communication cost
The resources needed for data transfer between components of a distributed AI system, in terms of bandwidth, energy, or latency, critical for efficient model deployment and operation in cloud and edge environments.
- Covariances for Free: Exploiting Mean Distributions for Training-free Federated Learning
- Layer-wise Update Aggregation with Recycling for Communication-Efficient Federated Learning
- Protocols for Verifying Smooth Strategies in Bandits and Games
- Sketched Gaussian Mechanism for Private Federated Learning
- Soft-consensual Federated Learning for Data Heterogeneity via Multiple Paths