Performance is one of the most important properties in software engineering, yet, it is also difficult to predict as it can be nonlinearly affected by various aspects of the software, including the code, configuration and architecture. As a result, machine learning is one of the most promising ways to model and predict software performance. However, existing machine learning based predictors are essentially `black-box', as it is difficult to explain why and how a model is produced for predicting the performance of a software.
This project seeks to investigate explainable methods for software performance under currently used machine learning paradigms in the field, including but not limited to, (semi-) supervised learning, unsupervised learning, representation learning, deep learning and transfer learning. The software artifacts to consider are vast, such as log file, program code, configurations, components, topology and architecture. The student will evaluate the approach using both experimental and empirical evaluation, under real-world open-source software systems and/or datasets that are widely used. Applicants with strong backgrounds in Software Engineering, Artificial Intelligence, Machine Learning and their intersection are a plus.
Applicants should have, or expect to achieve, at least a 2:1 Honours degree (or equivalent) in Computer Science, Software Engineering, Artificial Intelligence or a related subject. A relevant Master’s degree and/or experience in one or more of the following will be an advantage: software performance engineering, machine learning and data science.
How to apply
All applications should be made online: https://www.lboro.ac.uk/study/postgraduate/apply/research-applications/
Under school/department name, select 'Computer Science'. Please quote reference CO/TC-Un1/2020.
The deadline for applications is 4 May 2020.
Start date: April 2020, July 2020, October 2020, January 2021
Full-time/part-time availability: Full-time (3 years)
Fee band: Band RB (UK/EU: TBC; international: £22,350)
Link to supervisor's online staff profile page: http://taochen.github.io/