类型 | 说明 |
---|---|
论文信息 | Communication-Efficient Learning of Deep Networks from Decentralized Data H.Brendan McMahan, Eider Moore, Daniel Ramage, Seth Hampson, Blaise Agüera y Arcas |
研究的问题 | 移动设备的去中心化数据的高效率通信的分布式联合平均算法 |
算法名称 | FederatedAveraging Algorithm(FedAVG) |
前景知识 | – |
有限和形式 | min _ ω ∈ R d f ( ω ) w h e r e f ( ω ) = def 1 n ∑ i = 1 n f _ i ( ω ) . \min\_{\omega\in\mathbb{R}^{d}} f\left(\omega\right) where f\left(\omega\right) \stackrel{\text{def}}{=}\frac{1}{n}\sum^{n}_{i=1}f\_{i}\left(\omega\right). min_ω∈Rdf(ω) where f(ω)=defn1i=1∑nf_i(ω).For a machine learning problem: f i ( ω ) = ℓ ( x i , y i ; ω ) f_{i}\left(\omega\right)=\ell\left(x_{i},y_{i};\omega\right) fi(ω)=ℓ(xi,yi;ω) |
分布式形式 | f ( ω ) = ∑ k = 1 K n k n F k ( ω ) w h e r e F k ( ω ) = 1 n k ∑ i |
Communication-Efficient Learning of Deep Networks from Decentralized Data
最新推荐文章于 2022-11-17 11:04:50 发布