.png){kind=icon}
返回
Quantum Federated Learning With Decentralized Data
.png){kind=icon}
PRE
AI
DOI：10.1109/JSTQE.2022.3170150.png)
.png){kind=icon}
原文求助.png){kind=icon}
分享.png){kind=icon}
收藏
摘要
En 中文
Variational quantum algorithm (VQA) accesses the centralized data to train the model, and using distributed computing can significantly improve the training overhead; however, the data is privacy sensitive. In this paper, we propose communication-efficient learning of VQA from decentralized data, which is so-called quantum federated learning (QFL). Motivated by the classical federated learning algorithm, we improve data privacy by aggregating updates from local computation to share model parameters. Here, aiming to find approximate optima in the parameter landscape, we develop an extension of the conventional VQA. Finally, we deploy on the TensorFlow Quantum processor within variational quantum tensor networks classifiers, approximate quantum optimization for the Ising model, and variational quantum eigensolver for molecular hydrogen. Our algorithm demonstrates model accuracy from decentralized data, which have higher performance on near-term processors. Importantly, QFL may inspire new investigations in the field of secure quantum machine learning.
Keyword:
Training
Data models
Quantum computing
Encoding
Servers
Collaborative work
Machine learning
Quantum algorithm
quantum computing
quantum information
quantum machine learning
期刊
I
IF:
5.1
论文数:
5.6K
被引数:
1.2W
机构
引用论文
A variational eigenvalue solver on a photonic quantum processor光子量子处理器上的变分特征值求解器.png)
NATURE COMMUNICATIONS
OAAI
NATURE COMMUNICATIONSIF15.7
OAAI分享收藏Stochastic gradient descent for hybrid quantum-classical optimization混合量子-经典优化的随机梯度下降QUANTUM
OAAI
QUANTUMIF5.4
OAAI分享收藏分享收藏Effect of surface roughness on runoff and erosion in a mediterranean ecosystem: the role of fire
PREAI
分享收藏
查看更多内容