Integrating Agent Based Modelling and Machine learning

The integration is driven by two main factors:

(1) A wish to create smarter agents

(2) a problem of calibrating /validating complex models due to a vast variable space.Teaching Agent-Based Modelling and Machine Learning in an integrated way

ABMs use empirical data in many different ways even though the behaviour of the agents may be rule-driven, empirical data can be  used to structure the environments that agents move in hence bringing on board Machine learning.

By integrating ABM and ML, researchers and practitioners can leverage the strengths of both approaches.

• ABM provides a realistic representation of complex systems, capturing individual-level behaviours and interactions. Agent-awareness (observation of changes in environment states), agent-learning and for validation of ABMs. 
• ML techniques enhance the modelling process by leveraging large datasets and learning from patterns in the data. Involving domain and human intelligence, data fusion and preparation, adaptive learning. ABM are a way of creating datasets that otherwise would not be avalaible. 

There are three main methods to apply the integration:

1- Preprocessing of data using ML algorithms

2- Steering agents behaviours using ML

3- Postprocessing outputs from ABMs using ML