Trustworthy artificial intelligence (AI) is a critical concern when it comes to edge computing. Edge computing involves processing data at the edge of the network, close to where the data is generated, rather than sending it to a centralized location. This approach can provide faster and more efficient data processing, but it also requires that AI systems operate autonomously and reliably in remote environments. Therefore, it is important that AI systems in edge computing are trustworthy, meaning they are transparent, secure, and accurate. Trustworthy AI can help ensure that edge computing systems can make informed and reliable decisions, which is especially important in critical applications like healthcare, autonomous vehicles and industrial automation.

In SMARTY, we describe trustworthy AI as AI systems that can be deployed transparently from any point in the network, securely and accurately. Fundamentally, these requirements translate into offering protection of data-in-transit and data-in-process, while ensuring accessibility and low latency.

SMARTY has a primary objective of enabling secure and trustworthy dynamic integration of decentralized intelligence within the cloud-edge continuum, while ensuring reliability, privacy, and scalability at runtime. The technology developed in SMARTY will help develop secure on-board electronics for the automotive sector, secure edge computing platforms for telco-operators, secure and resilient financial infrastructures, and more.

metaphacts contribution to SMARTY

metaphacts will help develop approaches for continuum intelligence and explainable AI, supporting the seamless and continuous integration of data for learning and real-time insights. We will support the project partners with our technology stack, providing integration and knowledge management software that helps make the data required for the project use cases accessible to the partners and the developed tools via neuro-symbolic AI.

The focus here will be on using explicit knowledge representation based on knowledge graphs to model the continuum and leveraging these models to support data integration and analysis. With their underlying methods of explicit knowledge representation at the symbolic level, knowledge graphs also provide the basis for the development of explanatory components for AI applications.