41,983 Hits in 4.8 sec

Federated Visual Classification with Real-World Data Distribution [article]

Tzu-Ming Harry Hsu, Hang Qi, Matthew Brown
2020 arXiv   pre-print
To do so, we introduce two new large-scale datasets for species and landmark classification, with realistic per-user data splits that simulate real-world edge learning scenarios.  ...  In this work, we characterize the effect these real-world data distributions have on distributed learning, using as a benchmark the standard Federated Averaging (FedAvg) algorithm.  ...  Ours is the first work to our knowledge that attempts to train large-scale visual classification models for real-world problems in a federated setting.  ... 
arXiv:2003.08082v3 fatcat:inm5lx3owrh75hs4mzgvgihpiu

Exploiting Defenses against GAN-Based Feature Inference Attacks in Federated Learning [article]

Xianglong Zhang, Xinjian Luo
2021 arXiv   pre-print
We redesign the structure of the victim's GAN to encourage it to learn the classification features (instead of the visual features) of the real images.  ...  In this paper, we exploit defenses against GAN-based attacks in federated learning, and propose a framework, Anti-GAN, to prevent attackers from learning the real distribution of the victim's data.  ...  We redesign the structure of the GAN model to maximally preserve the classification features of the real images. • We conduct extensive experiments on three real-world datasets to evaluate the performance  ... 
arXiv:2004.12571v2 fatcat:432zj7oelbgt3k6ikpp2afc3ga

Prototype Guided Federated Learning of Visual Feature Representations [article]

Umberto Michieli, Mete Ozay
2021 arXiv   pre-print
Federated Learning (FL) is a framework which enables distributed model training using a large corpus of decentralized training data.  ...  To this end, we introduce a method, called FedProto, which computes client deviations using margins of prototypical representations learned on distributed data, and applies them to drive federated optimization  ...  Real World Image Classification Datasets To further investigate accuracy on classification data, we explore real-world image classification datasets, inspired from [5, 31, 35] . • MNIST: It is a classification  ... 
arXiv:2105.08982v1 fatcat:wbw7g42vmff5tlyq6kdzs2hcbq

Measuring the Effects of Non-Identical Data Distribution for Federated Visual Classification [article]

Tzu-Ming Harry Hsu, Hang Qi, Matthew Brown
2019 arXiv   pre-print
Federated Learning enables visual models to be trained in a privacy-preserving way using real-world data from mobile devices.  ...  In this work, we look at the effect such non-identical data distributions has on visual classification via Federated Learning.  ...  Synthetic Non-Identical Client Data In our visual classification task, we assume on every client training examples are drawn independently with class labels following a categorical distribution over N  ... 
arXiv:1909.06335v1 fatcat:c7esqx4vsrh5hkuhbo5suucpz4

TinyFedTL: Federated Transfer Learning on Tiny Devices [article]

Kavya Kopparapu, Eric Lin
2021 arXiv   pre-print
In addition, the deployment of TinyML hardware in the real world has significant memory and communication constraints that traditional ML fails to address.  ...  TinyML has rose to popularity in an era where data is everywhere. However, the data that is in most demand is subject to strict privacy and security guarantees.  ...  Conclusion We have shown a successful first-attempt at deploying on-device federated learning.  ... 
arXiv:2110.01107v1 fatcat:3ziycfwzova5ngf54uvk7h7tnu

Federated Generative Adversarial Learning [article]

Chenyou Fan, Ping Liu
2020 arXiv   pre-print
Generative adversarial networks (GANs) have achieved advancement in various real-world applications, such as image editing, style transfer, scene generations, etc.  ...  However, like other deep learning models, GANs are also suffering from data limitation problems in real cases.  ...  Also known as Wasserstein distance [2] , it measures the distance between the distribution of real data P r and generated data P g .  ... 
arXiv:2005.03793v3 fatcat:bkrsva6iubfdxgzq43blalx74m

Digital Twin: Technology Evolution Stages and Implementation Layers with Technology Elements

Deuk-Young Jeong, Myung-Sun Baek, Tae-Beom Lim, Yong-Woon Kim, Se-Han Kim, Yong-Tae Lee, Woo-Sug Jung, In-Bok Lee
2022 IEEE Access  
., people, objects, spaces, systems, and processes) in the real world into digital objects in the digital world.  ...  It also provides various simulations to solve problems in the real world or to improve situational operations.  ...  real world Intelligence framework technology for distributed storage, high-speed processing, and analysis efficiency of large data about digital twin objects Data collection and processing technology Technology  ... 
doi:10.1109/access.2022.3174220 fatcat:pkiogjw2qrhrjeoidhswasjmta

Using Spatial Concepts to Integrate Data and Information from Various Sources for a Knowledge-based Assessment of Impervious Surfaces

Thomas Strasser, Dirk Tiede
2020 GI_FORUM - Journal for Geographic Information Science  
size and shape) and on building a knowledge-base for the classification of real-world objects.  ...  For a study area covering the municipality of Hallein (Austria), we discuss preliminary results with a focus on real-world object characterization (including surface material, spectral reflectivity, object  ...  of real-world objects are addressable with big-data timeseries analyses.  ... 
doi:10.1553/giscience2020_02_s147 fatcat:qnk2u6yrbzh7hjre5qgr3soqj4

FedGL: Federated Graph Learning Framework with Global Self-Supervision [article]

Chuan Chen, Weibo Hu, Ziyue Xu, Zibin Zheng
2021 arXiv   pre-print
Graph data are ubiquitous in the real world. Graph learning (GL) tries to mine and analyze graph data so that valuable information can be discovered.  ...  However, in practical scenarios, graph data are usually distributed in different organizations, i.e., the curse of isolated data islands.  ...  In order to simulate the graph data distribution in the real world, the graph data of each client comes from the random sampling results of the experimental datasets under different proportions to Table  ... 
arXiv:2105.03170v1 fatcat:lvs66x5oc5f27evilxsjc5ormy

GRP-FED: Addressing Client Imbalance in Federated Learning via Global-Regularized Personalization [article]

Yen-Hsiu Chou, Shenda Hong, Chenxi Sun, Derun Cai, Moxian Song, Hongyan Li
2021 arXiv   pre-print
Each local model is learned from the local data and aligns with its distribution for customization.  ...  Since data is presented long-tailed in reality, it is challenging for Federated Learning (FL) to train across decentralized clients as practical applications.  ...  Experiments Experimental Setting Dataset We evaluate our GRP-FED on two federated classification datasets, real-world MIT-BIH (Goldberger et al., 2000) , and synthesis CIFAR-10 (Krizhevsky & Hinton  ... 
arXiv:2108.13858v1 fatcat:fwc47tl7breifi2lhb5lqsixgi

FedGraphNN: A Federated Learning System and Benchmark for Graph Neural Networks [article]

Chaoyang He, Keshav Balasubramanian, Emir Ceyani, Carl Yang, Han Xie, Lichao Sun, Lifang He, Liangwei Yang, Philip S. Yu, Yu Rong, Peilin Zhao, Junzhou Huang (+2 others)
2021 arXiv   pre-print
However, centralizing a massive amount of real-world graph data for GNN training is prohibitive due to privacy concerns, regulation restrictions, and commercial competitions.  ...  Federated learning (FL), a trending distributed learning paradigm, provides possibilities to solve this challenge while preserving data privacy.  ...  Federated visual classification with real-world data distribution. arXiv preprint arXiv:2003.08082, 2020. [38] 04977, 2019.  ... 
arXiv:2104.07145v2 fatcat:l7p3eb6tjbgztocugyohem2qea

Designing ECG Monitoring Healthcare System with Federated Transfer Learning and Explainable AI [article]

Ali Raza, Kim Phuc Tran, Ludovic Koehl, Shujun Li
2021 arXiv   pre-print
This limits the application of deep learning in real-world health systems.  ...  The federated setting is used to solve issues such as data availability and privacy concerns.  ...  The resulting data distribution after upsampling is shown in Figure 8 . Furthermore, this dataset is highly preprocessed, but in real-world scenarios, the ECG data collected is always noisy.  ... 
arXiv:2105.12497v1 fatcat:xz2gikw3trh3hgi3be6s6khg2a

Machine learning in translation

2021 Nature Biomedical Engineering  
GANs are deep-learning architectures that mimic a zero-sum game for training a generator network so that it creates new datasets that mirror the data distribution of a real dataset by 'competing' with  ...  data onto a lower number of dimensions to aid interpretability and visualization.  ... 
doi:10.1038/s41551-021-00758-1 pmid:34131319 fatcat:3xxs7qdoevakjjjev7fi3fks6u

Multi-stakeholder interactive simulation for federated satellite systems

Paul T. Grogan, Alessandro Golkar, Seiko Shirasaka, Olivier L. de Weck
2014 2014 IEEE Aerospace Conference  
Federated satellite systems (FSS) are a new class of space-based systems which emphasize a distributed architecture.  ...  New information exchanging functions among FSS members enable data transportation, storage, and processing as on-orbit services.  ...  The first federate focuses on data visualization and does not control any federation objects.  ... 
doi:10.1109/aero.2014.6836253 fatcat:inunnhbhorceteftfymxhxriq4

Addressing modern and practical challenges in machine learning: A survey of online federated and transfer learning [article]

Shuang Dai, Fanlin Meng
2022 arXiv   pre-print
Online federated learning (OFL) and online transfer learning (OTL) are two collaborative paradigms for overcoming modern machine learning challenges such as data silos, streaming data, and data security  ...  This survey explored OFL and OTL throughout their major evolutionary routes to enhance understanding of online federated and transfer learning.  ...  of data-driven real-world machine learning systems.  ... 
arXiv:2202.03070v1 fatcat:aiuhmm5dejbhxpn4nkxutpltry
« Previous Showing results 1 — 15 out of 41,983 results