首页 > TAG信息列表 > Averaging
Adversarial Deep Averaging Networks for Cross-Lingual Sentiment Classification
https://github.com/ccsasuke/adan 作者提出了Adversarial Deep Averaging Network (ADAN)模型，将高资源的源语言标记数据迁移到低资源未标记数据。ADAN有两个不同的分支：一个sentiment classifier和一个adversarial language discriminator。这两个分支都将feature extractorSCAFFOLD: Stochastic Controlled Averaging for Federated Learning
背景 传统的联邦学习在数据异构(non-iid)的场景中很容易产生“客户漂移”(client-drift)的现象，这会导致系统的收敛不稳定或者缓慢 贡献 提出了考虑到client sampling和数据异构的一个更接近的收敛边界 证明即便没有client sampling，使用全批次梯度(full batch gradients)，传ICML联邦学习论文解读 SCAFFOLD: Stochastic Controlled Averaging for Federated Learning
论文解读：SCAFFOLD: Stochastic Controlled Averaging for Federated Learning 作者：Sai Praneeth Karimireddy, Satyen Kale, Mehryar Mohri,Sashank J. Reddi, Sebastian U. Stich, Ananda Theertha Suresh 论文地址：https://arxiv.org/abs/1910.06378 一、背景介绍 联邦学习