Federated learning (FL) enables edge-devices to collaboratively learn a model without disclosing their private data to a central aggregating server. Most existing FL algorithms require models of identical architecture to be deployed across the clients and server, making it infeasible to train large models due to clients' limited system resources. In this work, we propose a novel ensemble knowledge transfer method named Fed-ET in which small models (different in architecture) are trained on clients, and used to train a larger model at the server. Unlike in conventional ensemble learning, in FL the ensemble can be trained on clients' highly heterogeneous data. Cognizant of this property, Fed-ET uses a weighted consensus distillation scheme with diversity regularization that efficiently extracts reliable consensus from the ensemble while improving generalization by exploiting the diversity within the ensemble. We show the generalization bound for the ensemble of weighted models trained on heterogeneous datasets that supports the intuition of Fed-ET. Our experiments on image and language tasks show that Fed-ET significantly outperforms other state-of-the-art FL algorithms with fewer communicated parameters, and is also robust against high data-heterogeneity.

Federated Learning (FL) is a variant of distributed learning where edge devices collaborate to learn a model without sharing their data with the central server or each other. We refer to the process of training multiple independent models simultaneously in a federated setting using a common pool of clients as multi-model FL. In this work, we propose two variants of the popular FedAvg algorithm for multi-model FL, with provable convergence guarantees. We further show that for the same amount of computation, multi-model FL can have better performance than training each model separately. We supplement our theoretical results with experiments in strongly convex, convex, and non-convex settings.

We study a sequential resource allocation problem where, at each round, the decision-maker needs to allocate its limited budget among different available entities. In doing so, the decision-maker obtains the reward for each entity in that round. The goal of the decision-maker is to maximize the expected cumulative reward or equivalently minimize cumulative regret over a total of T rounds. Sequential resource allocation can be modeled as a combinatorial bandit by viewing the allocation of a budget to an entity as a base arm. In the context of resource allocation, the rewards received under different budget allocations are likely to be correlated. We propose a novel correlated combinatorial bandit framework that explicitly models such correlations. We develop a novel Correlated-UCB algorithm for online resource allocation, which yields significantly reduced regret relative to correlation-agnostic algorithms. In certain cases, our proposed algorithm even achieves bounded regret, which is an order-wise reduction in the regret relative to the correlation-agnostic approach which incurs logarithmic regret under all scenarios. We validate these performance gains through experiments on several applications such as online power allocation across wireless channels, job scheduling in multi-server systems and online access point assignment.

No abstract

Scope of the WorkshopRecently, Deep Learning (DL) has received tremendous attention in the research community because of the impressive results obtained for a large number of machine learning problems. The success of state-ofthe-art deep learning systems relies on training deep neural networks over a massive amount of training data, which typically requires a large-scale distributed computing infrastructure to run. In order to run these jobs in a scalable and efficient manner, on cloud infrastructure or dedicated HPC systems, several interesting research topics have emerged which are specific to DL. The sheer size and complexity of deep learning models when trained over a large amount of data makes them harder to converge in a reasonable amount of time. It demands advancement along multiple research directions such as, model/data parallelism, model/data compression, distributed optimization algorithms for DL convergence, synchronization strategies, efficient communication and specific hardware acceleration.