View Proposal

---

Proposer

Mike Chantler

Title

Automated Qualitative Description of Response (Transfer) Functions or Behaviour Models

Goal

To provide an English language description of the behaviour (using a HeNe laser system as a case study)

Description

We often require a model of the input/output behaviour of physical systems. These are used for monitoring, control and fault diagnosis etc etc. It's common to use a CNN, Transformer or other DNN to obtain either a forward or reverse regression black box model. This project aims to develop a system that can provide qualitative descriptions and dialog that would help engineers understand the basic behaviour of such models. It would identify important variables and provide interactive qualitative and graphical description of their behaviours. Initially simple steady state descriptions (e.g. linear/monotonic/monomodal/multimodal ... but dynamic and discrete systems, variable sensitivity, cross-correlation etc. could also be covered). It would likely need industry user study (which would not just be interface usability - hence the need for straight-forward, but full ethical full approval).

Resources

Laser data set. Laser systems (if more data needed) Potential collaboration with Leonardo for assessment

Background

Url

Difficulty Level

Moderate

Ethical Approval

Full

Number Of Students

1

Supervisor

Mike Chantler

Keywords

machine learning, qualitative explanation, response functions

Degrees

Bachelor of Science in Computer Science
Bachelor of Science in Computer Systems
Bachelor of Science in Information Systems
Bachelor of Science in Computing Science
Bachelor of Engineering in Robotics