Keith Rush
Keith is a researcher in distributed and private learning, with background in mathematics.
Authored Publications
Google Publications
Other Publications
Sort By
Preview abstract
Model sizes are limited in Federated Learning due to communication bandwidth constraints and on-device memory constraints. The success of scaling model sizes in other machine learning domains, especially when it comes to generalizing to new data distributions, motivates the development of methods of training large scale models in Federated Learning. Inspired by dropout, [3] proposed Federated Dropout as a way of scaling up model sizes: clients train randomly selected subsets of the larger server model. In spite of the promising empirical results and the many other works that build on it [1, 8, 13], we argue in this paper that the metrics used to measure performance of Federated Dropout and its variants are misleading. We propose and perform new experiments which suggest that Federated Dropout is actually detrimental to scaling efforts. We show how a simple ensembling technique outperforms Federated Dropout and other baselines. We perform ablations which suggest that the best performing variations of Federated Dropout attempt to approximate ensembling. The simplicity of ensembling allows for easy, practical implementations. Furthermore, our ensembling technique naturally leverages the parallelizable nature of Federated Learning—recall that it is easy to train several models independently because there are a lot of clients and server-compute is not the bottleneck. Ensembling’s strong performance against our baselines suggests that Federated Learning models may be more easily scaled than previously thought e.g., via boosting.
View details
Preview abstract
Personalization methods in federated learning aim to balance the benefits of federated and local training for data availability, communication cost, and robustness to client heterogeneity. Approaches that require clients to communicate all model parameters can be undesirable due to privacy and communication constraints. Other approaches require always-available or stateful clients, impractical in large-scale cross-device settings. We introduce Federated Reconstruction, the first model-agnostic framework for partially local federated learning suitable for training and inference at scale. We motivate the framework via a connection to model-agnostic meta learning, empirically demonstrate its performance over existing approaches for collaborative filtering and next word prediction, and release an open-source library for evaluating approaches in this setting. We also describe the successful deployment of this approach at scale for federated collaborative filtering in a mobile keyboard application.
View details
Adaptive Federated Optimization
Jakub Konečný
(2021)
Preview abstract
Federated learning is a distributed machine learning paradigm in which a large number of clients coordinate with a central server to learn a model without sharing their own training data. Due to the heterogeneity of the client datasets, standard federated optimization methods such as Federated Averaging (FedAvg) are often difficult to tune and exhibit unfavorable convergence behavior. In non-federated settings, adaptive optimization methods have had notable success in combating such issues. In this work, we propose federated versions of adaptive optimizers, including Adagrad, Yogi and Adam, and analyze their convergence in the presence of heterogeneous data for general nonconvex settings. Our results highlight the interplay between client heterogeneity and communication efficiency. We also perform extensive experiments on these methods and show that the use of adaptive optimizers can improve the performance of federated learning.
View details
Preview abstract
We study personalization of supervised learning with user-level differential privacy. Consider a setting with many users, each of whom has a training data set drawn from their own distribution $P_i$. Assuming some shared structure among the problems $P_i$, can users collectively learn the shared structure---and solve their tasks better than they could individually---while preserving the privacy of their data? We formulate this question using joint, \textit{user-level} differential privacy---that is, we control what is leaked about each user's entire data set.
We provide algorithms that exploit popular non-private approaches in this domain like the Almost-No-Inner-Loop (ANIL) method, and give strong user-level privacy guarantees for our general approach. When the problems $P_i$ are linear regression problems with each user's regression vector lying in a common, unknown low-dimensional subspace, we show that our efficient algorithms satisfy nearly optimal estimation error guarantees. We also establish a general, information-theoretic upper bound via an exponential mechanism-based algorithm.
Finally, we demonstrate empirically (through experiments on synthetic data sets) that our framework not only performs well in the studied linear regression setting, but also extends to other settings like logistic regression that are not captured by our estimation error analysis.
View details
No Results Found
