检索结果-内蒙古大学图书馆

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

作者： Dong, Jinwei Wang, Xinsheng Mao, Qirong School of Computer Science and Communication Engineering Jiangsu University China Jiangsu Engineering Research Center of Big Data Ubiquitous Perception and Intelligent Agriculture Applications China Provincial Key Laboratory of Computational Intelligence and New Technologies in Low-Altitude Digital Agriculture Zhenjiang China Audio Speech and Language Processing Group School of Computer Science Northwestern Polytechnical University Xi'an China

ISBN: (纸本)9798350368741

Generative models have attracted considerable attention for speech separation tasks, and among these, diffusion-based methods are being explored. Despite the notable success of diffusion techniques in generation tasks, their adaptation to speech separation has encountered challenges, notably slow convergence and suboptimal separation outcomes. To address these issues and enhance the efficacy of diffusion-based speech separation, we introduce EDSep, a novel single-channel method grounded in score matching via stochastic differential equation (SDE). This method enhances generative modeling for speech source separation by optimizing training and sampling efficiency. Specifically, a novel denoiser function is proposed to approximate data distributions, which obtains ideal denoiser outputs. Additionally, a stochastic sampler is carefully designed to resolve the reverse SDE during the sampling process, gradually separating speech from mixtures. Extensive experiments on databases such as WSJ0-2mix, LRS2-2mix, and VoxCeleb2-2mix demonstrate our proposed method's superior performance over existing diffusion and discriminative models, validating its efficacy. © 2025 IEEE.

关键词： diffusion score matching speech separation stochastic differential equation

来源：评论

学校读者我要写书评

暂无评论

EDSep: An Effective Diffusion-Based Method for Speech Source Separation

EDSep: An Effective Diffusion-Based Method for Speech Source...

引用

International Conference on Acoustics, Speech, and Signal processing (ICASSP)

作者： Jinwei Dong Xinsheng Wang Qirong Mao School of Computer Science and Communication Engineering Jiangsu University Audio Speech and Language Processing Group School of Computer Science Northwestern Polytechnical University Xi’an China Provincial Key Laboratory of Computational Intelligence and New Technologies in Low-Altitude Digital Agriculture Jiangsu Engineering Research Center of Big Data Ubiquitous Perception and Intelligent Agriculture Applications Zhenjiang China

ISBN: (数字)9798350368741

ISBN: (纸本)9798350368758

Generative models have attracted considerable attention for speech separation tasks, and among these, diffusion-based methods are being explored. Despite the notable success of diffusion techniques in generation tasks, their adaptation to speech separation has encountered challenges, notably slow convergence and suboptimal separation outcomes. To address these issues and enhance the efficacy of diffusion-based speech separation, we introduce EDSep, a novel single-channel method grounded in score matching via stochastic differential equation (SDE). This method enhances generative modeling for speech source separation by optimizing training and sampling efficiency. Specifically, a novel denoiser function is proposed to approximate data distributions, which obtains ideal denoiser outputs. Additionally, a stochastic sampler is carefully designed to resolve the reverse SDE during the sampling process, gradually separating speech from mixtures. Extensive experiments on databases such as WSJ0-2mix, LRS2-2mix, and VoxCeleb2-2mix demonstrate our proposed method’s superior performance over existing diffusion and discriminative models, validating its efficacy.

关键词： Training Adaptation models Technological innovation Source separation databases Stochastic processes Differential equations Speech enhancement Diffusion models Mathematical models

来源：评论

学校读者我要写书评

暂无评论

EDSep: An Effective Diffusion-Based Method for Speech Source Separation

arXiv

引用

arXiv 2025年

Generative models have attracted considerable attention for speech separation tasks, and among these, diffusion-based methods are being explored. Despite the notable success of diffusion techniques in generation tasks, their adaptation to speech separation has encountered challenges, notably slow convergence and suboptimal separation outcomes. To address these issues and enhance the efficacy of diffusion-based speech separation, we introduce EDSep, a novel single-channel method grounded in score matching via stochastic differential equation (SDE). This method enhances generative modeling for speech source separation by optimizing training and sampling efficiency. Specifically, a novel denoiser function is proposed to approximate data distributions, which obtains ideal denoiser outputs. Additionally, a stochastic sampler is carefully designed to resolve the reverse SDE during the sampling process, gradually separating speech from mixtures. Extensive experiments on databases such as WSJ0-2mix, LRS2-2mix, and VoxCeleb2-2mix demonstrate our proposed method’s superior performance over existing diffusion and discriminative models, validating its efficacy. Copyright © 2025, The Authors. All rights reserved.

关键词： Stochastic systems

来源：评论

学校读者我要写书评

暂无评论

An accurate cell segmentation method for high-resolution complex brain image 9

An accurate cell segmentation method for high-resolution com...

引用

9th International Conference on image, Vision and Computing, ICIVC 2024

作者： Xu, Xiaofeng Li, Shuai Zheng, Wei Henan Engineering Research Center of Intelligent Processing for Big Data of Digital Image School of Computer Science and Technology Nanyang Normal University Nanyang China

ISBN: (纸本)9798350385991

Cell segmentation is a key step in medical image analysis and cell biology research, and is important in areas such as understanding cell behaviour, cancer detection and cell cycle analysis. This task is challenging due to the presence of overlap between cells and image noise. With the development of imaging techniques, the amount and quality of acquired cellular image data have increased significantly and even contain diverse information such as neuronal structures and blood vessels, which puts higher demands on segmentation techniques. Real rat brain cell images obtained using micro-optical slice tomography (MOST) contain complex and diverse information such as neuronal structure, cell morphology, and the distribution of blood vessels, demonstrating the characteristics of real biological samples, for which segmentation of the target region is more difficult. In order to solve these challenges, this paper proposes a new cell image segmentation method that combines watershed order differential feature extraction and fuzzy entropy spatial theory. Due to the complexity of image information, we first use traditional image enhancement techniques. Then feature extraction is performed and fuzzy entropy processing is added to achieve accurate identification of cell boundaries. Then distance transform and curvature analysis are added to further segment the cell clusters into individual cells while avoiding over-segmentation and missed detection. Experiments demonstrate that the method not only shows high performance on our dataset, but also validates our method on several publicly available cellular datasets respectively, proving that the segmentation algorithm shows great potential for segmentation on different types of cells. Future research will further optimise the model to improve segmentation performance on complex and irregularly shaped cells and explore the application of these methods to other types of medical image segmentation tasks. © 2024 IEEE.

关键词： image segmentation

来源：评论

学校读者我要写书评

暂无评论

An Automatic Method for Fast Synaptic Signal Detection and Count 22

An Automatic Method for Fast Synaptic Signal Detection and C...

引用

5th International Conference on big data Technologies, ICBDT 2022

作者： Xu, Xiaofeng Lan, Yihua Li, He Qi, Qinglei Liu, Quan Henan Engineering Research Center of Intelligent Processing for Big Data of Digital Image School of Computer Science and Technology Nanyang Normal University Nanyang473061 China

ISBN: (纸本)9781450396875

Synapses are the special connection structure between brain neurons, and they are the core element of brain plasticity and information transmission. In order to better understand the process of brain information transmission, it is necessary to quantitatively calculate the synapse, but the automatic identification of a large range of synaptic images is still a difficult problem. According to the characteristics of synaptic signal in microscopy optical mouse brain images, this paper proposes the automatic fast synapse recognition method based on density peak clustering, which solves the identification problem of dense small synaptic signal, and realizes the rapid synapse positioning and counting in a large brain area. The results show that the proposed methods in accuracy and efficiency bring great convenience to the researchers. CCS CONCEPTSĝ€¢Applied computing © 2022 ACM.

关键词： Signal detection

来源：评论

学校读者我要写书评

暂无评论

Automatic Segmentation Method for the Nuclei Contour of High Resolution Rat Brain 23

Automatic Segmentation Method for the Nuclei Contour of High...

引用

6th International Conference on big data Technologies, ICBDT 2023

作者： Xu, Xiaofeng Zheng, Wei Henan Engineering Research Center of Intelligent Processing for Big Data of Digital Image School of Computer Science and Technology Nanyang Normal University Nanyang473061 China

ISBN: (纸本)9798400707667

The brain consists of massive nuclei with different functions. In neuroscience research, the precise recognition and delineation of nucleus boundaries is the crux of brain atlas illustration. Here, we propose a method based on Markov Random Field. This method introduces fractional differentiation into texture feature extraction and a new potential energy function is defined based on kernel region information. To handle the problem of large data volume, we propose a strategy of dual MRF processing with down sampling for pre classification. Finally, the fuzzy entropy criterion is used to optimize the segmentation results. Experiments on model data and real tissue slice data show that this method can not only segment the regions with obvious differences in the density of the nuclei with insignificant differences in the density of nuclei, but also divide the nuclei without insignificant density differences by the cell shapes and textures. In addition, this method can not only segment a specific target kernel in one calculation, but also simultaneously partition multiple target kernels. The segmentation algorithm shows great potential for illustrating high-resolution 3D brain atlases. © 2023 ACM.

关键词： Markov processes

来源：评论

学校读者我要写书评

暂无评论

The NP-Completeness of (1,0)-Regular (k,s)-SAT 14

The NP-Completeness of (1,0)-Regular (k,s)-SAT

引用

14th International Conference on intelligent Human-Machine Systems and Cybernetics, IHMSC 2022

作者： Fu, Zufeng Wang, Haiying Li, He Liu, Jinjiang Nanyang Normal University School of Computer Science and Technology Nanyang China Nanyang Normal University Henan Engineering Research Center of Intelligent Processing for Big Data of Digital Image Nanyang China

ISBN: (数字)9781665461696

ISBN: (纸本)9781665461696

Dealing with uncertainty in optimization is increasingly recognized as a key necessity for tackling real-world problems. A (1,0)-super solution has better robust, and can protect against default risk of variable assignment. When flipped a value of any one of variables, the (1,0)-super solution still is a solution. In this work, we study the completeness of (1,0)-regular (k,s)-SAT. (1,0)-regular (k,s)-SAT is a problem to decide whether a regular (k,s)-CNF formula has a (1,0)-super solution. We give a reduction method that transform from (k,s)-SAT to (1,0)-regular (k+1,s)-SAT if an unsatisfiable (k,s)-CNF formula exists, and a structure method of forced-regular (k,s)-CNF formula. This implies that (1,0)-regular (k +1,s)-SAT problem has a sudden jump from triviality to NP-completeness for every k > 3. © 2022 IEEE.

关键词： Uncertainty Transforms Proposals Optimization Man-machine systems Cybernetics

来源：评论

学校读者我要写书评

暂无评论

The Jessen inequality of semi-E-preinvex functions and its applications 14

The Jessen inequality of semi-E-preinvex functions and its a...

引用

14th International Conference on intelligent Human-Machine Systems and Cybernetics, IHMSC 2022

作者： Fu, Zufeng Wang, Haiying Jia, Xiao School of Computer Science and Technology Nanyang Normal University Nanyang China2. Henan Engineering Research Center of Intelligent Processing for Big Data of Digital Image Nanyang Normal University Nanyang China

The Jensen inequality for the semi-E-preinvex functions is obtained by introducing the η-E-convex linear combination suitable for E-invex sets and semi-E-preinvex functions, and an upper bound of the error of the sem... 详细信息

ISBN: (数字)9781665461696

ISBN: (纸本)9781665461696

关键词： Upper bound Man-machine systems Cybernetics

来源：评论

学校读者我要写书评

暂无评论

Lightweight Transformer Network and Self-supervised Task for Kinship Verification 8

Lightweight Transformer Network and Self-supervised Task for...

引用

8th IEEE International Conference on Computer and Communications, ICCC 2022

作者： Zhu, Xiaoke Li, Yunwei Li, Danyang Dong, Lingyun Chen, Xiaopan School of Computer and Information Engineering Henan University Kaifeng China Henan Key Laboratory of Big Data Analysis and Processing Henan University Kaifeng China Henan Engineering Research Center of Intelligent Technology and Application Henan University China

ISBN: (纸本)9781665450515

Kinship verification is one of the interesting and critical problems in computer vision research, with significant progress in the past decades. Meanwhile, Vision Transformer (VIT) has recently achieved impressive success in many domains, including object detection, image recognition, and semantic segmentation, among others. Most of the previous work on kinship verification are based on convolutional or recurrent neural networks. Compared with the local processing of images like convolutions, transformers can effectively understand and process images globally. However, due to overuse, there are many Transformer layers of fully connected layers, and VIT speed is still an issue. Therefore, in this paper, inspired by the recent success of Transformer models in vision tasks, we propose a Transformer-based kinship verification for training and optimizing kinship verification models. We first train the basic vision transformer (VIT-B) with 12 transformer layers, then we reduce the transformer layers to 6 layers, namely VIT-S (Small Vit) and 4 layers, namely VIT-T (Tiny Vit), to make a tradeoff between optimization accuracy and efficiency. As the first attempt to apply Transformer to the kinship verification task, it provides a feasible strategy for kinship research topics and verifies the effectiveness of the method in terms of the accuracy of the experimental results. © 2022 IEEE.

关键词： image recognition

来源：评论

学校读者我要写书评

暂无评论

Point-Level and Set-Level Deep Representation Learning for Cross-Modal Person Re-identification 8

Point-Level and Set-Level Deep Representation Learning for C...

引用

8th IEEE International Conference on Computer and Communications, ICCC 2022

作者： Hu, Jihui Ye, Pengfei Li, Danyang Dong, Lingyun Chen, Xiaopan Zhu, Xiaoke School of Computer and Information Engineering Henan University Kaifeng China Henan University Henan Key Laboratory of Big Data Analysis and Processing Kaifeng China Henan University Henan Engineering Research Center of Intelligent Technology and Application China

ISBN: (纸本)9781665450515

In practice, significant modality differences usually exist between visible and infrared images, which makes visible-infrared Person Re-Identification (VI-ReID) a challenging research task. Due to the existing influence of pose variation, background changes, and occlusion, there are often outlier samples within the set of images from the same person. These outlier samples can adversely affect the process of learning the cross-modal matching model. The existing VI-ReID methods mainly focus on learning cross-modal feature representation by using image-level discriminant constraints, i.e., the distance between the truly-matching cross-modal images should be smaller than that between wrong-matching cross-modal images. However, most of these methods ignore the adverse influence caused by outliers. To solve the above problems, we proposed a Point-level and Set-level Deep Representation Learning (PSDRL) approach for VI-ReID in this paper. By using the set-level constraint in the process of deep representation learning, the discrepancy between visible and infrared modalities can be decreased, and the adverse effect of outliers can be weakened. By employing the image-level constraint, the discriminability of the obtained deep feature representations can be improved. Extensive experiments are conducted on the publicly available cross-modal Person Re-Identification datasets, including SYSU-MM01 and RegDB. Experimental results demonstrate the effectiveness of the proposed approach. © 2022 IEEE.

关键词： Infrared imaging

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：