The FENIX project

Project owned by AI Sweden and part of Industrial Immersion Exchange program English
342d ago update
The FENIX Project - Federated Edge Network Infrastructure for neXt gen foundation models.

The field of AI has a need for an established benchmark for developing and deploying federated foundation models, as well as a comprehensive understanding of the potential security risks involved. In particular, the challenge is to leverage a centralized foundation model like GPT-SW3, and investigate its potential evolution towards a federated model. 

This will require designing and demonstrating a next-generation infrastructure for training, testing, validating, and securely deploying federated foundation models. Alongside, assessing cybersecurity threats inherent to these distributed systems will be crucial. Ultimately, the goal is to contribute to the development of innovative solutions for the secure, federated deployment of AI models, setting the beginnings of new industry standards and guiding future advancements.

Two main goals:
 

  • Create benchmark for training and, testing foundation models in a federated setup
     
  • Impact Assessment of Adversarial Attacks on Federated Models and Possible defenses

Attributes

Technology
Information Technology
Innovation
Smarter Product or Service
Language
Federated
pages

Resources

2023-12-07 13:54 File (PDF, Word, PPT, etc)
2023-12-07 13:50 Code / Framework
Repo for the Foundation project at A.I Sweden. Contribute to BQ-QB/Foundation_A.I_SWE development by creating an account on GitHub.

There were not enough time to do the planned code refactoring during the project time.

2023-12-07 13:43 Weblink
Twelve graduate students presented their research on artificial intelligence projects on Friday at Dakota State University.
2023-12-07 13:40 Weblink
Research report for the project