School of Computer Science and Statistics
The sudden rise of adversarial examples (i.e., input points intentionally crafted so to trick a model into misprediction) has shown that even state-of-the-art deep learning models can be extremely vulnerable to intelligent attacks. Unfortunately, the fragility of such models makes their deployment in safety-critical real-world applications (e.g., self-driving cars or eHealth) difficult to justify, hence … Read more
Increasingly, patients gather data relevant to their care on their phones and via wearable or home-based sensors. But getting that data into a form that can be shared seamlessly with the healthcare system is a challenge. And clinicians have concerns about the quality of the data where it came from and to what extent they … Read more
I have supervised many projects involving the design and prototype development of apps and am happy to do more. Past examples include a mobile app to support development of patient self-efficacy, a mobile app to support suicide watch teams on the Corrib in Galway, a web application for supporting virtual work team productivity, and a … Read more
Regression verification has been proposed as an alternative to regression testing to combat the drawbacks of the latter. Using regression verification, a verification task is employed to formally prove that an incremental change does not disturb established software behaviour. The benefits of this approach include that, for reasonably confined modifications, verification may be less costly … Read more
Increasingly Knowledge Graphs are underpinning Digital Humanities projects to support the representation and interlinking of data. For example in the recently launched Virtual Record Treasury of Ireland (VRTI), a KG is used to represent people, places, roles that appear in historical documents and their interconnection over the centuries. Typically these KGs are implemented using W3C … Read more
This project will look into parallel algorithms in the OCaml programming language. OCaml will soon release version 5.0 which will support parallelism. Currently OCaml 5.0 is under development but accessible to use as beta. The goal will be to explore the performance and use of parallel algorithms in order to speed up the execution of … Read more
In the world of big data and artificial intelligence, federated learning is a very attractive way of enabling data sharing in a privacy preserving manner. Clients in a federated learning network never access data from other clients directly and they model parameters they receive are formed from data aggregates. However, under certain conditions the models … Read more
Intelligent Personal Assistants (IPAs) such as Apple Siri, Google Home, Amazon Alexa etc playa significant role in our everyday lives by assisting us in the information search and other day today tasks. These systems, however, possess the reactive behaviour which is achieved through arequest-response approach. While this behaviour is still helpful for the users to … Read more
Combine machine learning/Large Language Models and programming language approaches to create a system that detects null pointer exceptions in C programs. There are multiple projects available in this area which range from integrating LLMs in existing verification tools to decide optimal tool paramteres, to employ ML heuristics to search the large statespace in verification tools … Read more
Smartwatches are increasingly used to track a variety of health-related data, but the focus is mainly on automated tracking. This project will investigate different modalities for entering self-tracking information via smartwatch. The project will suit better students taking the CSU44051 Module (Human Factors), or who have previously taken a module in Human-Computer Interaction.