Overview

hivenet offers a highly original data storage architecture in which data is stored in a distributed and secure manner on the spare storage resources of participants, based on apeer-to-peer structure. This structure ensures scalability, resilience and voluntary sharing of data between users. The aim of this new challenge (after Alvearium) between hivenet and Inria is to push the limits of distributed AI computing. Its goal is to demonstrate that even the most demanding AI and Big Data applications can run efficiently on heterogeneous, distributed, and volatile resources — while maintaining accuracy, ensuring privacy, and reducing environmental impact.

The document describing the project proposal can be found here

Research directions

• Frugality

  Techniques to adapt training and inference to limited, dynamic resources by optimizing memory, computation and communications.

• Security & Confidentiality

  Protect data and models through encryption, trusted execution environments, and robust defenses against attacks.

• Volatility

  Robustness and fault tolerance when computing nodes unpredictably join, leave, or fail.