Date
2023
details
CEA-List
Designing an interface to explain the CEA's breakthrough blockchain-based federated learning platform.
Recent developments in artificial intelligence are opening up immense opportunities. But until now, training mechanisms AI systems rely on have struggled to be efficient while preserving users’ data privacy at the same time.
Following a first fruitful collaboration with the LIST laboratory at the CEA - France’s leading public research organization - Units worked with a team of researchers to design and develop a demonstrator to present Fantastyc, a groundbreaking decentralized, open source machine learning framework.
Fantastyc leverages blockchain to democratize federated learning, making it auditable, robust, and decentralized. The aim of the laboratory was to demonstrate this new framework to a wide audience with unequal knowledge of AI and Web3 technologies.
In close coordination with the lab’s research team, we identified the main conceptual challenges and prioritized information in order to strike the right balance between theoretical and practical considerations.
The demonstrator adopts a chronological approach to the topic, offering an instructive overview of the key paradigms that have shaped recent AI training history, while also providing practical and detailed explanations through concrete examples within each paradigm. We reframed the overall narrative around the idea of a trilemma between robustness, privacy, and scalability, resulting in a compelling and intuitive navigation principle, enabling comparative views between the different systems.
In parallel, we designed and developed a clear visual vocabulary based on an isometric grid and 3D shapes, and personified stakeholders to make behaviors and relationships easily understandable while guaranteeing scientific consistency.
The result is a clear, easy-to-navigate animated interface with strong graphic appeal inspired by both strategy games and editorial codes. The narrative structure helps underline Fantastyc’s key practical benefits but also how the Lab’s work contributes to a fair democratization of AI by empowering all stakeholders involved in the training process and using blockchain to foster digital trust.
Recent developments in artificial intelligence are opening up immense opportunities. But until now, training mechanisms AI systems rely on have struggled to be efficient while preserving users’ data privacy at the same time.
Following a first fruitful collaboration with the LIST laboratory at the CEA - France’s leading public research organization - Units worked with a team of researchers to design and develop a demonstrator to present Fantastyc, a groundbreaking decentralized, open source machine learning framework.
Fantastyc leverages blockchain to democratize federated learning, making it auditable, robust, and decentralized. The aim of the laboratory was to demonstrate this new framework to a wide audience with unequal knowledge of AI and Web3 technologies.
In close coordination with the lab’s research team, we identified the main conceptual challenges and prioritized information in order to strike the right balance between theoretical and practical considerations.
The demonstrator adopts a chronological approach to the topic, offering an instructive overview of the key paradigms that have shaped recent AI training history, while also providing practical and detailed explanations through concrete examples within each paradigm. We reframed the overall narrative around the idea of a trilemma between robustness, privacy, and scalability, resulting in a compelling and intuitive navigation principle, enabling comparative views between the different systems.
In parallel, we designed and developed a clear visual vocabulary based on an isometric grid and 3D shapes, and personified stakeholders to make behaviors and relationships easily understandable while guaranteeing scientific consistency.
The result is a clear, easy-to-navigate animated interface with strong graphic appeal inspired by both strategy games and editorial codes. The narrative structure helps underline Fantastyc’s key practical benefits but also how the Lab’s work contributes to a fair democratization of AI by empowering all stakeholders involved in the training process and using blockchain to foster digital trust.
Designing an interface to explain the CEA's breakthrough blockchain-based federated learning platform.
Recent developments in artificial intelligence are opening up immense opportunities. But until now, training mechanisms AI systems rely on have struggled to be efficient while preserving users’ data privacy at the same time.
Following a first fruitful collaboration with the LIST laboratory at the CEA - France’s leading public research organization - Units worked with a team of researchers to design and develop a demonstrator to present Fantastyc, a groundbreaking decentralized, open source machine learning framework.
Fantastyc leverages blockchain to democratize federated learning, making it auditable, robust, and decentralized. The aim of the laboratory was to demonstrate this new framework to a wide audience with unequal knowledge of AI and Web3 technologies.
In close coordination with the lab’s research team, we identified the main conceptual challenges and prioritized information in order to strike the right balance between theoretical and practical considerations.
The demonstrator adopts a chronological approach to the topic, offering an instructive overview of the key paradigms that have shaped recent AI training history, while also providing practical and detailed explanations through concrete examples within each paradigm. We reframed the overall narrative around the idea of a trilemma between robustness, privacy, and scalability, resulting in a compelling and intuitive navigation principle, enabling comparative views between the different systems.
In parallel, we designed and developed a clear visual vocabulary based on an isometric grid and 3D shapes, and personified stakeholders to make behaviors and relationships easily understandable while guaranteeing scientific consistency.
The result is a clear, easy-to-navigate animated interface with strong graphic appeal inspired by both strategy games and editorial codes. The narrative structure helps underline Fantastyc’s key practical benefits but also how the Lab’s work contributes to a fair democratization of AI by empowering all stakeholders involved in the training process and using blockchain to foster digital trust.
Recent developments in artificial intelligence are opening up immense opportunities. But until now, training mechanisms AI systems rely on have struggled to be efficient while preserving users’ data privacy at the same time.
Following a first fruitful collaboration with the LIST laboratory at the CEA - France’s leading public research organization - Units worked with a team of researchers to design and develop a demonstrator to present Fantastyc, a groundbreaking decentralized, open source machine learning framework.
Fantastyc leverages blockchain to democratize federated learning, making it auditable, robust, and decentralized. The aim of the laboratory was to demonstrate this new framework to a wide audience with unequal knowledge of AI and Web3 technologies.
In close coordination with the lab’s research team, we identified the main conceptual challenges and prioritized information in order to strike the right balance between theoretical and practical considerations.
The demonstrator adopts a chronological approach to the topic, offering an instructive overview of the key paradigms that have shaped recent AI training history, while also providing practical and detailed explanations through concrete examples within each paradigm. We reframed the overall narrative around the idea of a trilemma between robustness, privacy, and scalability, resulting in a compelling and intuitive navigation principle, enabling comparative views between the different systems.
In parallel, we designed and developed a clear visual vocabulary based on an isometric grid and 3D shapes, and personified stakeholders to make behaviors and relationships easily understandable while guaranteeing scientific consistency.
The result is a clear, easy-to-navigate animated interface with strong graphic appeal inspired by both strategy games and editorial codes. The narrative structure helps underline Fantastyc’s key practical benefits but also how the Lab’s work contributes to a fair democratization of AI by empowering all stakeholders involved in the training process and using blockchain to foster digital trust.
