UCD Computer Science Success in SFI Frontiers to the Future
Science Foundation Ireland (SFI) recently announced the results of the 2021 Frontiers to the Future Programme. The School of Computer Science received three awards in the ‘projects’ stream and a fourth colleague was a co-applicant in another award. On May 10th, Minister for Further and Higher Education, Research, Innovation and Science, Simon Harris TD, announced 76 grants valued at €53.7 million to support frontiers research across ten Higher Education Institutions in Ireland. UCD won the highest number of awards, and the school of CS won the most awards of any school of Computer Science in Ireland.
UCD Computer Science Awards
Alexey Lastovetsky, Associate Professor
Models Algorithms and Software for Energy-efficient Parallel Computing in Heterogeneous Hybrid Multicore Era
This project will address the ever growing and unsustainable use of global energy by ICT. The share of ICT in the total energy consumption is quickly increasing and expected to exceed 20% of the global electricity demand by 2030. This situation is unsustainable and extraordinary efforts are needed to increase the energy efficiency in ICT. The mainstream approach to energy optimization of computing is to optimize the execution environment rather than applications running in the environment. This project, will develop models, methods, algorithms, and software for optimization of the energy and performance of applications in modern highly heterogeneous and hybrid systems, used in HPC, Internet of Things, Smart Cities, and many other current and emerging digital platforms. These algorithms and methods will optimize applications, not executing platforms, aiming to find all their energy/performance Pareto-optimal configurations.
Colm Ryan, Assistant Professor
Understanding and predicting context-specific synthetic lethality in cancer
A major goal of precision medicine in cancer is to identify more targeted therapies, so that rather than treating patients based on their cancer type (e.g lung cancer), they will be treated based on the mutations present in their tumours. One way to identify these therapeutic targets, is to explore synthetic lethality, where one gene becomes essential in the presence of a mutation in another. Advances in genetic screening technologies, most notably CRISPR gene-editing, have enabled large-scale efforts to identify new synthetic lethal targets. However, testing all human gene pairs for synthetic lethality in even single cell line is well beyond the field’s current capacity. Moreover, it is now clear that synthetic lethal interactions are highly context specific, so that the synthetic lethal interaction - a combination of genetic perturbations – might be lethal in one cell type or genetic background, but well tolerated in another. The overall goal of this proposal is therefore to understand what factors contribute to the context-specificity of synthetic lethality in cancer and to develop computational models capable of predicting context-specific synthetic lethality.
The work will involve collaboration with Dr. David Adams (Wellcome Sanger Institute) and Prof. Chris Lord (Institute of Cancer Research, London).
Anthony Ventresque, Assistant Professor
RobuSTests: Robust Software Tests
Testing is a crucial part of any software development process, but it is difficult to know how effective the tests are. Mutation analysis is a test quality indicator that involves introducing simple artificial defects into programs before checking that the tests can detect them. Previously, mutation analysis focused on simple artificial defects and functional properties, while today’s software systems have become complex and require attention to be paid to non-functional properties (e.g., performance). The RobuSTests project aims to create complex artificial bugs to “test the tests” and make sure they are robust. We will improve the scale and scope of mutation analysis, through the use of complex higher order mutants (HOMs) to improve the quality of software tests. Using sophisticated AI techniques (e.g., search-based techniques) we will aim to minimise and select the HOMs, and to apply them where they are most relevant. We will also leverage HOMs to create better software tests and improve the test management process – again, using various AI technique applied to software engineering.
Abdollah Malekjafarian, UCD Civil Engineering (with CS collaborator, Eleni Mangina, Professor)
Automated and Rapid Fault Diagnosis of Railway Tracks using In-service Train Measurements
This project will develop an interdisciplinary framework for automated and rapid fault diagnosis of railway tracks using acceleration responses measured on in-service trains. Current methods of track monitoring include ‘walking the track’ which is inefficient, and using track recording vehicles, which are expensive. Instead this project will harness data measured on in-service trains, to create a framework for track fault detection and specific feature identification. This framework includes two innovative approaches; (a) an unsupervised machine learning approach which will employ a feature classifier to find the faulty sections of the track using the vertical acceleration data measured on an in-service train, (b) a supervised machine learning approach using the labelled data collected from faults observed in tracks owned by Irish Rail, which will identify the type of the fault where it is already detected.