Last March, Audensiel launched its new IA Power program. Indeed, Audensiel is positioning itself on artificial intelligence with innovative projects. Amira, PhD at Lab'Innov in Audensiel, talks about the Fog Stream Processing as a Service (FSPaaS) project based on AI. It is a flow processing architecture in the Fog*, an existing infrastructure at the network edge, which collects, processes data and provides real-time decision intelligence.
Can you tell us about the FSPaaS project in a few words ?
The Fog Stream Processing as a Service (FSPaaS) project is a resource orchestration platform that exists at the edge of networks. The best-known term is edge computing. This platform is a layer that exists between the cloud and the user. It is used to host and orchestrate applications on Edge resources. This solution reduces latency, ensures the locality of the data and keeps the connection with the cloud if necessary.
What is the type of AI in this project and how does it work ?
This project is based on reinforcement learning. It is one of the algorithms of artificial intelligence. FSPaaS is also used to enhance the use of AI at the network edge.
Image : Logical architecture FSPaaS
How is AI an added value for the FSPaaS project ?
AI ensures automatic system responsiveness and reinforces architectural best practices: resilience, security, cost and operational concept.
How is AI managed by people within the project ?
In the FSPaaS project, we use reinforcement learning to bring knowledge to automated agents. They interact with the environment. These agents learn from their mistakes and they are rewarded when they correctly perform their tasks that is to say, learning. Initially, humans are an integral part of the equation and contribute to the proper monitoring and effectiveness of these automatic agents.
Today, we are working on a dynamic environment that is infrastructure at the edge of networks. These are variable resources. Initially, we do not have all the parameters of the problem so the AI agent will learn and will be penalized. Therefore, on the other hand, we need the human being to follow the AI agent in time.
Why do you think it’s important to integrate AI into projects ?
AI integration depends on several factors. First of all, the need for the project must be taken into account. We need to know the project constraints and expected results. Indeed, AI requires time and expertise. AI is a means to meet efficiency objectives.
How do you see the evolution of AI in the coming years ?
Today, we are at a major transition point in the history of AI with the emergence of generative AI. Indeed, the latter revolutionizes the market and jobs. This AI is capable of creating the absolute that is to say that it can generate elements that do not exist. The human gives him a basis to create but over time this AI will itself generate new knowledge. As time goes on, humans get out of the generative AI equation. This type of AI is revolutionary. The evolution of AI is moving towards this trend. For example, we can imagine that in 2030, generative AI will generate knowledge that we do not have today. In addition, generative AI will shorten processes, where humans take longer to realize them today.
IA Power, Audensiel’s new AI-based program
This is an innovative approach to enable companies to capitalize on this technology. This support, led by Audensiel’s AI expert team, will help develop this evolution and adapt it to the business uses of companies.
Learn more about our AI Power program? Just here 👇
*Fog : le fog computing consiste à exploiter des applications et des infrastructures de traitement et de stockage de proximité, servant d’intermédiaire entre des objets connectés et une architecture informatique en cloud.
Comentários