检索结果-内蒙古大学图书馆

arXiv 2025年

作者： Huang, Yanglin Hu, Kai Zhang, Yuan Chen, Zhineng Gao, Xieping Key Laboratory of Intelligent Computing and Information Processing Ministry of Education Xiangtan University China School of Computer Science Fudan University China Key Laboratory for Artificial Intelligence and International Communication Hunan Normal University China

Current knowledge distillation (KD) methods for semantic segmentation focus on guiding the student to imitate the teacher’s knowledge within homogeneous architectures. However, these methods overlook the diverse knowledge contained in architectures with different inductive biases, which is crucial for enabling the student to acquire a more precise and comprehensive understanding of the data during distillation. To this end, we propose for the first time a generic knowledge distillation method for semantic segmentation from a heterogeneous perspective, named HeteroAKD. Due to the substantial disparities between heterogeneous architectures, such as CNN and Transformer, directly transferring cross-architecture knowledge presents significant challenges. To eliminate the influence of architecture-specific information, the intermediate features of both the teacher and student are skillfully projected into an aligned logits space. Furthermore, to utilize diverse knowledge from heterogeneous architectures and deliver customized knowledge required by the student, a teacher-student knowledge mixing mechanism (KMM) and a teacher-student knowledge evaluation mechanism (KEM) are introduced. These mechanisms are performed by assessing the reliability and its discrepancy between heterogeneous teacher-student knowledge. Extensive experiments conducted on three main-stream benchmarks using various teacher-student pairs demonstrate that our HeteroAKD outperforms state-of-the-art KD methods in facilitating distillation between heterogeneous architectures. Copyright © 2025, The Authors. All rights reserved.

关键词： Students

来源：评论

学校读者我要写书评

暂无评论

A uniform solution to the independent set problem through tissue P systems with cell separation

引用

Frontiers of Computer Science 2012年第4期6卷 477-488页

作者： Xingyi ZHANG Xiangxiang ZENG Bin LUO Zheng ZHANG Key Laboratory of Intelligent Computing and Signal Processing of Ministry of Education School of Computer Science and TechnologyAnhui UniversityHefei 230039China Department of Computer Science Xiamen UniversityXiamen 361005China Key Laboratory of Image Processing and Intelligent Control Department of Control Science and EngineeringHuazhong University of Science and TechnologyWuhan 430074China

Membrane computing is an emergent branch of natural computing, which is inspired by the structure and the functioning of living cells, as well as the organization of cells in tissues, organs, and other higher order structures. Tissue P systems are a class of the most investigated computing mod- els in the framework of membrane computing, especially in the aspect of efficiency. To generate an exponential resource in a polynomial time, cell separation is incorporated into such systems, thus obtaining so called tissue P systems with cell separation. In this work, we exploit the computational efficiency of this model and construct a uniform family of such tissue P systems for solving the independent set problem, a well-known NP-complete problem, by which an efficient so- lution can be obtained in polynomial time.

关键词： membrane computing tissue P system cell sep- aration independent set problem

来源：评论

学校读者我要写书评

暂无评论

The classification of gene expression profile based on the adjacency matrix spectral decomposition

The classification of gene expression profile based on the a...

引用

International Conference on Computer Application and System Modeling

作者： Su, Liangliang Wang, Nian Tang, Jun Chen, Le Wang, Ruiping Education Ministry Key Laboratory of Intelligent Computing and Signal Processing Anhui University Hefei Anhui 230039 China

ISBN: (纸本)9781424472369

In the process of disease diagnosis, determining the types of disease is very important. With the development of DNA microarray technology, the research on huge gene expression profile has become the focus of disease classification. This paper presents a method of classification of gene expression profile based on the adjacency matrix spectral decomposition. First, samples are mapped to a high-dimensional space of points to construct an adjacency matrix, and we can obtain eigenvectors describing the feature of the samples by decomposing the matrix. Finally, use eigenvectors as inputs of the SVM (Support Vector Machine) and KNN (K nearest neighbor) classifiers to classify gene expression profile. In this way sample information can be completely preserved, which enables an approach to making gene expression profile from one without structural information to one with structural information. The validity of this method is verified by comparative experiments. © 2010 IEEE.

关键词： Eigenvalues and eigenfunctions

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

Novel algorithm for hand vein recognition based on Retinex method and SIFT feature analysis

Novel algorithm for hand vein recognition based on Retinex m...

引用

2011 International Conference on Electric and Electronics, EEIC 2011

作者： Wang, Hua-Bin Tao, Liang Hu, Xue-You Key Laboratory of Intelligent Computing and Signal Processing Ministry of Education Anhui University Hefei Anhui 230039 China

ISBN: (纸本)9783642217463

Based on the Retinex method and the SIFT feature analysis, this paper presents a novel algorithm for hand vein recognition. First of all, the principle of the near-infrared hand vein image acquisition is introduced. Secondly, the Retinex method is used to normalize hand vein images, and the adaptive smoothing method is selected to estimate the illumination. Then gray cosine transform is used to enhance the discrimination of the skin and the vein in hand vein images. Thirdly, the SIFT feature analysis algorithm is used to extract the feature of hand vein. Finally, the match method of two hand vein images based on SIFT is given. A hand vein recognition system in Microsoft VC6.0 is also developed and the experimental results demonstrate the high efficiency of the proposed algorithm in runtime and correct recognition rate. © Springer-Verlag Berlin Heidelberg 2011.

关键词： Cosine transforms

来源：评论

学校读者我要写书评

暂无评论

Multiplicative reconstruction based on the transport of intensity equation 4

Multiplicative reconstruction based on the transport of inte...

引用

4th International Conference on Fuzzy Systems and Data Mining, FSDM 2018

作者： Cheng, Hong Wang, Jincheng Shen, Chuan Zhang, Cheng Zhang, Fen Bao, Wenxia Education Ministry Key Laboratory of Intelligent Computing and Signal Processing Anhui University HefeiAnhui230039 China

ISBN: (纸本)9781614999270

In order to effectively recover the phase information of the object and reconstruct the three-dimensional shape of the object, a multiplicative reconstruction based on the transport of intensity equation (TIE) is proposed. Firstly, combining with the relationship between the light field moment imaging and the TIE, the phase information and the first moments of the light field are solved between the two differently focused intensity images, and the phase perspective can be recovered. In this way, the phase can be observed from different slight angles, and the three-dimensional shape of the phase of the object is reconstructed in combination with the multiplicative technique . At the same time, the object is rotated at different angles for the purpose of reconstructing the three-dimensional shape of the object phase under these angles. Finally, the correctness and effectiveness of the algorithm are verified by experimental results. © 2018 The authors and IOS Press. All rights reserved.

关键词： Fuzzy systems

来源：评论

学校读者我要写书评

暂无评论

3D fresnel field computing method based on its propagation reversibility

3D fresnel field computing method based on its propagation r...

引用

Chinese Society for Optical Engineering Conferences, CSOE 2016

作者： Zhang, Kai Wang, Xiangxiang Tan, Shengxia Shen, Chuan Wei, Sui Key Laboratory of Intelligent Computing and Signal Processing Ministry of Education Anhui University Hefei Anhui230039 China

ISBN: (纸本)9781510610118

There are many 3D data structures for computing diffraction field, the description based on 3D point-cloud is the elemental method of other descriptions, the diffraction field of 3D scene can be thought as the superposition of the diffraction field of each sampled light point and other descriptions can be thought as grouping with point-cloud by different way. The paper focuses on devising a computation method of 3D Fresnel field which is based on Fresnel diffraction field propagation reversibility. The proposed method combines the reversibility of Fresnel propagation and the requirement of sampling spacing by Nyquist criterion, i.e. to find the best space positions to optimize the number of sampling points and the reconstruction for the 3D Fresnel field. The method can be applied to reduce the computational complexity and improve computational efficiency. © 2017 SPIE.

关键词： Computational efficiency

来源：评论

学校读者我要写书评

暂无评论

Novel algorithm for enhancement of hand vein images based on adaptive filtering and retinex method

Novel algorithm for enhancement of hand vein images based on...

引用

2012 IEEE International Conference on Information Science and Technology, ICIST 2012

作者： Wang, Hua-Bin Tao, Liang Key Laboratory of Intelligent Computing and Signal Processing Ministry of Education Anhui University Hefei Anhui 230039 China

ISBN: (纸本)9781457703454

Based on adaptive filtering and retinex method, this paper presents a novel algorithm for Enhancement of hand vein images. First of all, the principle of the near-infrared hand vein image acquisition is introduced. Secondly, the Retinex method is used to normalize hand vein images, and the adaptive smoothing method is selected to estimate the illumination. Thirdly, Gray cosine transform is used to enhance the discrimination of the skin and the vein in hand vein images. Finally, a determination criterion of hand vein is established to remove the false vein blocks from the segmented hand vein images. We developed a hand vein recognition system in Microsoft VC6.0 and the experimental results demonstrate the high efficiency of the proposed algorithm in runtime and correct recognition rate. © 2012 IEEE.

关键词： Adaptive filtering

来源：评论

学校读者我要写书评

暂无评论

Prediction of protein-protein interaction sites using granularity computing of quotient space theory

Prediction of protein-protein interaction sites using granul...

引用

International Conference on Computer Science and Software Engineering, CSSE 2008

作者： Du, Xiuquan Cheng, Jiaxing Key Laboratory of Intelligent Computing and Signal Processing Ministry of Education Anhui University Hefei Anhui province China

ISBN: (纸本)9780769533360

Proteins function through interactions with other proteins, compounds, RNA and DNA. Prediction of protein interface sites is the key process for providing clues to the function of a protein, and is becoming increasing relevant to drug discovery. In this paper, combining the protein features with the theory of granular computing of quotient space based on protein-protein interaction sites classification algorithm is proposed, (i.e. PPI-GS Model). Such algorithm uses the support vector machine (SVM) with better generalization ability to get different classification results, and construct different quotient space by using these results. The final result is got by granularity synthesis method to organize these quotient spaces. The experiment suggests that this new method is in effect. Also, the potency of the granular computing of quotient space in the application of protein-protein interaction sites prediction and processing is demonstrated © 2008 IEEE.

关键词： Support vector machines

来源：评论

学校读者我要写书评

暂无评论

Locating human eyes in face images based on illumination normalization and progressive thresholding

Locating human eyes in face images based on illumination nor...

引用

International Conference on Computer Science and education

作者： Wang, Hua-Bin Tao, Liang Hu, Xue-You Wang, Shi-Yi Key Laboratory of Intelligent Computing and Signal Processing Ministry of Education Anhui University Hefei Anhui 230039 China

ISBN: (纸本)9781424460052

Based on illumination normalization and progressive thresholding, this paper presents a novel algorithm for automatic localization of human eyes in still images with complex background. First of all, Retinex method is used to realize normalizing of the face image. Secondly, a determination criterion of eye location is established by the priori knowledge of geometrical facial features. Thirdly, a range of threshold values that would separate eye blocks from others in a segmented face image is estimated. With the progressive increase of the threshold by an appropriate step in that range, once two eye blocks appear from the segmented image, they will be detected by the determination criterion of eye location. Finally, the 2-D correlation coefficient is used as a symmetry similarity measure to check the factuality of the two detected eyes. The experimental results demonstrate the high efficiency of the algorithm in runtime and correct localization rate. ©2010 IEEE.

关键词： Image segmentation

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：