Computer Engineering Commons^™

Optimizing Few-Shot Learning In Pruned Large Language Models With Task-Specific Prompts, Danyal Aftab

Dissertations

Few-shot learning enables large language models to efficiently perform tasks given only a limited number of labeled examples. However, training these models entirely from scratch requires substantial computational resources, making it challenging for many organizations to fully leverage their potential. This thesis explores how structured pruning, task-specific prompting, and parameter-efficient fine-tuning can be combined to preserve few-shot learning capabilities in compressed LLMs, while also extending their utility to real-world recommendation systems.

In this research, we propose the Tailored LLM framework, which first reduces model size through structured pruning and then enhances few-shot learning performance using carefully designed prompts. We experiment …

Monitoring And Short-Term Forecasting Of Atmospheric Air Pollutants Using Deep Neural Networks, Prasanjit Dey

Dissertations

Accurate forecasting of near-surface atmospheric air pollutants such as PM2.5, NO2, SO2, CO, and O3 remains a critical scientific and societal challenge. This difficulty arises from several factors, including nonlinear pollutant dynamics, sparse ground monitoring net works, heterogeneous satellite observations, and strong cross-pollutant interdependencies. Substantial advances have been achieved in temporal deep learning, probabilistic modeling, satellite-based estimation, physics-informed methods, and foundation models. However, these paradigms have largely evolved in isolation. As a result, existing systems are often station-dependent or pollutant-specific and optimized for a single forecasting task. This limits their robustness and generalizability across regions and heterogeneous data regimes.

This …

Active Listening And Reassurance In Text-Based Virtual Health Coaches, Ghulam Hussain

Dissertations

Conversational agents (CAs) have strong potential to support health and physical wellbeing through text-based coaching, but they remain limited in their ability to demonstrate supportive social behaviours that are important in human coaching interactions. In particular, relatively little is known about how Active Listening and Reassurance are perceived, modelled, and evaluated in text-based virtual healthcare coaching, or how such behaviours should be adapted to individual users.

This dissertation investigates how supportive interaction behaviours can enhance text based virtual health coaching, with an initial focus on Active Listening and Reassurance and a later theoretical emphasis on Active Listening. Across five unique …

Virtual Humans In Virtual Reality: A Scoping Review On Sociability, Fidelity, And Expression, J K Sangeeth Chandran, Marisa Llorens Salvador, Cathy Ennis

Articles

Introduction:

Virtual reality (VR) systems have evolved significantly over the past decade, enabling immersive experiences with enhanced realism and interactivity. This has motivated an interest in socially oriented applications. As user proxies, Virtual Humans (VHs) play essential roles in such applications. However, despite technological advancements, achieving realistic, expressive, and socially responsive VHs continues to present design and implementation challenges. In this scoping review, we present the state-of-the-art of VR VHs, examining the impact of VHs on the user experience.

Methodology:

We reviewed 59 papers retrieved from five databases across three core themes: the implementation and impact of VH facial expressions, …

Beyond The Sayable: Wittgenstein, Theory Of Mind And Affective Simulation Of Llms, Haomiaomiao Wang, Lili Zhang, Tomás E. Ward

Women+ in Early Career Research Symposium

Wittgenstein’s distinction between what can be said and what can only be shown frames a limit of propositional language. Affect and aesthetic are not primarily matters of correct description but of how expressions function within shared forms of life. Large language models (LLMs), however, increasingly produce fluent language that resembles such understanding by reproducing the patterns through which people ordinarily talk about emotion and perspective.

This paper argues that apparent Theory of Mind (ToM) in LLMs is understood as competence in the publicly shared patterns of mental-state language, rather than as grounded understanding. Using abstract artworks, we show that LLMs …

Feedback In Digital Game-Based Learning: A Taxonomy And The Design And Empirical Evaluation Of A Feedback System In A Mathematics Serious Game, André Almo

Dissertations

Digital Game-Based Learning (DGBL) is an active, student-centred pedagogical approach in which feedback plays a central role by informing learners’ actions, guiding decision-making and shaping motivation and engagement. Despite its importance, feedback in serious games is often described inconsistently and insufficiently in research, limiting comparability across studies and the accumulation of design knowledge, particularly for children. This thesis addresses these gaps through two complementary contributions: the development of a taxonomy for feedback design in digital serious games and the empirical evaluation of a taxonomy-informed feedback system in a mathematics game for primary school students. First, this work introduces the Taxonomy …

Sharing Patient-Generated Health Data With Electronic Health Record: A Standardised Provenance And Context-Rich Information Model And Clinician Evaluation, Abdullahi Abubakar Kawu

Dissertations

With the advent of Patient-Generated Health Data (PGHD) through wearable, mobile, and home monitoring systems, there is immense potential for ongoing monitoring and patient engagement. But integrating PGHD with Electronic Health Record (EHR) is challenged by sub-optimal support for contextual metadata and its relevant elements, lack of semantic interoperability among disparate systems, poor knowledge regarding the factors that impact clinician acceptance, and absence of globally agreed standards for data exchange. This thesis explores how contextually relevant patient-generated health data can be shared with EHRs through a FAIR standardized information model that ensures semantic and syntactic interoperability.

The study addresses six …

Synthetically Expressive: Evaluating Gesture And Voice For Emotion And Empathy In Vr And 2d Scenarios, Haoyang Du, Kiran Chhatre, Christopher Peters, Brian Keegan, Rachel Mcdonnell, Cathy Ennis

Conference papers

The creation of virtual humans increasingly leverages automated synthesis of speech and gestures, enabling expressive, adaptable agents that effectively engage users. However, the independent development of voice and gesture generation technologies, alongside the growing popularity of virtual reality (VR), presents significant questions about the integration of these signals and their ability to convey emotional detail in immersive environments. In this paper, we evaluate the influence of real and synthetic gestures and speech, alongside varying levels of immersion (VR vs. 2D displays) and emotional contexts (positive, neutral, negative) on user perceptions. We investigate how immersion affects the perceived match between gestures …

Face Off: Evaluating Virtual Human Expressions And Non-Tracking Control Methods In Vr, J K Sangeeth Chandran, Marisa Llorens Salvador, Cathy Ennis

Conference papers

Social virtual reality (VR) applications have become more ubiquitous in recent years; central to this is the communication pipeline, how users perceive virtual human facial expressions, and how they control them in real time, especially when using VR devices without face-tracking. We investigated both aspects in a set of experiments. Firstly, we compared the perception of virtual human emotions on a traditional 2D screen and in VR. In a second experiment, we used a validated set of stimuli to compare three different control methods for manipulating an avatar’s facial expressions in VR. These control methods utilize non-tracking control techniques, which …

Explaining Time Series Classifiers Through Post-Hoc Xai Methods Capturing Temporal Dependencies, Ephrem Tibebe Mekonnen

Conference papers

Time series classification is essential in domains such as healthcare and finance, where accurate predictions can have significant real-world consequences. However, in many high-stakes applications, understanding why a model makes a certain decision is just as important as the prediction itself. While deep learning models excel at capturing complex temporal patterns, their black-box nature limits transparency, making it difficult to trust and interpret their decisions. Although eXplainable AI (XAI) methods have advanced considerably for image and tabular data, applying them to time series remains challenging due to the intricate temporal dependencies and high dimensionality of the data. Post-hoc model-agnostic XAI …

Reinforcement Learning And Virtual Human Animation: A Novel Approach To Data-Driven Animation, Portraying Dynamic, Flexible Human-Like Behaviours, Vihanga Gamage

Dissertations

Virtual characters require animation capable of portraying dynamic, context-sensitive human-like behaviours. Several approaches to generating such animation have been developed, but each carries limitations. Motion capture can produce high-fidelity animation but is expensive and ill-suited to systems that must respond in real time. Physics-based reinforcement learning (RL) enables flexible, dynamic behaviour portrayal, yet relies on simulation feedback signals that are unavailable for social gestures. Supervised approaches can learn social behaviours from motion capture data but yield agents with limited flexibility and generalisation.

This thesis presents RLAnimate, a model-based, data-driven RL framework for character animation that enables a single agent to …

Genwriter: Reducing Gender Cues In Biographies Through Text Rewriting, Shweta Soundararajan, Sarah Jane Delany

Conference papers

Gendered language is the use of words that indicate an individual’s gender. Though useful in certain context, it can reinforce gender stereotypes and introduce bias, particularly in machine learning models used for tasks like occupation classification. When textual content such as biographies contains gender cues, it can influence model predictions, leading to unfair outcomes such as reduced hiring opportunities for women. To address this issue, we propose GenWriter, an approach that integrates Case-Based Reasoning (CBR) with Large Language Models (LLMs) to rewrite biographies in a way that obfuscates gender while preserving semantic content. We evaluate GenWriter by measuring gender bias …

An Empirical Evaluation Of Communication Technologies And Quality Of Delivery Measurement In Networked Microgrids, Ruairí De Fréin, Yasin Emir Kutlu

Articles

Networked microgrids (NMG) are gaining popularity as an example of smartgrids (SG), where power networks are integrated with communication technologies. Communication technologies enable NMGs to be monitored and controlled via communication networks. However, ensuring that communication networks in NMGs satisfy quality of delivery (QoD) metrics such as the round trip time (RTT) of NMG control data is necessary. This paper addresses the communication network types and communication technologies used in NMGs. We present various NMG deployments to demonstrate real-life applicability in different contexts. We develop a real-time NMG testbed using real hardware such as Cisco 4331 Integrated Services Routers (ISR). …

Numerical And Experimental Investigation Of The Dynamic Behavior Of Vibrating Niti Guidewire, Graham Gavin

Conference Papers

Minimally invasive interventions often utilise slender wires to access and navigate to target sites. These guidewires act as rails, facilitating access for other therapies. In endovascular procedures, the guidewires facilitate angioplasty balloons, stenting and other dilation procedures. Due to their small profile and flexibility, endovascular guidewires have limited ability to transfer forces to their tip and as a result cannot penetrate hard calcified blockages, commonly referred to as chronic total occlusions (CTOs) [1-2]. One method proposed to overcome this limitation is the application of vibrations to the wire. The vibrations are transmitted along the length of the wire to the …

Evaluation Of The Crossing Forces Of A Vibrating Guidewire In Contact With Model Calcified Materials, A. Rezaei, Graham Gavin, S. Raut

Conference Papers

Peripheral arterial disease (PAD) affects millions globally, with over 21 million cases reported in the U.S. in 2020. It occurs when leg arteries become narrowed or blocked, leading to complications like critical limb ischemia, where blood flow is severely restricted. Standard treatments, such as bypass surgery and endovascular procedures, aim to restore blood flow. However, advanced chronic total occlusions with calcified plaques, often prevent flexible guidewires crossing into the true lumen of the distal vessel, accounting for approximately 80% of procedural failure. The use of high-frequency (20- 50 kHz) mechanical vibrations (0-50 μm) delivered to the distal tip of specialist …

Securing Iac:Comparing Checkov, Terrascan, And Tfsec On Aws And Azure, Maliha Binte Ruhul Amin, David White

Academic Poster Collection

Securing IaC:Comparing Checkov, Terrascan, and Tfsec on AWS and Azure

Comparison Of Kafka Operators: Strimzi Vs Koperator Vs Confluent: A Comprehensive Industry Analysis, Ankit Anthony, Gary Clynch

Academic Poster Collection

Comparison of Kafka Operators: Strimzi vs Koperator vs Confluent: A Comprehensive Industry Analysis

Analysis Of Impact Of Configuration Choice Upon Azure Service Bus Performance, Brian Skehan, Mary Rose Donnelly

Academic Poster Collection

Analysis of impact of configuration choice upon Azure Service Bus performance.

Comparative Analysis Of Mysql And Mongodb In A High-Concurrency System, Gabriel Solares, Cormac Keogh

Academic Poster Collection

Comparative Analysis of MySQL and MongoDB in a High-Concurrency System

Analysis Of Functional Programming Languages For Use In Serverless Lambda Functions On Aws Platform, William Spain, Gary Clynch

Academic Poster Collection

Analysis of functional programming languages for use in serverless lambda functions on AWS platform

A Comparison Of Terraform And Bicep Quality Attributes, Colm O'Hara, Kevin Bayliss

Academic Poster Collection

A Comparison of Terraform and BICEP Quality Attributes

Comparing Kubernetes And Nomad For Hosting .Net Legacy Workloads In Azure, Luke Osbourne, Omar Portillo

Academic Poster Collection

Comparing Kubernetes and Nomad for hosting .NET legacy workloads in Azure

Evaluating Kubernetes Security Mechanisms For Dos Prevention: A Comparative Analysis Of Multi-Layer Protection Strategies, Sergej Dikun, Omar Portillo

Academic Poster Collection

Evaluating Kubernetes Security Mechanisms for DOS Prevention: A Comparative Analysis of Multi-layer Protection Strategies

Improving Iac Script Quality: Evaluating Static Analysis Tools And Establishing Best Practices, Isha Rai, Kevin Bayliss

Academic Poster Collection

Improving IaC Script Quality: Evaluating Static Analysis Tools and Establishing Best Practices

Comparative Cost And Capacity Analysis Of Managed Service And Non-Managed Service Api Gateway Architectures On Leading Cloud Platforms, George Brown, Cormac Keogh

Academic Poster Collection

Comparative Cost and Capacity Analysis of Managed Service and Non-Managed Service API Gateway Architectures on Leading Cloud Platforms

The Irish Food Manufacturing Industry’S Preparedness For Nis2 Cybersecurity, Niall Mcgowan, Mary Rose Donnelly

Academic Poster Collection

The Irish Food Manufacturing Industry’s Preparedness for NIS2 Cybersecurity

A Comparison Of Aks And K3s On Vmss, Craig Dillon, Omar Portillo

Academic Poster Collection

A Comparison of AKS and K3s on VMSS

Comparison Of Asynchronous Architectural Patterns On Aws & Gcp Using Terraform, Alan Mcgee, Mary Rose Donnelly

Academic Poster Collection

Comparison of Asynchronous Architectural Patterns on AWS & GCP Using Terraform

Examining Terraform And Bicep: A Comparative Analysis Of Infrastructure As Code Tools For Provisioning Azure Environments, Richard Coffey, Kevin Bayliss

Academic Poster Collection

Examining Terraform and Bicep: A Comparative Analysis of Infrastructure as Code Tools for Provisioning Azure Environments

Nomad Vs. Kubernetes, Alexandru Constantin Cardas, Omar Portillo

Academic Poster Collection

Nomad vs. Kubernetes