检索结果-内蒙古大学图书馆

Self-Supervised Speech Denoising Using Only Noisy Audio Signals

学校读者我要写书评

暂无评论

arXiv 2021年

作者： Wu, Jiasong Li, Qingchun Yang, Guanyu Li, Lei Senhadji, Lotfi Shu, Huazhong LIST Key Laboratory of Computer Network and Information Integration Southeast University Ministry of Education Nanjing210096 China Jiangsu Provincial Joint International Research Laboratory of Medical Information Processing Southeast University Nanjing210096 China RennesF-35042 France The 28th Research Institute of China Electronics Technology Group Corporation Nanjing210007 China Univ Rennes INSERM LTSI UMR 1099 RennesF-35042 France

In traditional speech denoising tasks, clean audio signals are often used as the training target, but absolutely clean signals are collected from expensive recording equipment or in studios with the strict environments. To overcome this drawback, we propose an end-to-end self-supervised speech denoising training scheme using only noisy audio signals, named Only-Noisy Training (ONT), without extra training conditions. The proposed ONT strategy constructs training pairs only from each single noisy audio, and it contains two modules: training audio pairs generated module and speech denoising module. The first module adopts a random audio sub-sampler on each noisy audio to generate training pairs. The sub-sampled pairs are then fed into a novel complex-valued speech denoising module. Experimental results show that the proposed method not only eliminates the high dependence on clean targets of traditional audio denoising tasks, but also achieves on-par or better performance than other training strategies. Availability-ONT is available at https://***/liqingchunnnn/Only-Noisy-Training. Copyright © 2021, The Authors. All rights reserved.

关键词： Machine learning

Blockchain Assisted Federated Learning over Wireless Channels: Dynamic Resource Allocation and Client Scheduling

学校读者我要写书评

暂无评论

arXiv 2021年

作者： Deng, Xiumei Li, Jun Ma, Chuan Wei, Kang Shi, Long Ding, Ming Chen, Wen Poor, H. Vincent The School of Electronic and Optical Engineering Nanjing University of Science and Technology Nanjing210094 China Zhejiang Lab Hangzhou China Nanjing University of Science and Technology Key Laboratory of Computer Network and Information Integration Southeast University Ministry of Education China Data61 CSIRO KensingtonWA6152 Australia The Department of Electronics Engineering Shanghai Jiao Tong University Shanghai200240 China Department of Electrical and Computer Engineering Princeton University NJ08544 United States

The blockchain technology has been extensively studied to enable distributed and tamper-proof data processing in federated learning (FL). Most existing blockchain assisted FL (BFL) frameworks have employed a third-party blockchain network to decentralize the model aggregation process. However, decentralized model aggregation is vulnerable to pooling and collusion attacks from the third-party blockchain network. Driven by this issue, we propose a novel BFL framework that features the integration of training and mining at the client side. To optimize the learning performance of FL, we propose to maximize the long-term time average (LTA) training data size under a constraint of LTA energy consumption. To this end, we formulate a joint optimization problem of training client selection and resource allocation (i.e., the transmit power and computation frequency at the client side), and solve the long-term mixed integer non-linear programming based on a Lyapunov technique. In particular, the proposed dynamic resource allocation and client scheduling (DRACS) algorithm can achieve a trade-off of [O(1/V), O(√V)] to balance the maximization of the LTA training data size and the minimization of the LTA energy consumption with a control parameter V. Our experimental results show that the proposed DRACS algorithm achieves better learning accuracy than benchmark client scheduling strategies with limited time or energy consumption. Copyright © 2021, The Authors. All rights reserved.

关键词： Blockchain

Binary relevance for multi-label learning： an overview

学校读者我要写书评

暂无评论

Frontiers of computer Science 2018年第2期12卷 191-202页

作者： Min-Ling ZHANG Yu-Kun LI Xu-Ying LIU Xin GENG School of Computer Science and Engineering Southeast University Nanjing 210096 China Key Laboratory of Computer Network and Information Integration (Southeast University) Ministry of Education China Collaborative Innovation Center for Wireless Communications Technology Nanjing 211100 China

Multi-label learning deals with problems where each example is represented by a single instance while being associated with multiple class labels simultaneously. Binary relevance is arguably the most intuitive solution for learning from multi-label examples. It works by decomposing the multi-label learning task into a number of independent binary learning tasks （one per class label）. In view of its potential weakness in ignoring correlations between labels, many correlation-enabling extensions to binary relevance have been proposed in the past decade. In this paper, we aim to review the state of the art of binary relevance from three perspectives. First, basic settings for multi-label learning and binary relevance solutions are briefly summarized. Second, representative strategies to provide binary relevance with label correlation exploitation abilities are discussed. Third, some of our recent studies on binary relevance aimed at issues other than label correlation exploitation are introduced. As a conclusion, we provide suggestions on future research directions.

关键词： machine learning multi-label learning binary relevance label correlation class-imbalance relative labeling-importance

CPNet: cycle prototype network for weakly-supervised 3D renal compartments segmentation on CT images

学校读者我要写书评

暂无评论

arXiv 2021年

作者： Wang, Song He, Yuting Kong, Youyong Zhu, Xiaomei Zhang, Shaobo Shao, Pengfei Dillenseger, Jean-Louis Coatrieux, Jean-Louis Li, Shuo Yang, Guanyu LIST Key Laboratory of Computer Network and Information Integration Southeast University Ministry of Education Nanjing China Univ Rennes Inserm LTSI UMR1099 RennesF-35000 France Dept. of Radiology First Affliated Hospital Nanjing Medical University Dept. of Urology First Affliated Hospital of Nanjing Medical University Dept. of Medical Biophysics University of Western Ontario LondonON Canada

Renal compartment segmentation on CT images targets on extracting the 3D structure of renal compartments from abdominal CTA images and is of great significance to the diagnosis and treatment for kidney diseases. However, due to the unclear compartment boundary, thin compartment structure and large anatomy variation of 3D kidney CT images, deep-learning based renal compartment segmentation is a challenging task. We propose a novel weakly supervised learning frame- work, Cycle Prototype network, for 3D renal compartment segmenta- tion. It has three innovations: 1) A Cycle Prototype Learning (CPL) is proposed to learn consistency for generalization. It learns from pseudo labels through the forward process and learns consistency regularization through the reverse process. The two processes make the model robust to noise and label-efficient. 2) We propose a Bayes Weakly Supervised Module (BWSM) based on cross-period prior knowledge. It learns prior knowledge from cross-period unlabeled data and perform error correc- tion automatically, thus generates accurate pseudo labels. 3) We present a Fine Decoding Feature Extractor (FDFE) for fine-grained feature ex-traction. It combines global morphology information and local detail in- formation to obtain feature maps with sharp detail, so the model will achieve fine segmentation on thin structures. Our model achieves Dice of 79:1% and 78:7% with only four labeled images, achieving a significant improvement by about 20% than typical prototype model PANet[16]. © 2021, CC BY.

关键词： computerized tomography

Homotopic gradients of generative density priors for MR image reconstruction

学校读者我要写书评

暂无评论

arXiv 2020年

作者： Quan, Cong Zhou, Jinjie Zhu, Yuanzheng Chen, Yang Wang, Shanshan Liang, Dong Liu, Qiegen The Department of Electronic Information Engineering Nanchang University Nanchang330031 China The Key Laboratory of Computer Network and Information Integration Southeast University Ministry of Education Nanjing China Paul C. Lauterbur Research Center for Biomedical Imaging SIAT Chinese Academy of Sciences Shenzhen518055 China

Dee learnin articularl the enerative model, has demonstrated tremendous potential to significantly speed up image reconstruction with reduced measurements recently. Rather than the existing generative models that often optimize the density priors, in this work, by taking advantage of the denoising score matching, homotopic gradients of generative density priors (HGGDP) are exploited for magnetic resonance imaging (MRI) reconstruction. More precisely, to tackle the low-dimensional manifold and low data density region issues in generative density prior, we estimate the target gradients in higher-dimensional space. We train a more powerful noise conditional score network by forming high-dimensional tensor as the network input at the training phase. More artificial noise is also injected in the embedding space. At the reconstruction stage, a homotopy method is employed to pursue the density prior, such as to boost the reconstruction performance. Experiment results implied the remarkable performance of HGGDP in terms of high reconstruction accuracy;only 10% of the k-space data can still generate image of high quality as effectively as standard MRI reconstructions with the fully sampled data. Copyright © 2020, The Authors. All rights reserved.

关键词： Magnetic resonance imaging

A Memory-efficient Simulation Method of Grover’s Search Algorithm

学校读者我要写书评

暂无评论

computers, Materials & Continua 2018年第11期57卷 307-319页

作者： Xuwei Tang Juan Xu Bojia Duan College of Computer Science and Technology Nanjing University of Aeronautics and AstronauticsNanjingChina Institute for Quantum Computing University of WaterlooCanada Key Laboratory of Computer Network and Information Integration(Southeast University) Ministry of EducationChina

Grover’s search algorithm is one of the most significant quantum algorithms,which can obtain quadratic speedup of the extensive search *** Grover's search algorithm cannot be implemented on a real quantum computer at present,its quantum simulation is regarded as an effective method to study the search *** simulating the Grover's algorithm,the storage space required is exponential,which makes it difficult to simulate the high-qubit Grover’s *** this end,we deeply study the storage problem of probability amplitude,which is the core of the Grover simulation *** propose a novel memory-efficient method via amplitudes compression,and validate the effectiveness of the method by theoretical analysis and simulation *** results demonstrate that our compressed simulation search algorithm can help to save nearly 87.5%of the storage space than the uncompressed *** under the same hardware conditions,our method can dramatically reduce the required computing nodes,and at the same time,it can simulate at least 3 qubits more than the uncompressed ***,our memory-efficient simulation method can also be used to simulate other quantum algorithms to effectively reduce the storage costs required in simulation.

关键词： Grover’s search algorithm probability amplitude quantum simulation memory compression

DRTNet: Dual-route transformer network for thyroid ultrasound segmentation based on Bbox-supervised learning

学校读者我要写书评

暂无评论

Knowledge-Based Systems 2025年 324卷

作者： Bi, Hui Cai, Chengjie Sun, Jiawei Ge, Shihao Shu, Huazhong Ni, Xinye School of Computer Science and Artificial Intelligence Changzhou University Changzhou213164 China The Affiliated Changzhou NO.2 People's Hospital of Nanjing Medical University Changzhou213003 China Key Laboratory of Computer Network and Information Integration Southeast University Nanjing211096 China Jiangsu Province Engineering Research Center of Medical Physics Changzhou213003 China Center of Medical Physics Nanjing Medical University Changzhou213003 China Laboratory of Image Science and Technology Southeast University Nanjing210096 China Centre de Recherche en Information Biomédicale Sino-Français RennesF-35000 France

Background and Objective: Accurate nodule delineation plays a significant role in the intelligent diagnosis of thyroid disease. However, the labels accessing is difficult since it is time-consuming and laborious. To mitigate the over-dependence of the segmentation accuracy on the labels, we proposed a dual-route transformer network (DRTNet) based on Bbox-supervised learning that only requires the rough bounding rectangle as labels instead of a precise boundary for thyroid ultrasound segmentation. Methods: DRTNet incorporates double-branch foreground class activation mappings (CAMs) into Transformers to combine key areas. Meanwhile, double-branch architecture dynamically adjusts the feature distribution of nodules of different sizes in frequency channels and spatial dimensions, effectively addressing the localization of nodules of different sizes. Moreover, ultrasound prior background-aware pooling (UPBAP) is proposed in both branches to deal with the ambiguous boundary of thyroid nodules. Finally, adaptive uncertainty estimate multi-scale consistency (AUEMC) is proposed to help mitigate the risk of excessive over-fitting because of pseudo annotations, which further guarantees consistency among nodules with diverse resolutions. Results: Substantial improvement of segmentation accuracy is shown on the public thyroid dataset of TN3k and DDTI dataset with Dice similarity coefficient (DSC) of 84.94% and 83.98%, with 95% of the asymmetric Hausdorff distance (HD95) of 27.69 and 29.18, respectively. And our private dataset has a DSC of 84.39% and HD95 of 14.53. Conclusions: The proposed DRTNet used rectangular box labeled for thyroid ultrasound images based on Bbox-supervised learning. The experimental results show that the DRTNet is comparable to these fully supervised methods. Code is available at https://***/ccjcv/DRTNet. © 2025 Elsevier B.V.

关键词： Supervised learning

4D-CBCT Reconstruction via Motion Compensataion Learning Induced Sparse Tensor Constraint

学校读者我要写书评

暂无评论

4D-CBCT Reconstruction via Motion Compensataion Learning Ind...

International Congress on Image and Signal Processing, BioMedical Engineering and Informatics (CISP-BMEI)

作者： Jin Liu Yanqin Kang Dianlin Hu Yang Chen Key Laboratory of Computer Network and Information Integration (Southeast University) Ministry of Education Nanjing China School of Computer Science and Engineering Southeast University Nanjing China

The challenge in 4D-CBCT reconstruction is the lacked of projections at each phase, which result in a reconstruction image full of streak artifacts and motion blurring with the conventional analytical algorithms. Aiming at improving the overall quality of 4D-CBCT images, we propose a motion compensation learning induced sparse tensor constraint reconstruction (MCL-STCR) method which tries to explore the high correlation in 4D-CBCT images for different phases. In this method, we additionally conduct motion compensation on the 4D-CBCT volume by using trained image motion compensation convolutional neural network (CNN). Then the compensated 4D-CBCT volume is viewed as a pseudostatic sequence, of which the sparse tensor constraint was imposed on. The cost function of proposed MCL-STCR is optimized by a variable splitting algorithm. Its comparison to other methods through simulated dynamic phantom and lung data experiments demonstrated that the proposed method can lead to a promising improvement of 4D-CBCT reconstruction.

关键词：

Open Event Extraction from Online Text using a Generative Adversarial network

学校读者我要写书评

暂无评论

arXiv 2019年

作者： Wang, Rui Zhou, Deyu He, Yulan School of Computer Science and Engineering Key Laboratory of Computer Network and Information Integration Ministry of Education Southeast University China Department of Computer Science University of Warwick United Kingdom

To extract the structured representations of open-domain events, Bayesian graphical models have made some progress. However, these approaches typically assume that all words in a document are generated from a single event. While this may be true for short text such as tweets, such an assumption does not generally hold for long text such as news articles. Moreover, Bayesian graphical models often rely on Gibbs sampling for parameter inference which may take long time to converge. To address these limitations, we propose an event extraction model based on Generative Adversarial Nets, called Adversarial-neural Event Model (AEM). AEM models an event with a Dirichlet prior and uses a generator network to capture the patterns underlying latent events. A discriminator is used to distinguish documents reconstructed from the latent events and the original documents. A byproduct of the discriminator is that the features generated by the learned discriminator network allow the visualization of the extracted events. Our model has been evaluated on two Twitter datasets and a news article dataset. Experimental results show that our model outperforms the baseline approaches on all the datasets, with more significant improvements observed on the news article dataset where an increase of 15% is observed in F-measure. Copyright © 2019, The Authors. All rights reserved.

关键词： Generative adversarial networks