Edge Intelligence

Low-complexity distributed and frugal methods

About Us

Edge Intelligence is a research program of the MIAI institute that works on efficient resource management and distributed/federated Machine Learning. For instance, we promote locally distributed computations related to AI to avoid the waste of energy lost during data transfers. Thanks to a small data center close to the edge of the network, data produced on mobile devices can be pre-computed in low latency networks. Further computations may then be executed on a regular data center over the Cloud.

Fields of Interest

Distributed Learning

Distribute the computation for better efficiency. (E.g. computing power, workload, storage management)

Federated learning

Collaborate with the intent to keep the data of each agent heterogeneous, local, and private. Federated learning can be centralized, decentralized, and semi-centralized.

Task management

Efficient allocation and execution of the tasks in the appropriate heterogeneous and dynamic distributed computing devices connected at the edge level.

Online Learning

Predict better for future data. In online settings, data becomes available in sequential order.