Memory-Based Optimization Methods for Model-Agnostic Meta-Learning and Personalized Federated Learning
Memory-Based Optimization Methods for Model-Agnostic Meta-Learning and Personalized Federated Learning
复制标题
DOI：
--
复制
发表时间：
2021-06
期刊：
影响因子：
--
通讯作者：
Bokun Wang;Zhuoning Yuan;Yiming Ying;Tianbao Yang
中科院分区：
文献类型：
--
作者：
Bokun Wang;Zhuoning Yuan;Yiming Ying;Tianbao Yang
近年来，模型不可知元学习（MAML）已成为一个热门的研究领域。然而，MAML的随机优化研究还不够成熟。现有的MAML算法依赖于“插曲”的想法，通过采样一些任务和数据点来在每次迭代中更新元模型。尽管如此，这些算法要么不能保证以固定的小批大小收敛，要么需要在每次迭代中处理大量任务，这不适合连续学习或跨设备联合学习，因为每次迭代或每轮只有少量任务可用。为了解决这些问题，本文提出了基于记忆的MAML随机算法，该算法收敛于误差消失。所提出的算法要求每次迭代采样恒定数量的任务和数据样本，使它们适合于持续学习的场景。此外，我们引入了一种通信高效的基于内存的MAML算法，用于跨设备（带客户端采样）和跨筒仓（不带客户端采样）设置下的个性化联邦学习。理论分析完善了MAML的优化理论，实证结果证实了理论结论。感兴趣的读者可以在\url{https://github.com/bokun-wang/moml}访问我们的代码。
In recent years, model-agnostic meta-learning (MAML) has become a popular research area. However, the stochastic optimization of MAML is still underdeveloped. Existing MAML algorithms rely on the ``episode'' idea by sampling a few tasks and data points to update the meta-model at each iteration. Nonetheless, these algorithms either fail to guarantee convergence with a constant mini-batch size or require processing a large number of tasks at every iteration, which is unsuitable for continual learning or cross-device federated learning where only a small number of tasks are available per iteration or per round. To address these issues, this paper proposes memory-based stochastic algorithms for MAML that converge with vanishing error. The proposed algorithms require sampling a constant number of tasks and data samples per iteration, making them suitable for the continual learning scenario. Moreover, we introduce a communication-efficient memory-based MAML algorithm for personalized federated learning in cross-device (with client sampling) and cross-silo (without client sampling) settings. Our theoretical analysis improves the optimization theory for MAML, and our empirical results corroborate our theoretical findings. Interested readers can access our code at \url{https://github.com/bokun-wang/moml}.