检索结果-内蒙古大学图书馆

Enhanced Privacy and Communication Efficiency in Non-IID Federated Learning With adaptive quantization and Differential Privacy

引用

IEEE ACCESS 2025年 13卷 54322-54337页

作者： Ardic, Emre Genc, Yakup Gebze Tech Univ Dept Comp Engn TR-41400 Gebze Kocaeli Turkiye

Federated learning (FL) is a distributed machine learning method where multiple devices collaboratively train a model under the management of a central server without sharing underlying data. One of the key challenges of FL is the communication bottleneck caused by variations in connection speed and bandwidth across devices. Therefore, it is essential to reduce the size of transmitted data during training. Additionally, there is a potential risk of exposing sensitive information through the model or gradient analysis during training. To address both privacy and communication efficiency, we combine differential privacy (DP) and adaptive quantization methods. We use Laplacian-based DP to preserve privacy, which is relatively underexplored in FL and offers tighter privacy guarantees than Gaussian-based DP. We propose a simple and efficient global bit-length scheduler using round-based cosine annealing, along with a client-based scheduler that dynamically adapts based on client contribution estimated through dataset entropy analysis. We evaluate our approach through extensive experiments on CIFAR10, MNIST, and medical imaging datasets, using non-IID data distributions across varying client counts, bit-length schedulers, and privacy budgets. The results show that our adaptive quantization methods reduce total communicated data by up to 52.64% for MNIST, 45.06% for CIFAR10, and 31% to 37% for medical imaging datasets compared to 32-bit float training while maintaining competitive model accuracy and ensuring robust privacy through DP.

关键词： quantization (signal) Training Privacy Servers Computational modeling Adaptation models Accuracy Differential privacy Entropy Data models Federated learning adaptive quantization differential privacy non-IID distribution

来源：评论

学校读者我要写书评

暂无评论

adaptive quantization and fast error-resilient entropy coding for image transmission

引用

IEEE TRANSACTIONS ON CIRCUITS AND SYSTEMS FOR VIDEO TECHNOLOGY 1998年第4期8卷 411-421页

作者： Chandramouli, R Ranganathan, N Ramadoss, SJ Univ S Florida Dept Comp Sci & Engn Ctr Microelect Res Tampa FL 33620 USA

Recently, there has been an outburst of research in image and video compression for transmission over noisy channels, Channel matched source quantizer design has gained prominence, Further, the presence of variable-length codes in compression standards like the JPEG and the MPEG has made the problem more interesting. Error resilient entropy coding (EREC) has emerged as a new and effective method to combat catastrophic loss in the received signal due to burst and random errors. In this paper, we propose a new channel-matched adaptive quantizer for JPEG image compression. A slow, frequency-nonselective Rayleigh fading channel model is assumed. The optimal quantizer that matches the human visibility threshold and the channel bit-error rate is derived. Further, a new fast error-resilient entropy code (FEREC) that exploits the statistics of the JPEG compressed data is proposed, The proposed FEREC algorithm is shown to be almost twice as fast as EREC in encoding the data, and hence the error resilience capability is also observed to be significantly better. On an average, a 5% decrease in the number of significantly corrupted received image blocks is observed with FEREC, Upto a 2-dB improvement in the peak Signal-to-noise ratio of the received image is also achieved.

关键词： adaptive quantization error-resilient coding fading image compression JPEG

来源：评论

学校读者我要写书评

暂无评论

adaptive quantization - SOLUTION VIA NONadaptive LINEAR-CONTROL

引用

IEEE TRANSACTIONS ON COMMUNICATIONS 1993年第5期41卷 741-748页

作者： CRISAFULLI, S BITMEAD, RR AUSTRALIAN NATL UNIV RES SCH PHYS SCI & ENGN DEPT SYST ENGN CANBERRA ACT 0200 AUSTRALIA

A common nonlinear adaptive quantization problem is posed as a corresponding linear quadratic control problem. New adaptive quantizer algorithms are generated based on methods of disturbance rejection in linear systems. Simulation with speech data shows that these new algorithms can outperform a previous well-known method. The new approach subsumes and extends previous algorithms in a logical fashion.

关键词： linear quadratic control Disturbance rejection algorithms quantizer adaptive quantization logical Linear system

来源：评论

学校读者我要写书评

暂无评论

adaptive quantization FOR ONE-BIT SIGMA-DELTA MODULATION

引用

IEE PROCEEDINGS-G CIRCUITS DEVICES AND SYSTEMS 1992年第1期139卷 39-44页

作者： YU, J SANDLER, MB HAWKEN, RE King's College London Department of Electronic and Electrical Engineering University of London Strand London WC2R 2LS UK

A fixed step size is usually used for a quantiser in a sigma-delta modulator or noise shaper, but it cannot always match input signals adequately if they are nonstationary, as in the case of music. An attempt at introducing adaptive quantisers, based on a digital maximum-magnitude technique, into 1-bit sigma-delta modulators has been made, although the basic idea appeared about two decades ago. The initial results show it to be a promising technique. The dynamic range of the sigma-delta modulator can be effectively increased by using an adaptive quantiser, and the signal/noise ratio is nearly independent of input level for sinewave inputs. This advantage may increase future applications of sigma-delta modulators.

关键词： adaptive quantization SIGMA-DELTA MODULATION

来源：评论

学校读者我要写书评

暂无评论

adaptive quantization on a Grassmann-Manifold for Limited Feedback Beamforming Systems

引用

IEEE TRANSACTIONS ON SIGNAL PROCESSING 2013年第18期61卷 4450-4462页

作者： Schwarz, Stefan Heath, Robert W., Jr. Rupp, Markus Vienna Univ Technol Inst Telecommun Mobile Commun Grp Vienna Austria Vienna Univ Technol Inst Telecommun Vienna Austria Univ Texas Austin Dept Elect & Comp Engn Austin TX 78712 USA

In this paper we examine delay limited adaptive quantization on the Grassmann manifold of 1-dimensional subspaces in n-dimensional space. Due to strict delay limits, vector quantization over multiple time instants cannot be applied to exploit the temporal correlation of the source signal. Instead, a vector predictive quantizer is proposed that combines prediction and differential quantization algorithms to achieve an efficient quantization of the correlated Grassmannian source. The proposed predictor is based on adaptive finite impulse response filters to adjust to the temporal statistics of the source signal. It is shown that the prediction error in the tangent space associated with the Grassmann manifold behaves approximately Gaussian, provided its norm is sufficiently small. The proposed quantization algorithm is applied to channel state information quantization in multi-user beamforming wireless communication systems. Large throughput gains are demonstrated in comparison to memoryless quantization, due to reduced multi-user interference.

关键词： adaptive quantization channel state information Grassmann manifold limited feedback LTE multi-user MIMO OFDMA quantized feedback

来源：评论

学校读者我要写书评

暂无评论

adaptive quantization AND FAULT-DETECTION IN SMART SENSORS

引用

SENSORS AND ACTUATORS A-PHYSICAL 1995年第1-3期47卷 530-533页

作者： MARTINEZ, D ESTEVE, D Laboratoire d'Analyse et d'Architecture LAAS-CNRS 7 avenue du Colonel Roche 31077 Toulouse France

This paper presents an adaptive maximum entropy quantizer for fault detection in smart sensors. An unsupervised learning rule adapts the quantization thresholds for obtaining equal-probability quantization intervals, therefore giving a reliable nonparametric density estimation. Hypothesis testing based on the likelihood ratio is then applied for detecting abrupt changes in the signal distribution which reflects faulty sensor operations.

关键词： adaptive quantization FAULT DETECTION SMART SENSORS

来源：评论

学校读者我要写书评

暂无评论

adaptive quantization VIA FUZZY CLASSIFIED PRIORITY MAPPING FOR LIVER ULTRASOUND COMPRESSION

引用

INTERNATIONAL JOURNAL OF INNOVATIVE COMPUTING INFORMATION AND CONTROL 2016年第2期12卷 635-649页

作者： Sombutkaew, Rattikorn Chitsobhuk, Orachat Prapruttam, Duangkamon Ruangchaijatuporn, Thumanoon King Mongkuts Inst Technol Ladkrabang Fac Engn Chalongkrung Rd Bangkok 10520 Thailand Mahidol Univ Fac Med Ramathibodi Hosp Dept Diagnost & Therapeut Radiol Bangkok 10400 Thailand

This paper proposes adaptive quantization based on fuzzy classified priority mapping in order to achieve higher encoding efficiency. The priority map serves as a quantization mask, which is adaptively adjusted according to the statistical characteristics in terms of histograms based on the results of Fuzzy C-mean clustering. With its soft clustering property, the results illustrate robustness to ambiguity of the data and thus retain much more information than hard clustering. The priority map represents levels of significance as the Most Significant Group (MSG), the Normal Significant Group (NSG), and the Lowest Significant Group (LSG). The significant candidates of irregular liver tissues requiring special doctor attention will be assigned with higher priority than those from the regular ones. The higher the priority, the greater the number of bits assigned for encoding. An analysis of suitable quantization step size has been conducted. With the selection of appropriate quantization parameters for each priority level, the blocking artifacts can be greatly reduced. This helps to reduce the encoding bit rate and enhance the compression efficiency for the transmission and storage while maintaining an acceptable diagnostic image quality.

关键词： Ultrasound image compression Fuzzy C-mean clustering adaptive quantization Priority mapping

来源：评论

学校读者我要写书评

暂无评论

adaptive quantization and distributed estimation for bandwidth-constraint sensor networks

Adaptive quantization and distributed estimation for bandwid...

引用

IEEE International Symposium on Information Theory

作者： Li, Hongbin Fang, Jun Stevens Inst Technol Dept Elect & Comp Engn Hoboken NJ 07030 USA

ISBN: (纸本)9781424413973

In this paper, the problem of distributed parameter estimation in a wireless sensor network is considered, where due to bandwidth constraint, each sensor node sends only one bit of information per sample to a fusion center. We propose a new distributed adaptive quantization scheme by which each individual sensor node dynamically adjusts the threshold of its quantizer based on earlier transmissions from other sensor nodes. The maximum likelihood estimator (MILE) and the Cramer-Rao bound (CRB) associated with our distributed adaptive quantization scheme are derived. Numerical results depicting the performance and advantages of our approach over a fixed quantization scheme are presented.

关键词： wireless sensor networks distributed estimation adaptive quantization

来源：评论

学校读者我要写书评

暂无评论

adaptive quantization-Based HDR Video Coding with HEVC Main 10 Profile 19

Adaptive Quantization-Based HDR Video Coding with HEVC Main ...

引用

19th IEEE International Symposium on Multimedia (ISM)

作者： Fu, Qingtao Jung, Cheolkon Yang, Ge Xidian Univ Sch Elect Engn Xian Shaanxi Peoples R China

ISBN: (纸本)9781538629376

Perceptual quantization (PQ) for HDR video coding has good perceptual uniformity in the luminance range with the modest bit-depth. However, the dynamic range of HDR videos is not well-used in PQ according to contents, especially for chroma channels. Thus, there exist the wasted dynamic ranges in PQ which cause detail loss and color distortions. In this paper, we propose adaptive quantization-based HDR video coding with HEVC Main 10 Profile. We perform adaptive mapping for 16bit to 10bit conversion based on cumulative distribution function (CDF), i.e. adaptive quantization, instead of linear mapping. First, we obtain CDF of each video frame using histogram analysis. Then, we conduct adaptive quantization based on CDF according to video contents, and generate metadata for inverse adaptive conversion in decoder, i.e. 10bit to 16bit conversion. Experimental results demonstrate that the proposed method reconstructs high-quality HDR videos and achieves a significant improvement in performance in terms of visual assessment and quantitative measurement.

关键词： adaptive quantization dynamic range HDR video coding HEVC metadata perceptual quantization

来源：评论

学校读者我要写书评

暂无评论

adaptive quantization in DCT Domain for Distributed Video Coding

Adaptive Quantization in DCT Domain for Distributed Video Co...

引用

Data Compression Conference (DCC)

作者： Yang, Chun-Ling Xiao, Dong-Qin Po, Lai-Man Mo, Wang-Hua South China Univ Technol Guangzhou Guangdong Peoples R China City Univ Hong Kong Hong Kong Peoples R China

ISBN: (纸本)9780769543529

An improved quantization scheme with adaptive quantization levels (AQL) for transform-domain distributed video coding (DVC) is presented in this paper. In the proposed AQL, the quantization level for each DCT band is adaptively determined according to its importance measured by the magnitude of large coefficients. Experimental results demonstrate that the proposed AQL can achieve significant rate-distortion improvement as compared with uniform scalar quantization scheme with fixed quantization levels.

关键词： Video coding quantization level adaptive quantization discrete cosine transforms DCT gene uniform scalar quantization Distribute

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：