Reinforcement and Machine Learning Models for Coordinating Heterogeneous Hardware and Software Platforms in IoT Environments

Building complex and multi-component systems for the Internet-of-Things requires the materialization of systems requirements over heterogeneous hardware, software and network environments. The orchestration and configuration of a system which prioritizes the requirements specified by the users (for example covering the dimensions of energy efficiency, resiliency and application latency) under the constraints imposed by the available hardware, software and network settings, defines a complex decision problem. Underlying this decision problem resides the intrinsic complexity of interpreting the impact of each design element of the stack into the satisfaction of the user requirement. This complexity is defined by the size of the configuration space entailed by the set of possible design strategies, from the hardware to the application level.

This PhD project explores the use of reinforcement learning, machine learning and semantic representation models as the foundation to approach the multi-criteria optimisation problem underlying finding an optimal configuration for IoT systems operating in heterogeneous settings. The project is intrinsically multi-disciplinary and will cover all the dimensions involved in delivering an efficient end-to-end IoT system (from the chip level through software to the end user). The project will explore contemporary architectures and techniques in machine learning (targeting efficient representations for one/zero-shot learning and transfer learning) and their interplay with reinforcement learning.

Applicants are expected to have:

* An excellent undergraduate degree in Computer Science or Mathematics (or related discipline), and preferably, a relevant M.Sc. degree.
* Confidence and independence in programming complex systems in Java or Python. Industry experience is a plus.
* Previous academic or industry experience in Machine Learning (desired).
* Excellent report writing and presentation skills.

Please note that applicants must additionally satisfy the standard requirements for postgraduate studies at the University of Manchester, such as a first-class or high upper-second class (or an equivalent international qualification) and English language qualifications, as stated in the PGR guidelines.

Qualified applicants are strongly encouraged to informally contact Andre Freitas, ([Email Address Removed]), Christos Kotselidis ([Email Address Removed]) and Sarah Clinch ([Email Address Removed]) to discuss the application prior to applying.