Background
Many AI systems require significant time, data and computational resources to learn effective behaviours. This creates a major barrier to deploying intelligent infrastructure across different cities, regions and application domains.
Transfer Learning offers the possibility of enabling AI systems to reuse knowledge acquired in one environment and rapidly adapt to new situations.
Research Challenge
Reinforcement Learning (RL) has shown considerable promise for optimising service placement and resource management in edge computing environments. However, training effective RL models often requires large amounts of data, computational resources and exploration time, making it difficult to deploy intelligent decision-making systems across new cities, infrastructures and application domains.
A key challenge is determining how knowledge acquired in one environment can be reused in another. Cities may differ significantly in terms of population density, transportation systems, mobility patterns, infrastructure capabilities and service demands, making it difficult to apply previously learned deployment strategies directly. Developing AI systems that can rapidly adapt to new environments while minimising retraining remains an important open research problem.
This project investigates how Transfer Reinforcement Learning (TRL) can be used to accelerate decision-making in edge computing environments by transferring knowledge between different deployment scenarios. The research will explore challenges such as knowledge representation, policy transfer, domain adaptation, scalability and adaptation to changing environmental conditions.
Example application domains include transferring deployment strategies between different smart cities, adapting AI-driven transportation systems to new urban environments, migrating autonomous service management policies across edge infrastructures, and rapidly deploying AI services during large public events, emergencies or disaster response situations. The work will involve designing and evaluating transfer learning techniques through simulation and experimentation.
Topics
Transfer Learning; Reinforcement Learning; Knowledge reuse; Adaptive AI systems; Smart city infrastructure; Edge computing optimisation
Impact
The ability to transfer knowledge between environments is widely regarded as a key step towards more general and adaptable AI systems. This project contributes to the development of AI that can learn faster, adapt more effectively and scale across diverse real-world settings.