Adaptive Systems Using Generative AI
5.2 GENERATIVE AI FOR MULTI AGENT SYSTEMS
Multi-agent system ( MAS ) is a computerized system composed of multiple interacting intelligent agents . An agent is an entity that can perceive its environment and act upon it to achieve some goals . A MAS can solve problems that are beyond the individual capacities or knowledge of each agent , by coordinating and cooperating with each other . For example , a MAS can be used to model a flock of birds , a team of robots , or a network of smart devices .
A MAPE-K 17 model is a framework for designing self-adaptive multi agent systems , which are systems that can monitor their own behavior and environment and adjust themselves accordingly to achieve desired objectives . MAPE-K stands for Monitor , Analyze , Plan , Execute , and Knowledge , which are the main components of a self-adaptive system :
• The Monitor component collects data from the system and its environment .
• The Analyze component processes the data and detects any changes or anomalies .
• The Plan component generates possible actions or strategies to adapt to the changes . The Execute component implements the chosen actions or strategies .
• The Knowledge component stores and updates the information and models that are shared by the other components 18 . In the Generative AI model , we can introduce LLM in the loop 19 . This enables improved reasoning and decision-making skills .
6 DIGITAL TWINS – THE PATH TO SELF-ADAPTIVE SYSTEMS
In this section , we look at how Generative AI integrated with digital twins can enable adaptive systems .
6.1 DIGITAL TWINS WITH GENERATIVE AI
Digital twins are defined by IIC ’ s sister organization , the Digital Twin Consortium 20 , as follows : “ A digital twin is a virtual representation of real-world entities and processes , synchronized at a specified frequency and fidelity .”
While traditionally digital twins have been built towards bringing together structured information in engineering , production , operations , and enterprise information , few digital twins have integrated with 3D models to improve visualization aspects of the digital twins .
17
A Practical Guide to the IBM Autonomic Computing Toolkit
18 https :// www . mdpi . com / 2076-3417 / 12 / 17 / 8583
19 https :// arxiv . org / abs / 2307.06187
20 https :// www . digitaltwinconsortium . org / initiatives / the-definition-of-a-digital-twin / 78
March 2024