.png){kind=icon}
返回
Federated learning with t 1 regularization
.png){kind=icon}
PRE
AI
DOI：10.1016/j.patrec.2023.05.030.png)
.png){kind=icon}
原文求助.png){kind=icon}
分享.png){kind=icon}
收藏
摘要
En 中文
Federated Learning (FL) is a widely adopted deep learning method that does not require the collection of raw training data and solves specific learning tasks by federating distributed devices. Due to the het-erogeneous distribution of data across clients, the clients will drift toward the local optimal solutions during local training and result in different local models. The global model after aggregating these differ-ent local models may keep away from the global optimal solution. This phenomenon is known as client drift, which often hinders the performance of FL. Parameter regularization methods address this chal-lenge of client drift by controlling the update direction of each client. They consider the global model as both the starting point and a reference for the induction bias in the penalty. However, the exist-ing regularization approaches produce dense solutions so that all parameters need to be updated dur-ing local model training. At the same time, we note that some studies on deep learning have found that it is unnecessary to update all parameters at each round. Therefore, in this work, we design a novel FL training approach called Fed �1 which can alleviate the performance degradation of FL by up-dating only part of the parameters at each round. e 1 regularization is utilized to control the update di-rection of each client and avoid unnecessary parameter updates at the same time. To our knowledge, our study is the first to introduce sparse regularization term to correct the local training of individual clients in FL. We design a stochastic subgradient descent algorithm to train the t1-regularized nonsmooth model. The comparison experiments with state-of-the-art baselines verify the superiority of the proposed approach. & COPY; 2023 Published by Elsevier B.V.
Keyword:
Federated learning
Z 1 Regularization
Stochastic subgradient descent
期刊
IF:
3.3
论文数:
8.0K
被引数:
1.6W
机构
引用论文
Efficient construction of sparse radial basis function neural networks using L1-regularization.png)
NEURAL NETWORKS
NEURAL NETWORKSIF6.3
PREAI
分享收藏Efficient and robust TWSVM classification via a minimum L1-norm distance metric criterionMACHINE LEARNING
OAAI
MACHINE LEARNINGIF2.9
OAAI分享收藏Heterogeneous Feature Selection With Multi-Modal Deep Neural Networks and Sparse Group LASSO基于多模态深度神经网络和稀疏组LASSO的异构特征选择OAAI
OAAI分享收藏分享收藏Synthesis of an indole containing KDR kinase inhibitor by tandem Sonogashira coupling-5-endo-dig-cyclization as a key step以串联Sonogashira coupling-5-endo-dig-cyclization为关键步骤合成含吲哚的KDR激酶抑制剂Tetrahedron
TetrahedronIF0
PREAI
分享收藏Least squares twin bounded support vector machines based on L1-norm distance metric for classificationPATTERN RECOGNITIONOAAI
PATTERN RECOGNITIONIF7.6
OAAI分享收藏
查看更多内容