and wide applicability, evolutionary algorithms (EAs) have been regarded as a promising alternative for training NNs in recent years. However, EAs suffer from the curse of dimensionality and are inefficient in training deep NNs (DNNs). By inheriting the advantages of both the gradient-based approaches and EAs, this article proposes a gradient-guided evolutionary approach to train DNNs. The proposed approach suggests a novel genetic operator to optimize the weights in the search space, where the search direction is determined by the gradient of weights. Moreover, the network sparsity is considered in the proposed approach, which highly reduces the network complexity and alleviates overfitting. Experimental results on single-layer NNs, deep-layer NNs, recurrent NNs, and convolutional NNs (CNNs) demonstrate the effectiveness of the proposed approach. In short, this work not only introduces a novel approach for training DNNs but also enhances the performance of EAs in solving large-scale optimization problems.

关键词： Artificial neural networks Training Optimization Search problems Scalability Complexity theory Genetic algorithms Deep neural networks (DNNs) evolutionary algorithm (EA) genetic operator gradient descent sparsity

来源：评论

学校读者我要写书评

暂无评论

Face Forgery Detection Based on Fine-Grained Clues and Noise Inconsistency

IEEE Transactions on Artificial Intelligence

引用

IEEE Transactions on Artificial Intelligence 2025年第1期6卷 144-158页

作者： Zhang, Dengyong He, Ruiyi Liao, Xin Li, Feng Chen, Jiaxin Yang, Gaobo Changsha University of Science and Technology Hunan Provincial Key Laboratory of Intelligent Processing of Big Data on Transportation the School of Computer and Communication Engineering Changsha410114 China Hunan University Changsha410082 China

Deepfake detection has gained increasing research attention in media forensics, and a variety of works have been produced. However, subtle artifacts might be eliminated by compression, and the convolutional neural networks (CNNs)-based detectors are invalidated for fake face images with compression. In this work, we propose a two-stream network for deepfake detection. We observed that high-frequency noise features and spatial features are inherently complementary to each other. Thus, both spatial features and high-frequency noise features are exploited for face forgery detection. Specifically, we design a double-frequency transformer module (DFTM) to guide the learning of spatial features from local artifact regions. To effectively fuse spatial features and high-frequency noise features, a dual-domain attention fusion module (DDAFM) is designed. We also introduce a local relationship constraint loss, which requires only image-level labels, for model training. We evaluate the proposed approach on five large-scale benchmark datasets, and extensive experimental results demonstrate the proposed approach outperforms most SOTA works. © 2024 IEEE.

关键词： Frequency domain analysis

来源：评论

学校读者我要写书评

暂无评论

Simulation extractable SNARKs based on target linearly collision-resistant oracle

引用

Science China(Technological Sciences) 2024年第9期67卷 2853-2866页

作者： WANG LiGuan LI Yuan ZHANG ShuangJun CAI DongLiang KAN HaiBin Shanghai Key Laboratory of Intelligent Information Processing School of Computer ScienceFudan UniversityShanghai 200433China Shanghai Engineering Research Center of Blockchain Shanghai 200433China Yiwu Research Institute of Fudan University Yiwu 322000China

The famous zero-knowledge succinct non-interactive arguments of knowledge(zk-SNARK) was proposed by Groth in ***, the construction is based on quadratic arithmetic programs which are highly efficient concerning the proof length and the verification complexity. Since then, there has been much progress in designing zk-SNARKs, achieving stronger security,and simulated extractability, which is analogous to non-malleability and has broad applications. In this study, following Groth's pairing-based zk-SNARK, a simulation extractability zk-SNARK under the random oracle model is constructed. Our construction relies on a newly proposed property named target linearly collision-resistant, which is satisfied by random oracles under discrete logarithm assumptions. Compared to the original Groth16 zk-SNARK, in our construction, both parties are allowed to use such a random oracle, aiming to get the same random number. The resulting proof consists of 3 group elements and only 1 pairing equation needs to be verified. Compared to other related works, our construction is shorter in proof length and simpler in verification while preserving simulation extractability. The results also extend to achieve subversion zero-knowledge SNARKs.

关键词： quadratic arithmetic program simulation extractability subversion zero-knowledge succinct non-interactive arguments of knowledge target linearly collision-resistant

来源：评论

学校读者我要写书评

暂无评论

Exploring the Compositional Deficiency of Large Language Models in Mathematical Reasoning Through Trap Problems

Exploring the Compositional Deficiency of Large Language Mod...

引用

2024 Conference on Empirical Methods in Natural Language processing, EMNLP 2024

作者： Zhao, Jun Tong, Jingqi Mou, Yurong Zhang, Ming Zhang, Qi Huang, Xuanjing School of Computer Science Fudan University China Shanghai Key Laboratory of Intelligent Information Processing Fudan University China

ISBN: (纸本)9798891761643

Human cognition exhibits systematic compositionality, the algebraic ability to generate infinite novel combinations from finite learned components, which is the key to understanding and reasoning about complex logic. In this work, we investigate the compositionality of large language models (LLMs) in mathematical reasoning. Specifically, we construct a new dataset MATHTRAP‡ by introducing carefully designed logical traps into the problem descriptions of MATH and GSM8K. Since problems with logical flaws are quite rare in the real world, these represent "unseen" cases to LLMs. Solving these requires the models to systematically compose (1) the mathematical knowledge involved in the original problems with (2) knowledge related to the introduced traps. Our experiments show that while LLMs possess both components of requisite knowledge, they do not spontaneously combine them to handle these novel cases. We explore several methods to mitigate this deficiency, such as natural language prompts, few-shot demonstrations, and fine-tuning. Additionally, we test the recently released OpenAI o1 model and find that human-like 'slow thinking' helps improve the compositionality of LLMs. Overall, systematic compositionality remains an open challenge for large language models. © 2024 Association for Computational Linguistics.

关键词： Computational linguistics

来源：评论

学校读者我要写书评

暂无评论

Hierarchical Optimization Method for Federated Learning with Feature Alignment and Decision Fusion

引用

computers, Materials & Continua 2024年第10期81卷 1391-1407页

作者： Ke Li Xiaofeng Wang Hu Wang College of Computer Science and Engineering North Minzu UniversityYinchuan750021China The Key Laboratory of Images&Graphics Intelligent Processing of State Ethnic Affairs Commission North Minzu UniversityYinchuan750021China

In the realm of data privacy protection,federated learning aims to collaboratively train a global ***,heterogeneous data between clients presents challenges,often resulting in slow convergence and inadequate accuracy of the global *** shared feature representations alongside customized classifiers for individual clients emerges as a promising personalized ***,previous research has frequently neglected the integration of global knowledge into local representation learning and the synergy between global and local classifiers,thereby limiting model *** tackle these issues,this study proposes a hierarchical optimization method for federated learning with feature alignment and the fusion of classification decisions(FedFCD).FedFCD regularizes the relationship between global and local feature representations to achieve alignment and incorporates decision information from the global classifier,facilitating the late fusion of decision outputs from both global and local ***,FedFCD employs a hierarchical optimization strategy to flexibly optimize model *** experiments on the Fashion-MNIST,CIFAR-10 and CIFAR-100 datasets,we demonstrate the effectiveness and superiority of *** instance,on the CIFAR-100 dataset,FedFCD exhibited a significant improvement in average test accuracy by 6.83%compared to four outstanding personalized federated learning ***,extended experiments confirm the robustness of FedFCD across various hyperparameter values.

关键词： Federated learning data heterogeneity feature alignment decision fusion hierarchical optimization

来源：评论

学校读者我要写书评

暂无评论

Improving Speech Perceptual Quality and Intelligibility Through Sub-band Temporal Envelope Characteristics 18th

Improving Speech Perceptual Quality and Intelligibility Thr...

引用

18th National Conference on Man-Machine Speech Communication, NCMMSC 2023

作者： Wu, Ruilin Huang, Zhihua Song, Jingyi Liang, Xiaoming School of Computer Science and Technology Xinjiang University Urumqi China Key Laboratory of Signal Detection and Processing in Xinjiang Urumqi China

ISBN: (纸本)9789819706006

In the speech enhancement (SE) model, using auxiliary loss based on acoustic parameters can improve enhancement effects. However, currently used acoustic parameters focus on frequency domain information, neglecting the importance of time domain information. This study explores the effects of training the SE model utilizing temporal envelope parameters as auxiliary losses on perceptual quality and intelligibility, primarily focusing on accurately extracting temporal envelope parameters and optimizing losses. Initially, the effectiveness and robustness of the temporal envelope parameters in optimizing the SE model were verified. Subsequently, a temporal envelope loss based on sub-band weighted (ENVLoss) is proposed based on the perceptual characteristics of human hearing. Then, a method for constructing a joint loss function is proposed, integrating temporal and spectral acoustic features to promote the enhancement effect. Experimental results show that temporal envelope characteristics improve both the time domain and the time-frequency domain SE models. Compared to other acoustic parameter losses, the SE model using a sub-band temporal envelope auxiliary loss shows improvement in the PESQ, STOI, and MOS estimation metrics. © The Author(s), under exclusive license to Springer Nature Singapore Pte Ltd. 2024.

关键词： Speech intelligibility

来源：评论

学校读者我要写书评

暂无评论

MSFSAN: A Novel Multi-Scale Spatio-Temporal Feature Screening Attention Network for Urban Carbon Emission Prediction

MSFSAN: A Novel Multi-Scale Spatio-Temporal Feature Screenin...

引用

2024 IEEE International Conference on Systems, Man, and Cybernetics, SMC 2024

作者： Wang, Ben Qin, Xizhong Qin, Jiwei Zhang, Xiaoyu Ma, Haodong College of Computer Science and Technology Xinjiang University Xinjiang Key Laboratory of Signal Detection and Processing Ürümqi830046 China

ISBN: (纸本)9781665410205

In order to cope with the increasingly severe global energy conservation and emission reduction problems, research on urban carbon emission prediction is of great significance. The existing methods mainly use time series analysis to predict urban carbon emission, but there is a strong spatial correlation between the carbon emission data of several cities. Therefore, this paper designs a multi-scale spatial-temporal feature screening attention network to predict target cities' future carbon emission data. Firstly, this paper combines the daily carbon emission data of the near-neighbouring cities and the daily homologous emission data of the target city to analyze the urban carbon emission data from a spatio-temporal perspective. Then, this paper designs a multi-scale spatial interactive convolution module and a multi-scale temporal convolution module to extract multi-scale spatio-temporal features effectively. In addition, the feature screening module is designed to reduce the adverse effects of redundant features. Finally, the multi-scale features are used to predict the future carbon emissions of the target city through a predictor. The experimental results show that our prediction model is superior to the existing methods in six datasets. © 2024 IEEE.

关键词： Low emission

来源：评论

学校读者我要写书评

暂无评论

Spatiality-Aligned Graph Convolution Hashing for Unsupervised Cross-Modal Retrieval 2

Spatiality-Aligned Graph Convolution Hashing for Unsupervise...

引用

2nd IEEE International Conference on signal, Information and Data processing, ICSIDP 2024

作者： Li, Fang Dai, Liulian Lan, Rushi Yang, Rui Luo, Xiaonan Guangxi Key Laboratory of Image and Graphic Intelligent Processing School of Computer Science and Information Security Guilin541004 China

ISBN: (纸本)9798331515669

Unsupervised cross-modal retrieval leveraging hash learning has attracted a lot of interest from academics, primarily due to its minimal storage requirements, swift retrieval speeds, and label-free nature, making it a prime candidate for real-world applications. Nevertheless, the inherent "heterogeneous differences"and "semantic gap"within cross-modal data features pose significant challenges. Despite various attempts by scholars to address these issues, the existing solutions often fall short in terms of effectiveness. This can be attributed to several reasons: Firstly, image features, in contrast to text features, possess a plethora of attributes, making it challenging to extract features that possess strong expressive capabilities. Secondly, each modality of data exhibits not only inter-node correlations but also intricate relationships between individual nodes and their neighboring nodes. Consequently, a straightforward semantic correlation matrix is inadequate in capturing the intricate structural relationships within the data. Thirdly, the neglect of internal structural associations results in insufficient representation capabilities of binary hash codes in the Hamming space, hampering the learning of effective hash functions. To address these limitations, we introduce a Spatiality-Aligned Graph Convolution Hashing (SAGCH) algorithm. Extensive experiments demonstrate that our SAGCH method excels in both "image-to-text retrieval"and "text-to-image retrieval"tasks. © 2024 IEEE.

关键词： Semantics

来源：评论

学校读者我要写书评

暂无评论

A Lightweight Music Source Separation Model with Graph Convolution Network 1

引用

18th National Conference on Man-Machine Speech Communication, NCMMSC 2023

作者： Zhu, Mengying Wang, Liusong Hu, Ying School of Computer Science and Technology Xinjiang University Urumqi China Key Laboratory of Signal Detection and Processing in Xinjiang Urumqi China

ISBN: (数字)9789819706013

ISBN: (纸本)9789819706006

With the rapid advancement of deep neural networks, there has been a significant improvement in the performance of music source separation methods. However, most of them primarily focus on improving their separation performance, while ignoring the issue of model size in the real-world environments. For the application in the real-world environments, in this paper, we propose a lightweight network combined with the Graph convolutional network Attention (GCN_A) module for Music Source Separation (G-MSS), which includes an Encoder and four Decoders, each of them outputs a target music source. The G-MSS network adopts both time-domain and frequency-domain L1 losses. The ablation study verifies the effectiveness of our designed GCN Attention (GCN_A) module and multiple Decoders, and also make a visualization analysis of the main components in the G-MSS network. Comparing with the other 13 methods on the MUSDB18 dataset, our proposed G-MSS achieves comparable separation performance while maintaining the lower amount of parameters. © The Author(s), under exclusive license to Springer Nature Singapore Pte Ltd. 2024.

关键词： Convolution

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：