Project Allocation & Tracking System
Projects from 2025
| Student | Title | Supervisor | Moderator | Module | |
|---|---|---|---|---|---|
| Abbie Griffiths | Algorithms for Generating Robust and Diverse Counterfactuals | Nico Potyka | Hiroyuki Kido | 2425-CM3203 | |
| Much of "AI" today is driven by black-box machine learning models whose inner mechanics cannot really be understood. There are various ideas to explain their decisions, one of them are "counterfactual explanations". The standard motivation for counterfactuals is a loan application scenario. Suppose that a loan application is rejected by a black-box model, and the user wants to understand why. A counterfactual explanation explains the decision by generating an input that is as close to the origin...[more] | |||||
| Abdinasir Hassan | Create a mapping system for adding investigations by children working on the ' Little Investigators' Project in London | Catherine Teehan | Wei Zhou | 2324-CMT404 | |
| This dissertation explores the design and implementation of a web-based mapping system tailored for primary school children in London, as part of the "Little Investigators" project. The system allows young users to identify, annotate, and share local landmarks, enhancing their spatial literacy and community engagement. Utilising open-source technologies like HTML5, CSS3, JavaScript, PHP, MySQL, and Leaflet.js, the platform focuses on creating an intuitive user interface, ensuring robust data sec...[more] | |||||
| Abdullah Alnajfan | Code Quest | Catherine Teehan | Dr Soumya Barathi | 2425-CM3203 | |
| Create a game called Code Quest using either Twine or Minecraft for Education. The game should be designed as an adventure game which includes coding challenges that need to be completed by the players. It should be linked to the new Welsh curriculum for STEM which launches September 2022. The game will need to be supported by education materials such as lesson plans, Schemes of Work and workbooks. This can also include a game that focusses on cyber security.[more] | |||||
| Adam Parsons | OPTIMISING STUDENT WELLBEING AND PERFORMANCE THROUGH THEMINDFUL REVISION APP – GAMIFICATION | Nervo Verdezoto Dias | Nhat (Nick) Pham | 2425-CM3203 | |
| Description: The ultimate goal of the project is to further design, implement and evaluate features for the MindfulRevision app that can help students to improve their academic performance and well-being in more effective and engaging ways. The MindfulRevision App (www.mindfulrevision.com) is a new app that helps students revise for exams more mindfully and effectively. The app provides a variety of features, including: - Tools and techniques to manage stress - Techniques to help students focu...[more] | |||||
| Adriel Nuqui | Developing a Web App to Enhance Learning Efficiency with Large Language Models. | Fernando Alva Manchego | Yulia Cherdantseva | 2425-CM3203 | |
| Learning is often inefficient because students are rarely taught how to learn effectively. Despite the popularity of methods such as rereading, highlighting and summarising, research indicates that these techniques offer limited benefits for memory retention and recall. In contrast, Active Recall, which involves retrieving information from memory through self-testing, is one of the most effective learning strategies. However, currently, there is a lack of platforms that leverage this technique ...[more] | |||||
| Aidan Davey | Extracting information from music to create a playable rhythm game | Crispin Cooper | Theodoros Spyridopoulos | 2425-CM3203 | |
| This project presents the development of a rhythm game that generates levels from musical information extracted from user imported audio. Track tempo is estimated via the use of autocorrelation and resonance, and the two methods are compared for their accuracy. Key challenges faced include accurate onset detection for enemy generation and modification of tempo detection algorithms to improve accuracy. The final implementation presents a playable and responsive game experience, with semi-accurate...[more] | |||||
| Alejandro Arroyo | A System Dynamics investigation to understand the impact of the University's Academic Futures proposals. | Catherine Teehan | Elaine Haigh | 2425-CMT404 | |
| In January 2025, Cardiff University announced their Academic Futures proposal which included the loss of jobs, closures of schools, merging of schools and loss of degree programmes. This investigation using System Dynamics will look at the qualitative data using Causal Loop Diagrams to understand the perceived benefits, investigate the opposing views and identify where data can be quantified and analysed using simulation models.[more] | |||||
| Alexander Brijnath | Augmented Reality IMU Workout Tracker and Display | Dr Soumya Barathi | Hiroyuki Kido | 2425-CM3203 | |
| This report presents the development of a prototype system for automatic workout tracking and augmented reality (AR) feedback. The system is designed to recognise exercise type, count repetitions, and display workout statistics in real time through an AR interface. The aim is to offer a distraction and phone-free gym experience that supports improved mental health and workout performance by eliminating the need for user interaction with smartphones or tracking methods during training. The syste...[more] | |||||
| Alfie Gormley | Teaching Allocation system | Surya Thottam Valappil | Luis Espinosa-Anke | 2425-CM3203 | |
| The School of Computer Science and Informatics has both Teaching Associates and teachers who provide teaching support to our students as part of module teaching teams. Matching the supply of staff to the demand for teaching support is a complex problem with multiple constraints. The skills and knowledge of TAs & teachers must must be considered and matched to the subjects being taught. Time plays a significant role alongside with restriction on allocation, as many teaching support staff are ...[more] | |||||
| Anfisa Phillips | Finding the Optimal Program for a Figure Skater | Richard Booth | Aric Fowler | 2425-CMT400 | |
| In figure skating competitions, the skaters must choose a program to perform before the judges. This project will investigate the problem of finding the optimal program for a figure skater. Aspects to consider include the set of possible elements that can be included, the sequence in which to perform them, and various constraints such as the total duration of the routine, the music that will be played, etc. Ideally the routine should be tailored to the strengths/weaknesses of the skater. We will...[more] | |||||
| Anna Sewell | Benchmarking machine learning techniques for the automatic classification of right hemisphere damage from acoustic speech signals. | Fernando Alva Manchego | Crispin Cooper | 2425-CMT405 | |
| Right hemisphere damage (RHD) is caused by trauma or injury to the right hemisphere of the brain. This project was motivated by a significant gap in research on automatic detection of RHD. Although previous authors have researched this disorder, only one previous study used machine learning (ML) approaches to the classification of RHD from spoken data. This aim of this project was to benchmark a set of machine learning techniques for the automatic classification of RHD from spoken data which ha...[more] | |||||
| Anying Hu | Fee Enforcement Financial Returns Automation and Validation (FEFRAV) - Back-office solution | Richard Booth | Dr Katarzyna Stawarz | 2324Place-CMT403 | |
| This project aims to automate the submission, validation, monthly&quarterly reporting and accounting of fee enforcement financial returns at Liberata. The current manual process is slow, error-prone, and lacks a clear audit trail. The new system is designed to improve efficiency, ensure accuracy, and provide greater transparency through automation and structured data management. The overall project is divided into two main components: a back-end application that handles automated data proces...[more] | |||||
| Aurelie Bourgeais | Measuring comfort for LLM’s and AI avatar’s | Dr Daniel J. Finnegan | Nico Potyka | 2425-CM3203 | |
| As museums continue to adopt a new museology (Vergo, 1989) approach shifting towards inclusivity and social engagement, using technology as a means. Given the rapid advancements and widespread adoption of AI chatbots, museums are beginning to leverage these tools in innovative ways. This research will investigates the effect of human-like characteristics in LLM-powered avatars on user experience. Asking 'Does changing the anthropomorphic characteristics of an LLM affect how comfortable the user...[more] | |||||
| Bartosz Iwaszkiewicz | Desktop app for automatic marking of programming projects | Sylwia Polberg | Dr Katarzyna Stawarz | 2425-CM3203 | |
| The aim of this project is to create a desktop application that would allow teachers to automatically execute and test a series of students' programming submissions in Python and Java. The app will need to include an intuitive and clear interface, ability to rename and organize downloaded students' submissions from Learning Central, and automatically run and test them against files and settings provided by the teacher. This should return an appropriate report that the teacher can later use for f...[more] | |||||
| Benjamin Evans | Detecting IoT Device Events from Network Traffic | George Theodorakopoulos | Andrew Hood | 2425-CMT400 | |
| This project aims to develop heuristics to identify IoT devices from patterns in the wireless traffic that they generate, e.g. length of packets transmitted by the device to the router. With these features identified, the project aims to develop a machine learning program which is capable of receiving a sequence of packets generated via the monitoring of an IoT device (smart plug), and generating an accurate representation of the activities that took place.[more] | |||||
| Benjamin Hacker | Medical image processing - bruises | Paul L Rosin | Hantao Liu | 2425-CM3203 | |
| This project will use a set of images of bruises that have been marked with circles to roughly outline their position. The project aim is to develop automatic detection of bruises using a combination of image processing and machine learning techniques. Here are a few relevant technical papers relating to image-based bruise analysis: Li, Cheng-Ru, et al. "Quantifying the color changes in bruised skin using a color-calibrated imaging system." 2020 IEEE International Symposium on Medical Measurem...[more] | |||||
| Benjamin Iorns | Generating feedback for GCSE English resit students using small language models. | Fernando Alva Manchego | Irena Spasic | 2425-CMT403 | |
| This project assesses the viability of using small language models with a range of prompt-engineering techniques to generate writing feedback which meets the needs of students resitting GCSE English Language. Using established metrics, measure the quality of the feedback. Objectives: - Curate a dataset of GCSE students’ writing annotated with teacher feedback. - Implement feedback generation systems using zero-shot, few-shot and retrieval augmented generation prompt engineering techniques usin...[more] | |||||
| Bo Liu | Image Quality Assessment in Embodied AI Systems | Wei Zhou | Alexia Zoumpoulaki | 2425-CMT400 | |
| As Embodied AI systems, such as robots, autonomous vehicles, and augmented reality agents, become increasingly integrated into real-world applications, their ability to accurately perceive and interpret visual data is crucial. High-quality image perception directly impacts decision-making, navigation, and interaction with the environment. However, image quality in these systems is often subject to diverse challenges, including varying lighting conditions, motion blur, occlusions, and sensor limi...[more] | |||||
| Brandon Williams | AI-Based Urban Traffic Congestion Prediction and Optimization Model | Yipeng Qin | Liam Turner | 2425-CM3203 | |
| Traffic congestion is a major issue in Cardiff, causing delays, pollution, and inefficiencies in transport planning. My project aims to develop an AI-based system to predict and reduce congestion by identifying traffic hotspots using OpenStreetMap data and machine learning. By analysing historical traffic data, the system will suggest better travel routes to ease congestion. This approach involves processing traffic data, applying predictive models, and testing optimization strategies to improve...[more] | |||||
| Charlie Boxall | Generating Images from Audios Using AI | Bailin Deng | Tingting Li | 2425-CM3203 | |
| This project aims to generate a single image for each input acoustic music audio file using deep learning techniques. The student's role will involve: - Creating a dataset of acoustic music audio files to train the AI model on - Experimenting with different audio captioning models to generate text descriptions from the audio - Testing text-to-image generation models to convert the text descriptions into images - Comparing the performance of various model combinations - Constructing an end-to-end...[more] | |||||
| Chen Sun | RoboFin – a robo-advisor/recommender system for financial investment | Dr Daniela Tsaneva | Carolina Fuentes Toro | 2223Place-CMT403 | |
| The digitalization age ha brough a disruption in the financial industry, too. The rise of self-investors has lead to the need of new approach to provide financial advice for investment. There is a request for a new more personalised approach tailored to the specific profile of the investor. RoboFin is a robo-advisor (a recommender system) assisting the investor in the selection of the right investment portfolio by utilising hybrid recommender algorithm returning the results of recommended invest...[more] | |||||
| Chen Wang | Building Lightweight Neural Networks | Yukun Lai | Dr Soumya Barathi | 2425-CMT400 | |
| Deep neural networks have achieved significant success in many areas, including computer vision. However, many successful architectures rely on a huge amount of computing resources. This can restrict such learned models to be deployed to devices with limited computing power, such as IoT devices and mobile phones. The project aims to investigate and compare different techniques for simplifying deep neural networks while keeping the performance as much as possible.[more] | |||||
| Christopher Petrov | Patching LLM Chatbot Vulnerabilities in Web Applications | Andrew Hood | Jandson Santos Ribeiro Santos | 2425-CM3203 | |
| This project aims to explore methods of patching vulnerabilities in LLM chat bot integrations made easy for web-developers, particularly through monitoring and blocking malicious user input and poor LLM output. Investigation includes testing patching techniques across different scenarios (e.g. conversational chats, tool-calling via LLM) and across different models to test flexibility. A combination of fuzzing and external user testing will be used to evaluate the effectiveness of these technique...[more] | |||||
| Colin Kallnik | From Smartphones to Stand-Alone Devices: Developing an Assistive Tool for ADHD Time and Task Management | Dr Katarzyna Stawarz | Shancang Li | 2425-CMT403 | |
| This project focuses on designing and prototyping a dedicated wearable device to support individuals with ADHD in managing time, tasks, and daily responsibilities more effectively. Unlike smartphone apps, which often increase distraction and cognitive overload, the device was built on ADHD-first design principles to minimize overstimulation and maximize accessibility. Inspired by the “Pip-Boy” from the Fallout game series, the prototype combines a Raspberry Pi 4 and Arduino microcontrollers ...[more] | |||||
| Connor Jones | Towards Human-Level Game AI with Deep Reinforcement Learning | Frank C Langbein | Matthew J W Morgan | 2425-CM3203 | |
| We aim to develop reinforcement learning algorithms capable of training game AI bots that exhibit human-like performance, leveraging the same (or at least similar) information available to human players. Given the difficulty of this task, generally lower than human performance is expected. For this project a specific game should be chosen, based on its suitability for reinforcement learning and the availability of necessary resources. A suitable reinforcement learning algorithm has to be selecte...[more] | |||||
| Constantinos Psomadakis | Node-based data structure visualisation tool | Nico Potyka | Wei Zhou | 2425-CM3203 | |
| Binary trees and related data structures are foundational in computer science education but can be challenging and time-consuming for students to create and manipulate. Few tools exist that help students understand, experiment with, and test algorithms covered in lectures, especially with node-based structures like binary trees and graphs. This project involves building a web-based platform that simplifies constructing, visualising, and interacting with these structures. The tool will offer a us...[more] | |||||
| Cunshuo Xia | AI Generation of Music Videos | Bailin Deng | Jing Wu | 2425-CMT400 | |
| Recent advances in AI has made it possible to generate images and videos from text description (see https://stability.ai/blog/stable-diffusion-public-release for example). The aim of this project is to develop a pipeline that automatically generates a music video for songs. The video can consist of a sequence of images or videos segments generated from the lyrics, potentially with the help of other user inputs.[more] | |||||
| Cyrus King | Keep Me Safe App | Carolina Fuentes Toro | Amir Javed | 2425-CM3203 | |
| “Keep Me Safe” is a mobile app designed to enhance personal safety, particularly for women. In today’s society, where personal safety is a pressing concern, this app acts as a virtual guardian. Users can open a query describing their situation, such as going on a date, and include details like the person’s name, location, and time. The app periodically checks on the user and sends alerts if the user fails to respond or declines calls. A yellow alert is triggered initially, and if unaddre...[more] | |||||
| Daniel Sussmann | Simulating Instruments on the GPU | Kirill Sidorov | Fernando Alva Manchego | 2425-CM3203 | |
| Physical modeling synthesis represents an exciting frontier in music production and digital audio. Unlike traditional multisampling-based approaches, which deliver realism and variety at the cost of extensive storage, physics-based modelling offers the potential to recreate the complex sounds of acoustic instruments through computational simulation rather than recording and replaying. This opens up the possibility to generate highly dynamic, realistic and responsive sounds by simulating the phys...[more] | |||||
| David Bowl | SteeringNet, A Deep-Learning based Obstacle Avoidance Approach for Autonomous Driving | Neetesh Saxena | Yulia Cherdantseva | 2425-CM3203 | |
| This dissertation explores SteeringNET, which is a deep learning-based obstacle avoidance approach to autonomous vehicles, with a focus on visualizing and understanding how neural networks learn and evolve without human intervention. By building an Evolutionary Artificial Neural Network (EANN) from scratch in a 2D Unity simulation, this project demonstrates how self-driving agents can develop effective driving behaviours through iterative trial and error. The system bypasses traditional, manuall...[more] | |||||
| Dylan Mingay | Energy-Efficient Code | Frank C Langbein | Bailin Deng | 2425-CM3203 | |
| We wish to investigate the energy consumption of algorithms running on various hardware platforms. By understanding the factors that influence energy usage, we aim to develop techniques to optimize code for improved energy efficiency and the impact on other performance measures such as traditional time, memory and I/O requirements. This is linked to specific hardware platforms and measuring their energy usage, even if the focus is on code and algorithms. The first task is to find a way to estim...[more] | |||||
| Ebrahim Al Aghbari | Desktop app for automatic marking of programming projects | Sylwia Polberg | Carolina Fuentes Toro | 2425-CM3203 | |
| The aim of this project is to create a desktop application that would allow teachers to automatically execute and test a series of students' programming submissions in Python and Java. The app will need to include an intuitive and clear interface, ability to rename and organize downloaded students' submissions from Learning Central, and automatically run and test them against files and settings provided by the teacher. This should return an appropriate report that the teacher can later use for f...[more] | |||||
| Edon You Huang | Photo montage | Paul L Rosin | Irena Spasic | 2425-CMT403 | |
| The aim of this project is to take a set of images and combine them together into a single image. The first step is to roughly paste together the source images. The second step is to fill in any holes using "inpainting". The final step is to refine the image to make it more coherent, and also resize the image if required. The project could take an approach inspired by early papers in this field such as: Rother, Carsten, et al. "Autocollage." ACM transactions on graphics (TOG) 25.3 (2006): 847-8...[more] | |||||
| Elena Corbally | Ethical Hacking Lab | George Theodorakopoulos | Amir Javed | 2425-CMT400 | |
| Technology is advancing rapidly, and as a result the need for cybersecurity education is greater than ever. Penetration testing is an important skill for cybersecurity students to develop, but it can be difficult to gain practical experience in a safe and controlled way. This project addresses the need for accessible and engaging penetration testing resources by providing a vulnerable virtual machine (VM) that contains common Linux misconfigurations, alongside a scenario and challenge questions ...[more] | |||||
| Emile Gaston | Creating an ethical hacking lab | George Theodorakopoulos | Natasha Edwards | 2425Place-CMT400 | |
| In order to practice ethical hacking (penetration testing) it is important to have appropriate infrastructure at your disposal, e.g. computers, routers, configured with intentionally vulnerable software. Normally this is expensive to scale, but Virtual Machines facilitate this by making it possible to create digital equivalents. The aim of this project is to create a digital equivalent of an ethical hacking lab, similar e.g. to the one used in the CMT121 module.[more] | |||||
| Emily Haddrell | A Comparative Analysis of Steganographic Techniques for Embedding Text in Images | Paul L Rosin | Yipeng Qin | 2425-CM3203 | |
| This project compares various image steganography methods, assessing both open-source tools like OpenStego and OpenPuff and custom implementations like LSB, DCT, and HS. Important parameters like robustness, efficiency, capacity, and imperceptibility were evaluated. To facilitate testing, a unique GUI was created. The study highlights trade-offs between the methods investigated and suggests adaptive, context-aware strategies.[more] | |||||
| Ethan Patalano | A System Dynamics Investigation into the impact of the default 20 mph in Wales | Catherine Teehan | Carolina Fuentes Toro | 2425-CM3203 | |
| On September 17 2023, Wales became the first country in the UK to introduce a default speed limit of 20 miles per hour into law. Although the Welsh Government have highlighted the perceived benefits, the introduction of the new default speed limit has met with some strong opposition. This investigation using System Dynamics will look at the qualitative data using Causal Loop Diagrams to understand the perceived benefits, investigate the opposing views and identify where data can be quantified a...[more] | |||||
| Fan Zhou | Hierarchical Ensemble of Open-Source Models with Meta-Learning for Wildlife Classification in Camera Trap Images | Charith Perera | Christopher Wallbridge | 2425-CMT403 | |
| Camera traps are increasingly used for wildlife monitoring in ecology and conservation, but they produce massive image data with challenges like inefficient manual annotation and high false positives (e.g., empty frames from wind or lighting). Single deep learning models, such as MegaDetector (YOLO-based), MLWIC2, DeepFaune, SpeciesNet, and AddaxAI, perform well in specific scenarios but struggle with generalisation across environments. Research shows ensemble methods like voting and averaging i...[more] | |||||
| Farwizah Haji Sufri Khan | OSINT Phishing Tracker | Amir Javed | Yulia Cherdantseva | 2425-CMT400 | |
| This project focused on developing a tool that will use existing flagged phishing/fake websites to extract features, such as domain names, certificates, Ip, location etc. sources : https://www.validin.com/blog/x-phishing-threat-hunting-pivotoing-techniques/ https://inteltechniques.com/ Kicksecure + Tor Tails If you are interested in this project drop me an email confirming your interest and answer below mentioned question 1. What is your interpretation of the problem area? What is the sp...[more] | |||||
| Finn Bradbury | Controlling IoT devices with our brain, eyes, and facial muscles | Nhat (Nick) Pham | Nedjma Ousidhoum | 2425-CM3203 | |
| Rock climbing is as much a mental challenge as a physical one. Climbers face a dangerous paradox: fatigue clouds judgment just as muscles begin to fail, yet traditional safety systems can't detect these invisible tipping points. This research bridges that gap by developing a real-time monitoring system that simultaneously tracks cognitive stress through electroencephalography (EEG) and muscular fatigue via electromyography (EMG). The system employs machine learning algorithms to analyse physiolo...[more] | |||||
| Firas Al Amri | Security Vulnerabilities in IoT Devices: Exploitation and Mitigation approaches | George Theodorakopoulos | Charith Perera | 2425-CM3203 | |
| The project will focus on designing and implementing attacks on IoT devices (e.g., smart lightbulbs, plugs, cameras) or developing defenses to mitigate these attacks. It requires either experience with or a willingness to learn network traffic sniffing and pentesting, Linux commands.[more] | |||||
| Gaurav Malviya | Federated vs Centralized Learning: Performance Trade-Offs | Frank C Langbein | Elaine Haigh | 2425Place-CMT400 | |
| Medical image analysis is rapidly advancing, driven by machine learning and its potential to improve diagnosis, treatment planning, and disease understanding. However, training robust and generalizable AI models in this domain requires access to large, diverse datasets of medical images. Due to stringent patient privacy regulations and data sensitivity, directly sharing medical image data across institutions is often restricted or impractical. This project addresses the critical need for privac...[more] | |||||
| George Lynch | Exploring the impact of Competition and Cooperation on Performance and Engagement in VR Exergaming | Dr Soumya Barathi | Christopher Wallbridge | 2425-CM3203 | |
| This project investigates how competitive and cooperative social dynamics influence user performance, motivation, and engagement in virtual reality exergaming. A bespoke VR cycling application was developed, featuring three distinct cycling scenarios (1) Baseline (solo cycling), (2) Cooperative (cycling with teammates), (3) Competitive (cycling with opponents). Scenarios incorporated adaptive AI and real-time feedback mechanisms to simulate different social environments whilst maintaining consis...[more] | |||||
| Georgin Saju | Gamifying cybersecurity education | Yulia Cherdantseva | Amir Javed | 2425-CMT400 | |
| This project will be focusing of researching different ways of gamification in cybersecurity education. The main objective of the project is to develop a web-application that will facilitate cybersecurity education particularly in the area of network security by providing a game or a set of games that help to learn about concepts such as network protocols, network traffic, network configuration and firewall in an interactive way more appealing for younger audience. You can explore the use and th...[more] | |||||
| Gerard Glynn-Rivers | Injury Prediction in Professional Football: A Machine Learning (ML) Approach. | Annelies Gibson | Surya Thottam Valappil | 2425-CMT403 | |
| Though perceived to be a safe sport, Venturelli et al. (2011) evidence how football has a higher injury rate per hour of exposure than both Rugby and American football (sports considered dangerous), as players are 1000x more likely to suffer injuries playing than workers in the most dangerous industrial professions (Drawer and Fuller 2002). With the impacts resulting from said injuries having wide-ranging consequences like a team’s overall performance being impacted (Hägglund et al. 2013), a ...[more] | |||||
| Gethin Rhys Trigwell | A Comparative Analysis of Vulnerability Scoring Systems for Industrial Control Systems | Neetesh Saxena | Charith Perera | 2425-CM3203 | |
| Develop a new cyber-physical assets security framework to identify and evaluate cyber risks associated with assets including hidden, silent and edge devices.[more] | |||||
| Guanmin Zhang | Design of a Cardiff University Student Part-time Job Information Platform Based on Intelligent Recommendation Algorithm | David J Humphreys | Alexia Zoumpoulaki | 2425-CMT403 | |
| For Cardiff University students, part-time work not only provides them with financial support but also serves as an important means to enhance practical skills and accumulate social experience. However, Cardiff University students generally encounter issues such as fragmented information, low job matching rates, and conflicts between study and work schedules when seeking part-time employment. Existing recruitment platforms lack intelligent services tailored to students' specific needs, and a dat...[more] | |||||
| Hamed Al-Baklani | Explainable Loan Approval System | Alexia Zoumpoulaki | Crispin Cooper | 2425-CM3203 | |
| Using open datasets, develop an (online) application that allows you to load data and train a model that predicts loan approvals based on various input parameters, and more importantly, provides clear and understandable reasons for its decisions. Data Collection: Gather datasets with features like credit score, income, employment history, loan amount, etc. Model Development: Use a ML models like a Decision Tree or Random Forest for its inherent explainability. Explainability: Integr...[more] | |||||
| Hanna Kubica | Developing a personal financial management website | Dr Daniela Tsaneva | Surya Thottam Valappil | 2324Place-CMT403 | |
| Individuals often struggle to effectively manage their personal finances as financial complexity is increasing. Financial disorganisation can lead to stress and clutter with savings and investment. This project seeks to provide users with a user-friendly and accessible web platform that simplifies budgeting, expense tracking, and financial goal setting. Detailed issues: A lack of tools used for tracking expenses and budgeting and limited accessibility in terms of financial planning resources. T...[more] | |||||
| Hansen Han | VR Dental Care - Conquering your dental fears | Annelies Gibson | Elaine Haigh | 2425-CMT404 | |
| 53% of people are scared of the dentist, with many going untreated when a problem occurs. This can often lead to the person seeking medication online to treat their aliment instead of seeking treatment from their dentist. In extreme cases, people with severe anxiety regarding the dentist will attempt to perform a tooth extraction which can lead to complications and infections. The aim of this project is to create a VR dentist game in which the player will be exposed to the sights and sounds of a...[more] | |||||
| Haoran Sun | Personalized News Recommendation System Based on the MIND Dataset | Irena Spasic | Dr Daniel J. Finnegan | 2425-CMT404 | |
| Kaggle data resource link: https://www.kaggle.com/datasets/arashnic/mind-news-dataset Background I used to work as a product manager in an internet company, responsible for the company's user reach system, recommending appropriate messages to users. In our recommendation system, we found that our rule-based recommendation is more effective than algorithmic recommendation after experimentation, that is, at the beginning of the recommendation system, we tried to introduce algorithms, but the effe...[more] | |||||
| Hao Zhang | Image Quality Assessment based on Large Language Models (LLMs) | Wei Zhou | Usashi Chatterjee | 2425-CMT404 | |
| While Large Language Models (LLMs) have experienced significant advancement in visual understanding, their potential to serve as powerful and text-driven models for Image Quality Assessment (IQA) remains largely unexplored. We would like to investigate the IQA task based on LLMs.[more] | |||||
| Harry Franklin | CNI Digital Demonstrator: Showcasing the need to secure the renewable energy sector | Jenny Highfield | Mohammad Taher Pilehvar | 2425-CM3203 | |
| The 14 sectors of Critical National Infrastructure (CNI) in the UK are strongly interwoven and essential to the day-to-day functioning of society. A cyber attack on one sector could have catastrophic effects on not only the incident sector, but it could have cascading effects on downstream sectors. The purpose of this project is to create a digital demonstrator to model the cascading effects of a cyber attack on CNI sectors downstream of the energy sector, specifically, the renewable energy sect...[more] | |||||
| Henrik Vlijter | AI Boxing Technique Improvement Trainer | Dr Daniel J. Finnegan | Bailin Deng | 2425-CM3203 | |
| Boxing is expensive to learn and often inaccessible to people in remote locations. This project aims to bridge the gap between desire and ability to improve performance in boxing, by analysing video data of shadow boxing and subsequently providing feedback with improvements. This is similar to how a coach might teach students but instead the process is virtualised through AI. TLDR: AI Boxing Technique Coach[more] | |||||
| Heyao Li | Development of NPR Anime style Character Rendering Materials in Unreal Engine 5 and Exploration of Quantitative Evaluation Methods | Paul L Rosin | Irena Spasic | 2425-CMT403 | |
| In recent years, non-photorealistic rendering (NPR) has gained significant traction in the gaming industry, with anime-style aesthetics becoming increasingly prominent in commercial titles. However, implementing high-quality NPR remains challenging for small and medium-sized development teams due to limited technical resources and the absence of standardized evaluation methodologies.This project addresses these challenges by developing an artist-friendly NPR base material system for anime-style ...[more] | |||||
| Hin Fung Tsang | Smart Clinical Task Manager: A To-Do List for Clinicians | Dr Daniela Tsaneva | Surya Thottam Valappil | 2425Place-CMT403 | |
| Context: This idea stems from a pitch by a general practitioner during an NHS hackathon I attended. He explained that doctors often suffer from task overload, leading to missed follow-ups, inefficiencies, and burnout. He suggested a smart task manager to not only help him to keep track of tasks, but also automatically assign and prioritise tasks to improve efficiency and quality of care. Problem: Clinicians lack a centralised, tailored task management system to help manage daily tasks, reduce...[more] | |||||
| Isabelle Motley | Social Engineering with Pepper Robot | Christopher Wallbridge | Richard Booth | 2425-CMT403 | |
| Often one of the security measures used by institutions such as banks is to ask you ‘secret questions’ such as ‘What was your first job?’ or ‘What was the name of your first pet?’. Many scams online and on social media attempt to target this information, and awareness has grown. We probably don’t think too much about divulging some of this information in casual conversation with people -it is unlikely that someone we meet will remember all the details from just a casual conversatio...[more] | |||||
| Issah Bartlett | Desktop app for automatic marking of programming projects | Sylwia Polberg | Yipeng Qin | 2425-CM3203 | |
| The aim of this project is to create a desktop application that would allow teachers to automatically execute and test a series of students' programming submissions in Python and Java. The app will need to include an intuitive and clear interface, ability to rename and organize downloaded students' submissions from Learning Central, and automatically run and test them against files and settings provided by the teacher. This should return an appropriate report that the teacher can later use for f...[more] | |||||
| Jack Bell | Collaborative knowledge map platform | Charith Perera | Liam Turner | 2425-CM3203 | |
| Please watch the video in full before contacting the supervisor: https://www.youtube.com/watch?v=GrRsn4Dvpf8 This project invites students to build an interactive, full-stack web application for creating and visualizing knowledge maps. The platform is envisioned to help users organize, link, and explore educational content on various topics—starting with IoT as an example but adaptable to any domain. Inspired by concepts similar to early Khan Academy knowledge maps, this tool will enable user...[more] | |||||
| Jack Thurman | Building Transparent Campaign Intelligence: A Machine Learning Dashboard Using Synthetic Advertising Data | Wei Zhou | Yazmin Ibanez Garcia | 2425-CM3203 | |
| This project explores whether core functions of commercial advertising analytics—such as campaign prediction, performance visualisation, and optimisation—can be replicated using only open-source tools and synthetic data. The result is a fully interactive dashboard that integrates machine learning, interpretability tools, and user-centric design to offer a free and transparent alternative for advertisers and researchers.[more] | |||||
| Jacob Knapp | VNC Security: Analysing the security risks associated with use of the RFB Protocol | Jenny Highfield | Christine Mumford | 2425-CM3203 | |
| Virtual Network Computing (VNC) is a concept used to control remote Graphical User Interfaces (GUIs), and the Remote Framebuffer Protocol (RFB) is commonly used in education and industry for tech support and remote monitoring. There are flaws in the RFB protocol, including security vulnerabilities. This project will look at existing literature to find out what the security vulnerabilities are, with the aim of recreating the exploits and discussing pre-existing solutions.[more] | |||||
| Jac Phillips | Medical image processing - lesions | Paul L Rosin | Jandson Santos Ribeiro Santos | 2425-CM3203 | |
| This project presents the development of a computer-aided diagnosis (CAD) system for the detection of skin cancer in dermatoscopic images, using traditional image processing and machine learning techniques. A significant focus has been placed on detection of the deadliest form of skin cancer, melanoma. Dermatoscopic images have been acquired from two reputable datasets, the HAM10000 and Ganster datasets, enabling both single and cross-dataset evaluation. A robust pre-processing pipeline has been...[more] | |||||
| Jake Palmer | Safeguarding Against AI-Driven Cyber Threats: Mitigating Risks Posed by Large Language Models (LLMs) [INDUSTRY PROJECT] | Eirini S Anthi | Padraig Corcoran | 2425Place-CMT400 | |
| This project explores the dual-use nature of Large Language Models (LLMs) in the context of cybersecurity, focusing on their potential to both enhance security measures and be exploited for cyber-attacks. With the rapid advancement and adoption of LLMs in various applications, there is a growing concern about their misuse, including the generation of sophisticated phishing emails, crafting of malware, or automating social engineering attacks. This research aims to develop strategies and tools to...[more] | |||||
| Jakub Stoklosa | Interactive Knowledge Mapping Platform | Charith Perera | Mohammad Taher Pilehvar | 2425-CM3203 | |
| The Interactive Knowledge Mapping Platform is a web-based application designed to address fundamental challenges in educational content organisation. Traditional educational resources often present information linearly, failing to capture the interconnected nature of knowledge and hindering learners' ability to transfer concepts across domains. Our platform offers a visual, community driven solution to this problem by enabling the creation of knowledge maps, graphical representations connecting ...[more] | |||||
| James Beavis | Controlling IoT devices with our brain, eyes, and facial muscles | Nhat (Nick) Pham | Hantao Liu | 2425-CM3203 | |
| Imagine that someday in the near future, all the smart devices in our homes and offices could be controlled automatically through our thoughts, eye blinks, or just some facial expressions. We will take a small step in this project to achieve this sci-fi vision. We will use state-of-the-art biosensing platforms, i.e. ANT NEURO (https://www.ant-neuro.com/products/eego-sports) and/or OpenBCI (https://openbci.com/), in our newly established Physical-Sensing Lab (https://www.agilecps.org/facilities) ...[more] | |||||
| James Davies | Analysing the efficiency of strength exercises | Fernando Alva Manchego | Andrew Hood | 2425-CM3203 | |
| Implementing a program that can analyse given footage of a strength exercise (Back Squat, Bench Press, Deadlift, etc.) From this the program should be able to derive analytics from the form of the user, calculating the efficiency, targeted muscles and risk of injury. Using this information, the program should provide markers and tips to the user to improve their form.[more] | |||||
| Jana Jhaveri | A Novel Cyber Risk Assessment Tool for SMEs: A Lightweight Practitioner-informed Web Prototype | Yulia Cherdantseva | Amir Javed | 2425-CMT400 | |
| This project will design, build, and evaluate a lightweight, explainable cyber risk assessment method tailored to small and medium-sized enterprises (SMEs). While both quantitative approaches (e.g. numeric estimation of asset value, loss probability, threat frequency) and qualitative approaches (e.g. expert judgement and control checklists) exist, most established methods are too time-consuming, data-hungry, or specialist-led for SMEs with limited time, skills, and budget. The aim here is to cre...[more] | |||||
| Jan Smajdor | Teaching Allocation system | Surya Thottam Valappil | Andrew Hood | 2425-CM3203 | |
| The School of Computer Science and Informatics has both Teaching Associates and teachers who provide teaching support to our students as part of module teaching teams. Matching the supply of staff to the demand for teaching support is a complex problem with multiple constraints. The skills and knowledge of TAs & teachers must must be considered and matched to the subjects being taught. Time plays a significant role alongside with restriction on allocation, as many teaching support staff are ...[more] | |||||
| Jasmine Edwards | A Contextual Analysis of Language in Anti-Women Forums | Nedjma Ousidhoum | Jose Camacho Collados | 2425-CMT400 | |
| The Manosphere is a collection of predominantly male-focused and anti-women forums, which has garnered attention due to its controversial context and often extremist views. Within this community, numerous discussions occur where such language plays a key role. A feature of these communities is the development of their own specific language, creating new terms that are likely unknown to outsiders. However, by looking at the context in which these unknown terms are used, we can gain an understandi...[more] | |||||
| Jay Warren | Explainable Loan Approval System | Alexia Zoumpoulaki | Amir Javed | 2425-CM3203 | |
| Using open datasets, develop an (online) application that allows you to load data and train a model that predicts loan approvals based on various input parameters, and more importantly, provides clear and understandable reasons for its decisions. Data Collection: Gather datasets with features like credit score, income, employment history, loan amount, etc. Model Development: Use a ML models like a Decision Tree or Random Forest for its inherent explainability. Explainability: Integr...[more] | |||||
| Jia-Syun Lyu | Benchmarking World Models for Autonomous Driving: A Comparative Study of Terra and DrivingDojo | Victor Romero Cano | Christopher Wallbridge | 2425-CMT403 | |
| This dissertation dives into a pressing issue in autonomous driving: the lack of reliable ways to test world models, especially when traditional motion planning falls short in unpredictable settings like messy road layouts or sudden obstacles. I’ve been motivated by the absence of consistent benchmarks, which makes it tough to fairly compare different models. To tackle this, I’m proposing a new framework to evaluate two unique approaches: Terra, a deterministic autoregressive model, and Driv...[more] | |||||
| Jiaxun Zong | Game Development with AI, Networking, Graphics, and Physics | Frank C Langbein | Kirill Sidorov | 2425-CMT403 | |
| Focus on physics and collision detection. Developing interactive and engaging games presents a rich and multifaceted challenge in computer science, encompassing artificial intelligence, networking, game environment creation and representation, graphics/rendering, and physics simulation. This project offers a flexible framework for exploring these diverse aspects. Students should choose to focus on specific combinations of these elements to address a technically challenging game development topi...[more] | |||||
| Joel Westcott | Exploring machine learning models in stock market prediction across stock types and market conditions | Yuhua Li | Alexia Zoumpoulaki | 2425-CM3203 | |
| This project aims to explore and compare the effectiveness of machine learning models in stock market prediction across different stock types and under varying market conditions. This exploration is achieved by implementing and comparing the predictive accuracy of four Machine Learning models – SVR, XGBoost, LSTM and RNN – to understand how well they perform across diverse scenarios and stock profiles. These models will then be evaluated using XAI techniques, to hopefully identify the most i...[more] | |||||
| Joseph English | Combining LLMs and Graph Databases for explainable AI | Alia I Abdelmoty | Matthew J W Morgan | 2425-CM3203 | |
| Large Language Models (LLMs) are currently in high demand as they offer opportunities for faster content retrieval, generation, and summarization. Nonetheless, several challenges, such as bias, inaccuracies, and hallucinations, need to be addressed. Additionally, the lack of domain-specific knowledge and the inability to validate and attribute sources hinder widespread adoption. Graph databases are a great tool for representing and manipulating information in the form of knowledge graphs. Combi...[more] | |||||
| Joshua Rimes | Identifying Urban Functional Regions | Padraig Corcoran | Crispin Cooper | 2425-CM3203 | |
| Urban functional regions are urban regions that perform different functions in cities. Examples include tourism regions, bank/financial regions, shopping regions, nighttime economy regions, food regions etc. The goal of this project is to identify the functional regions in a given city in an automated manner using Point of Interest (POI) data. A POI is a specific point location that someone may find useful or interesting. Examples include pubs, shops, gyms, restaurants etc. The POI data will be ...[more] | |||||
| Juan Ayuque Arenas | Creating an activity recommendation app with a swipe-based user interface and an algorithm based off user's interests, location and past selections | Federico Liberatore | Rhodri Morris | 2324Place-CMT403 | |
| Today’s world is filled with immeasurable amounts of information which often leads to people struggle to find suitable activities or events that align with their interests and preferences in their local area. When there’s too many options to chose from, it can generate an overwhelming feeling and at the same time become time-consuming. The proposed solution aims to address this problem by developing a mobile application that recommends activities based on the user’s locations and prefere...[more] | |||||
| Justyn Henman | Smart Workout Tracking | Nervo Verdezoto Dias | Nico Potyka | 2425-CM3203 | |
| This project aims to design and develop a user-friendly workout tracking application tailored to casual gym-goers and individuals with limited time or low self-discipline. Existing fitness apps often overwhelm users with complexity, require manual input, or impose various barriers that discourage long-term engagement. This project focuses on creating a solution that prioritises simplicity, time efficiency, and seamless integration with wearable devices such as the Apple Watch. Building on the i...[more] | |||||
| Kahatapiti Siriwardana | Under the Influence: An Exploration on the Perceptions of Alcohol and Substance Usage in the Youth | Nervo Verdezoto Dias | Omer F Rana | 2425-CM3203 | |
| For my project, I aim to design a mobile application, targeting individuals from the ages of 16-24 (primarily students), to promote healthier habits and raise awareness on the effects of frequent alcohol consumption. I would like the app to combine educational content, self-monitoring tools, and personalized interventions to support users in understanding and reducing their alcohol intake. For the project, I aim to draw from the principles of design thinking - to ensure that the target audience ...[more] | |||||
| Kai Hailstone | Gamification of Digital Abacws: Creation of a digital walkthrough of Abacws using 360 images | Jenny Highfield | Amir Javed | 2425-CM3203 | |
| In association with the Abacws Facilities team. Use 360 images of Abacws (provided by Alex Deverson) which you will be provided with to create a digital walkthrough of Abacws. Gamify the walkthrough so it can be used by potential students with Autism and/ or Anxiety to reduce the stress of starting university in a new environment. Please email both Jenny Highfield (highfieldjm@cardiff.ac.uk) and Alex Deverson (deversonaa@cardiff.ac.uk) together if interested in the project.[more] | |||||
| Kisholi Perera-Merry | A Board Game Companion Application for Diplomacy | Dr Daniel J. Finnegan | Matthew J W Morgan | 2425-CMT403 | |
| Diplomacy is a strategy board game enjoyed for its emphasis on execution of strategy through negotiation with other players. However, its long gameplay means that interruptions are likely to occur, raising cognitive challenges for the players of the game. This project inspects this issue through the critical lens of human computer interaction, game design and cognitive psychology to explore the role of board game companion applications in supporting the prospective memory, motivation, and self-e...[more] | |||||
| Laura Tindley | RoboFin – A robo-advisor to provide goal-based financial recommendations and improve user financial literacy | Dr Daniela Tsaneva | Yuhua Li | 2425Place-CMT403 | |
| The digitalization age ha brough a disruption in the financial industry, too. The rise of self-investors has lead to the need of new approach to provide financial advice for investment. There is a request for a new more personalised approach tailored to the specific profile of the investor. RoboFin is a robo-advisor (a recommender system) assisting the investor in the selection of the right investment portfolio by utilising hybrid recommender algorithm returning the results of recommended invest...[more] | |||||
| Laurence Harley | AI Generation of Music Videos | Bailin Deng | Amir Javed | 2425-CM3203 | |
| Recent advances in AI has made it possible to generate images and videos from text description (see https://stability.ai/blog/stable-diffusion-public-release for example). The aim of this project is to develop a pipeline that automatically generates a music video for songs. The video can consist of a sequence of images or videos segments generated from the lyrics, potentially with the help of other user inputs.[more] | |||||
| Lewis Troy | Personalised Media Recommender | Alexia Zoumpoulaki | Jing Wu | 2425-CM3203 | |
| Recommendation algorithms form an integral aspect of the modern streaming service, with most major services utilising one. However, one area where these services typically lack is explainability, with recommendations usually given with no to little explanation, making the process ambiguous to the user. There is also a lack of control within these systems, with the user unable to define what they would like to see. To address these issues, this project creates a music album recommendation website...[more] | |||||
| Liam Colley | Developing a web app to support dietitians in caring for women with gestational diabetes | Nervo Verdezoto Dias | Crispin Cooper | 2425-CM3203 | |
| Description: This project will develop a web application to support dietitians in caring for women with gestational diabetes. Potential features of the app, as identified in the fieldwork, include: creating meal plans, inputting nutrition recommendations, and personalising the patients’ mobile app. The project will be conducted in collaboration with Beatriz Félix and Francisco Nunes from Fraunhofer AICOS. This thesis occurs in the context of a larger project, NUTRIA, which goal is to deve...[more] | |||||
| Lily Benaid | Using MC Edu to Teach Horse Care & Nutrition | Catherine Teehan | Liam Turner | 2425-CM3203 | |
| In collaboration with the Wales Pony and Cob Society, you will be using Minecraft for Education to develop a series of challenges that will teach young children about horses including care and nutrition. This resource will be designed as learning and teaching resource to support WPCS in their efforts to education children about horses.[more] | |||||
| Linglin Yu | Predicting habitat type from archived biological records | Chris B Jones | Oktay Karakus | 2324Place-CMT404 | |
| There are millions of records of biological samples in museums and herbaria, many of which include a natural language description of the habitat where the sample was collected. The data have the potential to be used to reconstruct historical habitats. The project will experiment with using machine learning methods to predict habitat type based on such text data. Classifiers will be trained using ground truth data on habitat in the form of digital land cover maps from recent decades, with a view ...[more] | |||||
| Luca Selano Robson | Minecraft Edu Escape Rooms to Teach Cyber Security | Catherine Teehan | Padraig Corcoran | 2425-CM3203 | |
| Using Minecraft for Education to develop a series of escape rooms that challenge players using cyber security techniques to help develop cyber security knowledge and skills.[more] | |||||
| Lu Jing | Deconstructing Gambling Through Games: A Risk Education Framework Based on Coin Pushers | Dr Daniel J. Finnegan | Liam Turner | 2425-CMT403 | |
| This literature review evaluates evidence on: (1) how coin-pusher (penny-falls) machines operate and the risks they pose, (2) common probabilistic misconceptions related to gambling, (3) the effectiveness of game-based and simulation-based interventions for teaching probability and gambling awareness. The review finds converging evidence that interactive, gamified simulators can improve statistical reasoning and increase risk awareness, while also highlighting a clear gap: coin-pusher machines ...[more] | |||||
| Lukasz Mikolajczak | Tri-Metric Framework for Evaluation of Robustness for Steganography Methods | Paul L Rosin | Carolina Fuentes Toro | 2425-CM3203 | |
| This project proposes a structured evaluation framework for assessing the robustness of image steganography algorithms under real-world distortions. It introduces three complementary metrics—TRI, CSRI, and IWRI—and applies them to four steganographic methods to demonstrate their effectiveness in capturing transformation resilience, capacity scaling, and imperceptibility trade-offs.[more] | |||||
| Luke Pope | System for Novel Extended View Planner and Live 3D Object Reconstruction Using Franka Emika Manipulator | Juan Hernandez Vega | Neetesh Saxena | 2425-CM3203 | |
| 3D Object reconstruction (in this scope) is the process of representing an object’s appearance using a 2D image. This is achieved by analysing the object’s shape to form a coordinate system representing all points within the object’s frame. Reconstruction information can be applied to various fields of work, such as medical imaging, graphics, virtual reality, etc. Robot devices are often used in conjunction with reconstruction to develop automated and seamless methods of defining an objec...[more] | |||||
| Maame Dartey | Cyber Physical System Education: Gamification of the teaching of Industrial Control System Security | Jenny Highfield | Padraig Corcoran | 2425-CM3203 | |
| You will be given teaching content for ICS Security and asked to gamify it (open to options, but could be CLI game, Webform or a 2D/3D game). The project will look at the benefits of gamification in education and how it can be applied to learning about Industrial Control Systems and their security risks.[more] | |||||
| Manuel Miranda Mendes Da Luz | Artificial Intelligence for Games [multiple availability] | Federico Liberatore | Padraig Corcoran | 2425-CM3203 | |
| Dream of creating a game-winning AI? This is your chance to bring it to life! Challenge: Choose any game you love, design your own AI agent, and compete for victory. Pick from top competitions listed by the IEEE Conference on Games (https://2024.ieee-cog.org/competitions/) or propose a game of your own! No need to officially enter—just focus on building the ultimate AI. Requirements: •Solid programming skills •Background in AI, optimization, or machine learning •A competitive spirit an...[more] | |||||
| Max Emmans | Digital Transformation | Catherine Teehan | Paddy Slator | 2324Place-CMT403 | |
| Working with an organisation to map current business processes using a variety of business situational awareness methodologies with the aim to identify where Business Process Automation would fit and develop an automation strategy with potential to implement the identified BPAs.[more] | |||||
| Meabh Carragher | Developing a mobile application to support women with gestational diabetes | Nervo Verdezoto Dias | Rhodri Morris | 2425-CM3203 | |
| Description: This project will develop a mobile application to support the self-care and learning of women with gestational diabetes. Potential features of the app, as identified in the fieldwork, include: meal reminders, glucose measurement reminders, meal plan display, meal/recipe recommendations (algorithm under development in NUTRIA), and nutrition learning modules. The project will be conducted in collaboration with Beatriz Félix and Francisco Nunes from Fraunhofer AICOS. This thesis o...[more] | |||||
| Mihai Craciun | Relevance-focused methods for reasoning with Conditional Knowledge Bases | Richard Booth | Víctor Gutiérrez Basulto | 2425-CMT400 | |
| A conditional knowledge base (KB) is a set of statements of the form "if A, then normally/usually B". A classic question in Knowledge Representation has been "given a conditional knowledge base KB, which other conditionals should be said to "follow from" KB. Several solutions have been proposed over the years, including the preferential closure, the rational closure and, more recently, the disjunctive rational closure (DRC). The DRC is especially applicable in situations where we want our result...[more] | |||||
| Mittul Rungta | Anomaly Detection in Environmental Sensor Data Using Machine Learning | Alia I Abdelmoty | Federico Liberatore | 2425-CM3203 | |
| Anomaly detection in environmental sensor networks is essential for ensuring system reliability, occupant comfort, and energy efficiency in smart buildings. This study implements and evaluates four unsupervised machine-learning methods: Isolation Forest, One-Class Support Vector Machine (OCSVM), Long Short-Term Memory (LSTM) Autoencoder, and TimeGPT transformer applied to univariate and multivariate datasets. A six-month real-world dataset from the University of Bristol is split into a four-mont...[more] | |||||
| Mohammed Ahmed | AI Generated Image Detection | Wei Zhou | Irena Spasic | 2425-CM3203 | |
| Nowadays, there emerged lots of AI generated images that are different from natural images. This project aims to develop methods to distinguish AI generated images from natural images. It would be better to send me your CV when contacted.[more] | |||||
| Moriah Tsekiri | Data-driven Fashion Trend Forecaster | Federico Liberatore | Nedjma Ousidhoum | 2425-CM3203 | |
| This aims to be a dashboard that predicts trends in fashion, based on previous garments that have been worn and photographed on runways, as well as similar pieces seen on other lower brands (like fast fashion brands) and similar items seen on social media. The problem with the fashion industry today is the overproduction and waste of materials through fast fashion, failed sample garments, etc. causing environmental damage in our society. This forecasting tool could help designers better predict...[more] | |||||
| Muhammad Saim Javed | Deepfake detection using a GAN Fingerprints | Shancang Li | Wei Zhou | 2425-CMT400 | |
| Generative Adversarial Networks (GANs) have been increasingly used in generating deepfacks, which are believed one of the biggest AI concerns. This project aims to look into the AI engine, GANs, to develope approaches to analyse GAN fingerprints in terms of its existence, uniqueness, persistence, and superiority. This project needs strong AI background and python programming.[more] | |||||
| Myrsini Apostoloudi | Efficient Anomaly Detection Based on Brain Inspired Emerging Computing Framework | Yuhua Li | Jandson Santos Ribeiro Santos | 2425-CM3203 | |
| Modern machine learning methods including deep learning have shown super-human performance in some applications. However, they face several challenges in algorithm and data limitations. In the last few years, Hyperdimensional Computing (HDC) has emerged as a promising paradigm to tackle the challenges faced by existing machine learning and deep learning. It is a brain-inspired emerging computing paradigm for representing and manipulating concepts and their meanings using fixed-size vector repres...[more] | |||||
| Naeem Ali | Natural Language Processing (NLP) for Education - Generation of quizzes to assess reading comprehension | Fernando Alva Manchego | Padraig Corcoran | 2425Place-CMT400 | |
| This project is about using Natural Language Processing (NLP) to automatically create quizzes that check reading comprehension. The idea is to generate both questions and answers directly from a passage of text. To make the models efficient, we use Low-Rank Adaptation (LoRA) and Quantized LoRA (QLoRA). These techniques reduce the memory and computing power needed while keeping good performance. We fine-tuned and tested three models: FLAN-T5-XL, LLaMA-3.2-3B-Instruct, and Phi-2. Training was don...[more] | |||||
| Nathan Blundell | CNI 2D Digital Demonstrator: Showcasing why we need to secure industrial systems | Jenny Highfield | Wei Zhou | 2425-CM3203 | |
| The use of demonstrators within cybersecurity allows us to showcase, without technicality, the impact of cyberattacks on Critical National Infrastructures (CNI). Cyber Innovation Hub have a large physical demonstrator for this purpose, however, this project is to create a digital version. This can include CNI such as can be found in the transport sector, civil nuclear sector and manufacturing sector, among others. You will have artistic licence, and will be working with an industry partner to ge...[more] | |||||
| Ned Kingdon | Chess AI that explains its decisions | Kirill Sidorov | George Theodorakopoulos | 2425-CM3203 | |
| The way in which artificial intelligence systems make decisions is typically drastically different from the approach a human takes. Thus, the operation of an AI system may be difficult to interpret and to learn from. While the search algorithms in traditional chess engines are easy for a human to understand (algorithmic transparency), they lack simulatability, i.e. a human cannot reproduce calculations necessary for the result, and thus this knowledge does not lead to a better understanding of w...[more] | |||||
| Nicolas Gutierrez Rolon | Unsupervised Classification and Explanation Generation of Mutation Triggers in Welsh | Fernando Alva Manchego | Alia I Abdelmoty | 2425-CMT405 | |
| This project focuses on developing a rule-based mutation explainer for Welsh, aimed at supporting language learners and advancing Welsh NLP. Initial consonant mutation is a distinctive feature of Welsh grammar, where the first consonant of a word changes depending on its grammatical environment. While existing tools such as CyTag can detect mutated forms, they cannot explain why a mutation occurs, which limits their usefulness for learners and educators. To address this gap, the system integrate...[more] | |||||
| Niranjan Uppini | Enhancing Large Language Model Performance on Tabular Question Answering through Supervised Fine-Tuning | Jose Camacho Collados | Elaine Haigh | 2425-CMT400 | |
| Large Language Models exhibit strong text generation capabilities but demonstrate poor performance on tabular data analysis, frequently producing incorrect calculations despite generating syntactically valid code. This research addresses this limitation through a systematic approach combining automated data curation, programmatic verification, and parameter-efficient fine-tuning. The methodology centers on creating a verified training dataset where every example undergoes computational validatio...[more] | |||||
| Nurseiit Bakkali | Natural Language Processing: Evaluating LLMs in languages other than English | Mohammad Taher Pilehvar | Jing Wu | 2425-CMT400 | |
| [The details need to be discussed] For students who are fluent in languages other than English, the goal of this project is to assess the capabilities and weaknesses of LLMs (such as GPT, Gemini, or Llama) on language-specific tasks. This could either target language proficiency or specific cultural concepts.[more] | |||||
| Oliver Turk | Algorithms for Interpretable Machine Learning with Quantitative Argumentation Frameworks | Nico Potyka | Jing Wu | 2425-CM3203 | |
| Interpretable Machine Learning aims at designing machine learning models that are human-interpretable [A]. While classical approaches like decision trees, rules or regression models satisfy this criterion, their learning performance is often inferior to black-box models like ensemble classifiers or neural networks. Recently, it has been shown that there is a close relationship between neural networks and quantitative argumentation frameworks [B]. This connection can be used to combine parameter ...[more] | |||||
| Oliwia Stajuda | Efficiency - Accuracy Balance in 2D Collision Detection: A Comparative Analysis of Grid-Based and Border-Only Algorithms | Frank C Langbein | Federico Liberatore | 2425-CMT403 | |
| One of the fundamental challenges faced by 2D game developers is maintaining an accuracy - efficiency balance in collision detection. Traditional approaches require form them to choose between complex polygon and pixel-perfect methods that provide accuracy but are computationally expensive, and simple geometric shapes, which are fast but less precise. This trade-off significantly impacts modern games with numerous interactive elements and detailed artwork, especially when they are intended for v...[more] | |||||
| Omar Ahmed | Optimising Student Wellbeing and Performance through the Mindful Revision App | Nervo Verdezoto Dias | Wei Zhou | 2425-CM3203 | |
| Description: The ultimate goal of the project is to further design, implement and evaluate features for the MindfulRevision app that can help students to improve their academic performance and well-being in more effective and engaging ways. The MindfulRevision App (www.mindfulrevision.com) is a new app that helps students revise for exams more mindfully and effectively. The app provides a variety of features, including: - Tools and techniques to manage stress - Techniques to help students focu...[more] | |||||
| Omar Ibrahim | Detecting Media Bias in Coverage of the Israeli war on Gaza: A Transformer-Based Evaluation of Gold and Synthetic FIGNEWS 2024 Dataset | Mohammad Taher Pilehvar | Martin Caminada | 2425-CMT405 | |
| Amid ongoing developments in the Middle East, particularly in Gaza, media coverage and online discussions have reflected a wide range of perspectives on the conflict. Following October 7th, 2023, reporting and commentary on Gaza became a subject of significant debate. To examine this discourse from a linguistic perspective, especially on social media platforms such as Facebook, a group of authors (Zaghouani et al., 2024) introduced the FIGNEWS2024 dataset, which focuses on media bias in the ea...[more] | |||||
| Oscar Hoodsmith | Rendering Engine for Voxel-Based Real-Time Multiresolution Terrain Generation | Frank C Langbein | Padraig Corcoran | 2425-CM3203 | |
| For this project, you should choose an arbitrary game that is played in a 2D or 3D environment with one or more players. The aim of this project is to implement a game engine with 2D or 3D graphics and suitable dynamics (physics; collision detection) simulations to run this game and networking capabilities. You are free to focus on any one or any combination of these elements, but the focus of this project is the engine and not a particular game. The focus could be specifically on 2D or 3D envi...[more] | |||||
| Patryk Gluszkiewicz | Localisation using Wi-Fi emanations | Kirill Sidorov | Mohammad Taher Pilehvar | 2425-CM3203 | |
| Even if not connected to a WiFi network (but with WiFi enabled), your phone periodically emits beacon packets and otherwise scans for available networks. Such packets, among other things, contain information about the strength of the signal (RSSI). Having positioned several base stations, listening for such packets, one may be able to (crudely) localise people by trilateration (or multilateration). This project will implement such a localisation system, and investigate its limitations and possib...[more] | |||||
| Peixuan Yu | Using Twine to Create an Interactive Adventure Story about the Welsh Dragon to Teach the new Welsh Cross-Curriculum Competency Frameworks | Catherine Teehan | Annelies Gibson | 2425-CMT404 | |
| Create an interactive narrative using Twine that includes activities challenges that need to be completed by the players. It should be linked to the new Welsh curriculum and cover the competency frameworks (Digital, Literacy and Numeracy) which launched in September 2022. The interactive story will need to be supported by education materials such as lesson plans, Schemes of Work and workbooks.[more] | |||||
| Peng Xu | Underwater Image Enhancement | Wei Zhou | Yukun Lai | 2425-CMT403 | |
| Underwater images usually have problematic issues, such as color casts, color artifacts and blurred details, etc. This project aims to develop methods to enhance the visual quality of underwater images. It would be better to send me your CV when contacted.[more] | |||||
| Phuong Thanh Le | Emotion-Driven Interaction: Designing an AI-Assisted Music Application for Active Emotional Engagement | Alia I Abdelmoty | Eirini S Anthi | 2425-CMT403 | |
| There is a gap in current AI music research, which often isolates generative models from real-time human input or emotional context. This project proposes the design of an interactive AI-Assisted Music Application and its effect on emotional engagement. The application combines AI music generation models with other tools that allows users to personalise the music outputs in real time. These elements are tempo, rhythm, choices of instruments and others. The goal is to explore how humans and AI ca...[more] | |||||
| Qiang Fu | Build an abstractive summarizer based on PLM and Fine-Tuning | Nedjma Ousidhoum | Andrew Hood | 2425-CMT400 | |
| In today’s Internet world, we read a lot of information every day, and it is not easy to quickly understand the main ideas of long articles or papers. This project aims to create a tool that can automatically produce short, clear summaries in natural language, so people can save time and learn more efficiently. The tool will be built using advanced language technology that learns from large amounts of text.[more] | |||||
| Rakshit Rangarajan | AI-driven Personalised Welsh Language Learning Platform | Paddy Slator | Nervo Verdezoto Dias | 2425-CMT400 | |
| This project aims to develop a web-based platform inspired by Say Something in Welsh, designed to provide an interactive and personalised way to learn Welsh. Users will begin by providing a word or phrase of interest, and the platform will use a large language model (LLM) to generate a set of relevant Welsh phrases or sentences that incorporate the input. The platform will first introduce these phrases in both Welsh and English, with accompanying audio to help users understand pronunciation and ...[more] | |||||
| Rami Azizi | hacking human vulnerability to phishing | Amir Javed | Carolina Fuentes Toro | 2425-CM3203 | |
| Phishing poses a significant and persistent threat in the realm of cybersecurity, targeting individuals, businesses, and organisations alike. This insidious tactic involves cybercriminals masquerading as trustworthy entities, such as banks, government agencies, or reputable companies, to deceive unsuspecting victims into divulging sensitive information, such as login credentials, financial details, or personal data. The sophistication of phishing attacks continues to evolve, with perpetrators em...[more] | |||||
| Rana Cakmak | Intrusion Detection in SCADA Systems Using AI Techniques | Eirini S Anthi | Nhat (Nick) Pham | 2425-CMT400 | |
| This project focuses on detecting cyber attacks in SCADA systems by applying AI techniques. I will analyze the effectiveness of IDS in SCADA networks using the dataset provided in the paper: https://orca.cardiff.ac.uk/id/eprint/137897/[more] | |||||
| Raveena Kesavan | Explainable Loan Approval System | Alexia Zoumpoulaki | Tingting Li | 2425-CMT400 | |
| Using open datasets, develop an (online) application that allows you to load data and train a model that predicts loan approvals based on various input parameters, and more importantly, provides clear and understandable reasons for its decisions. The project will go beyond simple ml algorithms. Different areas of focus will be considered: e.g. interactive explainability, dataset augmentation, simplification, chatbots for explanations. Dataset identification, Model Development, Explainabili...[more] | |||||
| Rebekah Pidgeon | Parameter exploration of a Genetic Algorithm for evolving Hearthstone decks | Federico Liberatore | Christopher Wallbridge | 2425-CM3203 | |
| Genetic Algorithms have been used as an Artificial Intelligence strategy for many games, including competitive, collectible card games. Hearthstone is a particularly popular online competitive, collectible card game, and poses many interesting Artificial Intelligence challenges such as deckbuilding and game-playing agents. While Genetic Algorithms have been used previously to build competitive Hearthstone decks, the many parameters that are required to be set for a Genetic Algorithm have not bee...[more] | |||||
| Reiyan Al-Shaikh | Unreal Engine Escape Rooms to Teach Cyber Security | Catherine Teehan | Tingting Li | 2425-CM3203 | |
| This project explores the development of the educational game Project ZeroPoint, created using Unreal Engine, as a teaching and learning resource for cybersecurity. The game introduces learners to fundamental concepts within cybersecurity through interactive challenges, promoting critical thinking and problem-solving in a virtual environment. This project aims to align with the information and guidance provided by the new Welsh Curriculum and Digital Competence Framework. The report will detail...[more] | |||||
| Rene Campbell | Evaluating Social Bias in Large Language Models across different multilingual settings | Walter Colombo | Yazmin Ibanez Garcia | 2425-CM3203 | |
| This study examines bias in question-answering (QA) tasks in two recently developed large language models (LLMs), Gemini Flash and Llama-3.3. A foundational English (EN) dataset is constructed across seven social dimensions by intersecting existing bias focused benchmarks: age, disability status, gender, physical appearance, religion, socioeconomic status and sexual orientation. The dataset is translated into German (DE) and Japanese (JA), where these languages are used to test the LLMs. This ap...[more] | |||||
| Ricardo Garcia | MusicMerge - Collaborative music writing | Kirill Sidorov | Eirini S Anthi | 2425-CM3203 | |
| Collaborative music composition presents unique challenges in version control, requiring specialized tools capable of detecting and merging differences while preserving musical semantics. This project ad- dresses the limitations of existing notation comparison tools by developing a Python-based MusicXML diff and merge application. The tool integrates a weighted difference algorithm, analyzing pitch, duration, articulation, and stem direction at the note, measure, and part levels to provide intui...[more] | |||||
| Rohin Rashid | Optimising Student Wellbeing and Performance through the Mindful Revision App | Nervo Verdezoto Dias | Yipeng Qin | 2425-CM3203 | |
| Description: The ultimate goal of the project is to further design, implement and evaluate features for the MindfulRevision app that can help students to improve their academic performance and well-being in more effective and engaging ways. The MindfulRevision App (www.mindfulrevision.com) is a new app that helps students revise for exams more mindfully and effectively. The app provides a variety of features, including: - Tools and techniques to manage stress - Techniques to help students focu...[more] | |||||
| Romilly Nash | Development of a Humanoid Robot System for Enhancing Social Synchrony in Children with Autism Spectrum Disorder | Christopher Wallbridge | Andrew Hood | 2425-CM3203 | |
| One of our projects is looking at social synchrony in autistic children with a robot. Social synchrony is the mechanism by which we keep time with each other and may be an important skill in the process of conversational turn taking. We want the robot to follow along with simple gestures such as clapping in time or doing a synchronised wave. Potential students will have to look at potential methods for identifying a pose non-intrusively (e.g. using a 3D camera) and then mimicking it (preferably)...[more] | |||||
| Rupert Davies | MRI Superresolution | Frank C Langbein | Oktay Karakus | 2425-CM3203 | |
| Magnetic Resonance Imaging (MRI) is a vital medical imaging technique. However, limitations in hardware and acquisition time often result in lower-resolution images (e.g. at 1.5T or with specific fast scanning protocols). This project aims to develop a superresolution algorithm capable of constructing higher-resolution MRI images (e.g. 3T or the full resolution from a fast scan) from lower resolution ones. By simulating lower resolution images from high resolution images using suitable k-space ...[more] | |||||
| Ryan Gallagher | EyeTrack: How can a user-friendly optometry app enhance awareness, accessibility, and proactive management of eye health among users? | Liam Turner | Yulia Cherdantseva | 2425-CM3203 | |
| Design for Eye Care Apps focuses on creating digital solutions that enhance the user experience in managing eye health and vision care. The problem theme centres on the insufficient awareness and accessibility surrounding eye health management among individuals. Many people do not prioritize regular eye care, often neglecting the importance of monitoring their vision and understanding common eye conditions. My app will store user data securely, implement strong encryption methods, and adhere to...[more] | |||||
| Sabari Girish Srinivasan | ”Wards and Firewalls”: Enhancing Cybersecurity Awareness Through Gamification for the Digital Generation | Yulia Cherdantseva | Amir Javed | 2425-CMT400 | |
| This project focuses on the creation of a serious cybersecurity game to spread awareness regarding basic cybersecurity concepts. With the rising frequency of cyber attacks and their increasing lethality to organisations, cybersecurity is now a joint responsibility where every employee in an organisation plays a crucial role and thus needs to be vigilant. Traditional forms of cybersecurity training often fall short on engaging non-technical participants (who often perceive cybersecurity as an ext...[more] | |||||
| Saikiran Rangarajan | Deepfake detection using a GAN Fingerprints | Shancang Li | Tingting Li | 2425-CMT400 | |
| Generative Adversarial Networks (GANs) have been increasingly used in generating deepfacks, which are believed one of the biggest AI concerns. This project aims to look into the AI engine, GANs, to develope approaches to analyse GAN fingerprints in terms of its existence, uniqueness, persistence, and superiority. This project needs strong AI background and python programming.[more] | |||||
| Samuel Ashman | Flipper Zero: Evaluating the Cybersecurity Risks of Universal Hacking Tools | Eirini S Anthi | Neetesh Saxena | 2425-CMT400 | |
| This project investigates the security implications of Flipper Zero, a portable multi-tool for pentesting, by assessing its capabilities in intercepting and manipulating wireless communications (RFID, NFC, Bluetooth). The study will explore defensive measures to mitigate unauthorised access risks in consumer and industrial IoT environments. A flipper zero will be provided to the student.[more] | |||||
| Samuel Berry | Evaluating Deep Audio Embeddings as Fitness Function for Genetic Music Generation | Crispin Cooper | Carolina Fuentes Toro | 2425-CM3203 | |
| This study evaluates the effectiveness of deep audio embeddings as a fitness function for evolutionary music generation, specifically within the domain of ambient music. Using deep audio embedding similarity as the sole fitness metric, I implemented a graph-based genetic algorithm that evolves probabilistic MIDI generators to produce audio similar to a reference track. The research addresses whether such embeddings can guide evolutionary search without relying on explicit music-theoretic rules o...[more] | |||||
| Samuel Larkins | Considering Social Groups in Social Robot Navigation | Juan Hernandez Vega | Kirill Sidorov | 2425-CM3203 | |
| Human-robot interaction (HRI) and human-robot collaboration (HRC) are relevant areas of robotics that explore the necessary requirements that allow robots to work alongside humans. One key aspect of such collaboration is stablishing an efficient transmission of information between robots and humans. A recent and novel approach to this challenge involves the use of a social heatmap. This project aims to extend the capabilities of a social heatmap so that it is also able to display high cost are...[more] | |||||
| Sashwath Ravilla Seenivasan | Cafe Location Analysis: A Time Series Forecasting Approach to Spatial Business Intelligence | Padraig Corcoran | Nhat (Nick) Pham | 2425-CMT400 | |
| Urban functional regions are urban regions that perform different functions in cities. Examples include tourism regions, bank/financial regions, shopping regions, nighttime economy regions, food regions etc. The goal of this project is to identify the functional regions in a given city in an automated manner using Point of Interest (POI) data. A POI is a specific point location that someone may find useful or interesting. Examples include pubs, shops, gyms, restaurants etc. The POI data will be ...[more] | |||||
| Selmane Tabet | Project Mjölnir: An Automated Brute-Force Dataset-Model Combinatorics Training and Evaluation Pipeline for Computer Vision | Charith Perera | Matthew J W Morgan | 2223Place-CMT404 | |
| The project aims to experiment with a technique inspired by Brute-force Attack methods from the cybersecurity field, which can be applied in the context of datasets and training for binary image classification. This project implements an AutoML pipeline that performs training and evaluation on an exhaustive set of model-dataset pairings to determine the best performing ones, drawing comparisons to published literature and traditional approaches to ML development. The primary goal of this projec...[more] | |||||
| Semih Sarisoy | Project Title: Emotion Detection for Social Media Posts using Classification | Irena Spasic | Surya Thottam Valappil | 2425-CM3203 | |
| Project Proposal: Emotion Classification for Social Media Posts BACKGROUND: Social media platforms like Instagram, YouTube, and Twitter play a major role in shaping public opinion. However, on some platforms, features such as dislike counts are either hidden, unavailable, or absent altogether. This can make it difficult for content creators, brands, and everyday users to understand the emotional reactions to their posts. Despite the absence of explicit dislike indicators, analysing the text inte...[more] | |||||
| Shaikha Alshehhi | Adversarial Attacks and Defences for Industrial Control Systems (ICS) | Eirini S Anthi | Theodoros Spyridopoulos | 2425-CMT400 | |
| The increasing integration of Industrial Control Systems (ICS) with IT networks has increased their vulnerability to sophisticated cyber threats, which makes the implementation of robust Intrusion Detection Systems (IDS) essential. Although Machine Learning (ML) improves detection performance, these models remain inherently vulnerable to adversarial attacks, where input data is deliberately manipulated to cause misclassification. This study investigates the robustness of ML-based IDS against suc...[more] | |||||
| Shiv Kubal | Detection of In-Memory Credential Dumping Attacks | Neetesh Saxena | Charith Perera | 2425-CMT400 | |
| In-Memory credential theft attacks targeting Windows authentication systems have become a critical threat to enterprise security in recent years. Tools such as Mimikatz and its variants exploit the Local Security Authority Subsystem Service (LSASS) process to extract credentials directly from the system memory. This allows adversaries to escalate privileges, impersonate legitimate users and perform lateral movement within networks. Despite the introduction of security measures, attackers continu...[more] | |||||
| Shreyas Ramachandran | Welsh Flashcards for Vocabulary Learning | Paddy Slator | Crispin Cooper | 2425Place-CMT400 | |
| This project aims to create a simple web-based flashcard tool to help users learn Welsh vocabulary. Users will enter English words, and the platform will generate flashcards with the Welsh translation, pronunciation audio, and an example sentence. The flashcards will allow users to quiz themselves, testing their knowledge of Welsh words and phrases. As users go through the flashcards, they will receive immediate feedback, helping them reinforce their memory and improve pronunciation. This tool p...[more] | |||||
| Simon Fenner | Detecting Respiratory Disease from Coughs: A Lightweight Model for Smartphones | Yuhua Li | Paddy Slator | 2425-CMT400 | |
| The central research question is whether cough audio can be used to distinguish healthy individuals from those with respiratory illness in general (not limited to one disease) using a lightweight model. This dissertation targets the following objectives: •General Illness Detection: Develop a cough classiőcation model that detects signs of respiratory illness (vs. healthy), rather than focusing only on COVID-19. In our context, ’illness’ includes any respiratory conditions, including COVI...[more] | |||||
| Sk Shadid | An AI Framework for Predicting Nanoparticle Biodistribution in Cancer Treatment via Interpretable Ensemble Learning | Frank C Langbein | Surya Thottam Valappil | 2425-CM3203 | |
| This project presents a machine learning-based platform to predict nanoparticle biodistribution in cancer treatment, with a focus on tumour delivery efficiency. It leverages an ensemble of neural networks and gradient-boosted decision trees (XGBoost) to model how design features—such as size, surface charge, and coating—affect delivery outcomes in preclinical models. A key innovation is the integration of SciBERT text embeddings to extract semantic features from nanoparticle descriptions, e...[more] | |||||
| Su Ya | Exploring Large Language Models for Enhanced Game Experiences | Frank C Langbein | Richard Booth | 2425-CMT405 | |
| The integration of Large Language Models (LLMs) into game development presents an exciting frontier with the potential to revolutionize player experiences. LLMs offer unprecedented capabilities for generating dynamic content, creating believable and reactive characters, and enabling novel forms of player interaction. This project aims to explore the technical aspects of incorporating LLMs into games, focusing on their potential to enhance gameplay through dynamic content generation, AI-driven ch...[more] | |||||
| Suyog Magar | Exploiting and Mitigating Security Vulnerabilities in IoT Devices | George Theodorakopoulos | Christopher Wallbridge | 2425-CM3203 | |
| IoT devices (smart lightbulbs, smart plugs, smart cameras, etc) are entering the market without much regard for security. Following a systematic methodology, this project focuses on designing and implementing attacks against IoT devices (to be decided) or on defences that mitigate existing attacks. This project requires willingness to get hands-on; learning systematic methodologies, linux Command Lines, network traffic sniffing and pen testing.[more] | |||||
| Sydney Pollock | LLM-Driven Argument Mining and Feedback Generation for Persuasive Writing Assessment | Fernando Alva Manchego | Dr Soumya Barathi | 2425-CMT405 | |
| This research develops a multi-stage pipeline combining Argument Mining and Feedback Generation for persuasive writing assessment using smaller, open-source LLMs (Llama-3.2-3B-Instruct and Gemma-2-2B-IT). The study evaluates twelve different prompt engineering combinations investigating zero-shot versus few-shot prompting, chain-of-thought reasoning and variations in the quality of few-shot examples. Results show Gemma models consistently outperformed Llama in argument component identification (...[more] | |||||
| Tammy Davies | A SYSTEM DYNAMICS INVESTIGATION INTO THE GENDER GAP IN COMPUTER SCIENCE FOR HIGHER EDUCATION IN THE UNITED KINGDOM TO INFORM MITIGATING POLICY ACTIONS | Catherine Teehan | Padraig Corcoran | 2425-CMT404 | |
| An investigation into the gender diversity gap in UK computer science higher education, focusing on identifying and evaluating policy measures that could mitigate this disparity. It will use a qualitative system dynamics approach to map the interrelated factors influencing female participation, including recruitment, retention, and progression, through causal loop diagrams. The research will draw on academic literature, national statistics, and institutional policy documents to identify key feed...[more] | |||||
| Taylor Barber | Health and Wellbeing App | Dr Katarzyna Stawarz | Federico Liberatore | 2425-CM3203 | |
| This project aims to develop a user-friendly mobile application, that will allow users to monitor and track their health, fitness and nutritional progress. This app will cater to individuals looking to set and achieve their personal health goals.[more] | |||||
| Theodor Baur | Super-resolution Enhancement of Magnetic Resonance Images using Deep Learning | Frank C Langbein | Yazmin Ibanez Garcia | 2425-CMT400 | |
| Magnetic Resonance Imaging (MRI) is an indispensable tool in modern medical diagnosis, providing detailed anatomical and functional information. However, inherent limitations in MRI hardware, clinical workflow demands for faster acquisition times, energy and cost considerations often result in images acquired at lower resolutions (e.g., using 1.5T scanners or fast-scanning protocols). Lower resolution can compromise diagnostic accuracy, particularly for subtle pathologies or fine anatomical deta...[more] | |||||
| Thomas Hudson | Smart Flowerpot | Kirill Sidorov | Yuhua Li | 2425-CM3203 | |
| With the advent of the Internet of Things (IoT) and the proliferation of smart home devices, there has been an increased integration of intelligence and connectivity into everyday objects. Everything from refrigerators to door locks now possess "smart capabilities". Building on this momentum, this project aims to develop a Smart Flowerpot that can not only detect the hydration needs of a plant but also provide a visual record of its growth, all while having the ability to supplement its light so...[more] | |||||
| Thomas Light | GPU-Accelerated Quantum Mechanics Game | Crispin Cooper | Andrew Hood | 2425-CM3203 | |
| This project seeks to create a browser-based GPU-accelerated equivalent of “Quantum Marble Maze”, a game developed in 2016 by Dr. Crispin Cooper to help teach the fundamentals of quantum mechanics. The original game’s educational focus was undercut by two main flaws – a lack of browser support and poor simulation performance. This project sought to correct both these issues with a JavaScript re-write and GPU-acceleration respectively. The original game’s Java code was thoroughly inve...[more] | |||||
| Tianyang Zhang | Stylized Black and White Images from Photographs | Paul L Rosin | Christopher Wallbridge | 2324Place-CMT403 | |
| The aim of this project is to take an input gray level or colour image and generate from it a binary (i.e. black & white) output image. While this can be easily done with a simple thresholding approach, this project will investigate techniques that attempt to retain the essence of the input image and produce a more artistic binary image. The topic of this project comes under the heading of non-photorealistic rendering - which combines aspects of image processing with computer graphics. My ...[more] | |||||
| Tianyi Yu | Supporting Physical Activity at Home to Reduce Sedentary Behaviours | Dr Katarzyna Stawarz | Jenny Highfield | 2425Place-CMT403 | |
| Research question: How can technology be used to effectively reduce sedentary behavior? What technologies (e.g., apps, reminder systems, fitness trackers) are most effective in reducing sedentary behavior? Research content: Design a simple app that reminds users to do short exercises at regular intervals and analyze the impact of different reminder strategies (e.g., time intervals, reminder content, exercise types) on reducing sedentary behavior. To begin, determine the target group: Is the pr...[more] | |||||
| Timothy Shimman Huang | Gamification of a Typing Application | Fernando Alva Manchego | Yipeng Qin | 2425-CM3203 | |
| Typing is an incredibly vital soft skill in this current era, as jobs and communication become ever more digitalized, yet it remains undervalued in formal education, leaving many at an amateur level. Existing typing applications often contain flaws that can disengage users such as excessive competition or monotonous practice. The aim of this project was to create a unique solution by implementing gamification mechanics into a typing application that could motivate users of all typing proficiency...[more] | |||||
| Toby Phipp | Explainable Loan Approval System | Alexia Zoumpoulaki | Surya Thottam Valappil | 2425-CM3203 | |
| Using open datasets, develop an (online) application that allows you to load data and train a model that predicts loan approvals based on various input parameters, and more importantly, provides clear and understandable reasons for its decisions. Data Collection: Gather datasets with features like credit score, income, employment history, loan amount, etc. Model Development: Use a ML models like a Decision Tree or Random Forest for its inherent explainability. Explainability: Integr...[more] | |||||
| Tong Li | Development of inventory management systems for robotics laboratories | Victor Romero Cano | Paul L Rosin | 2425-CMT400 | |
| This project aims to design and implement an advanced inventory management system tailored to the unique needs of robotics laboratories. Robotics labs often handle a diverse range of components, including robots, sensors and specialised tools, making efficient inventory tracking a challenging task. The proposed system will utilise modern technologies such as open-source software, barcode or RFID tagging, and cloud-based databases to streamline inventory management. The aim is to enable precise ...[more] | |||||
| Tsz Kwong | Develop kinetic energy harvesting shoes to power healthcare wearables on the go | Nhat (Nick) Pham | Neetesh Saxena | 2425-CM3203 | |
| The wearable healthcare market has been experiencing significant growth in recent years, reaching $71.91 billion in 2023 and over $186 billion globally by 2030. In the near future, there could be hundreds of wearable devices that “invisibility” weave themselves into the fabrics of everyday wearables (e.g., clothes, hats, eyeglasses, earphones, shoes, etc.), providing us with all information about our health and wellbeing. To enable this massive number of wearable devices, frequent charging o...[more] | |||||
| Vincent Chan | National Rail Data Coverage Gap Mapping | Liam Turner | Paul L Rosin | 2425-CM3203 | |
| As a frequent train commuter to and from Cardiff, I experience regular interruptions to my mobile data, causing my lecture videos and web browsing to cut out frequently. Coverage gaps are not unique to my mobile network operator (MNO,) these interruptions are widespread across different operators, affecting many UK commuters. (https://www.railforums.co.uk/threads/trivia-significant-gaps-in-mobile-coverage-and-wifi-on-the-rail-network.236818/). While MNOs provide online coverage maps, these often...[more] | |||||
| Vlad-Mihai Sargu | Build an AI based solution and User Interface for the Liar’s Dice game | Walter Colombo | Amir Javed | 2425-CM3203 | |
| With the field of imperfect information games, creating strong agents is a major concern in Artificial Intelligence (AI) research and practice. Each game needs a specific agent trained by a set of related techniques such as Randomized Algorithm (RA) or Machine Learning (ML) algorithms. The objective of this project is to build AI based solutions for the single-agent problem Liar’s Dice game, which is a popular imperfect information game, and to optimise a UI for the resulting implementations. ...[more] | |||||
| Weiye Luo | CT Image Segmentation and Classification | Yukun Lai | Crispin Cooper | 2425-CMT400 | |
| We have seen significant progress in medical imaging, which provides doctors/radiologists with a powerful tool for early diagnosis of diseases. However, it requires substantial expertise and is also time consuming to analyse CT images to identify and segment tumours. This project involves implementing some deep neural network based architectures and carry out experiments to test their strengths and weaknesses. The project will have a co-supervisor from PETIC (Wales Research and Diagnostic PET Im...[more] | |||||
| Wenzhi Hu | An Android-Based Deep Learning System for Emotion Recognition: Face Detection and Speech Sentiment Fusion | Hiroyuki Kido | Mohammad Taher Pilehvar | 2425-CM3203 | |
| This project aims to develop an Android-based emotion recognition application that integrates facial detection in camera feeds (using Resnet-18 or other object detection algorithms) with speech sentiment emotion analysis (MFCC+MLP). By fusing the outputs, the app will provide a comprehensive result to the user's current emotional state and generate results to the users.[more] | |||||
| William Collyer | Investigating and Evaluating Methods for Indoor Wi-Fi Based Localisation | Padraig Corcoran | Tingting Li | 2425-CM3203 | |
| Localization concerns the problem of determining the physical location of a person or device. GPS is a commonly used localization method. However, this method does not work indoors due to the lack of a clear path between the GPS receiver and satellites. Wi-Fi positioning systems are commonly used to perform localization indoors. They involve measuring the position of a Wi-Fi receiver relative to a set of wireless access points (https://en.wikipedia.org/wiki/Wi-Fi_positioning_system). The a...[more] | |||||
| William Sawtell | Motion Planning for Robot Object Manipulation Using Machine Learning | Juan Hernandez Vega | Federico Liberatore | 2425-CM3203 | |
| Motion planning is a relevant research area in robotics and its purpose is to find a collision-free path from a start to a goal robot configuration. Especially for manipulator arms, planning problems require contact-rich interactions between a robot and the objects in the environment without collision. There are several techniques to solve these problems, such as search-based and sampling-based algorithms. Besides these techniques, learning-based methods are used to enhance efficiency at solving...[more] | |||||
| William Sims | DAIST: Datacentre as an Interconnected System Testbed | Jenny Highfield | Jandson Santos Ribeiro Santos | 2425-CM3203 | |
| Testbeds are used within cybersecurity to replicate Cyber Physical Systems (CPS) to provide a safe environment for cyberattack research. This project is to create a prototype testbed from a microcontroller/ microprocessor (ie. Arduino/ Raspberry Pi) and connected sensors. Supervisory Control and Data Acquisition (SCADA) Systems, take in realtime sensor data and make changes according to conditions. You will be building a prototype to represent a SCADA System, and will discuss how representative ...[more] | |||||
| Will Scully | Evaluating Rendering Techniques & Optimisations for Non-Euclidean Environments | Frank C Langbein | Tingting Li | 2425-CM3203 | |
| Originating from the game Portal, portals are pairs of blue & orange holes which can be used to create non-Euclidean environments by linking locations over a distance via an impossibly shallow doorway. This project tries to re-create their visual effect utilising Geometry Shaders, as opposed to using the Stencil Buffer or Render Textures, which other approaches have used previously. In the end, three distinct renderer implementations are used – A single-pass forward renderer, a multi-pass ...[more] | |||||
| Xiaoxiang Chen | Developing Step-by-Step Examples for Microstructural Models in MicroTorch: Advancing Self-Supervised Machine Learning for MRI | Paddy Slator | Wei Zhou | 2425-CMT403 | |
| Magnetic resonance imaging (MRI) is crucial for diagnosing and monitoring various diseases. Many emerging MRI techniques use machine learning to generate microstructural maps—representations of small-scale tissue properties such as axon diameter, cell size, and blood flow. Self-supervised machine learning is proving to be a promising approach for producing these maps [1,2]. This project will contribute to the development of MicroTorch, an open-source self-supervised machine learning package fo...[more] | |||||
| Xingyu Wang | Predicting Word-Level Errors in Chinese Learners’ English: Integrating Linguistic Features and L1 Transfer | Fernando Alva Manchego | Richard Booth | 2425-CMT405 | |
| This study predicts word-level errors in Chinese learners’ English writing, testing whether features from first language (L1) negative transfer improve performance and interpretability. A unified dataset was built from three corpora (FCE, Lang-8, CAWSE). Models include Logistic Regression, LightGBM, BERT, and few-shot prompting. Beyond baseline linguistic features (length, frequency, POS), four L1 transfer categories were introduced: high-risk word lists, grammatical templates, tense/aspect ma...[more] | |||||
| Xinhao Guo | AI-Generated Image Detection | Wei Zhou | Dr Daniel J. Finnegan | 2425-CMT404 | |
| Nowadays, there emerged lots of AI generated images that are different from natural images. This project aims to develop methods to distinguish AI generated images from natural images. It would be better to send me your CV when contacted.[more] | |||||
| Xinru Wang | Labeling and Quantifying Biases in Text Generated by Pre-trained Language Models | Nedjma Ousidhoum | Jandson Santos Ribeiro Santos | 2425-CM3203 | |
| In this work, the student will look into NLP tools and train models to study bias in text. The student will be initiated to different techniques commonly used in computational social science and natural language processing and will choose the kind of study they'd like to conduct in any language. They will then train a model for further analysis and share their insights.[more] | |||||
| Yasin Biswas | See through my eyes - A VR Shopping Experience from the perception of the neurodiverse | Annelies Gibson | Hiroyuki Kido | 2425-CMT403 | |
| The aim of this project is to create a VR shopping experience with a twist, this will be from the perspective of a neurodiverse individual, which will show a neurotypical person how the various sensory inputs can effect an autistic individual when going shopping.[more] | |||||
| Yi Wang | CT Image Segmentation and Classification | Yukun Lai | Dr Daniel J. Finnegan | 2425-CMT400 | |
| We have seen significant progress in medical imaging, which provides doctors/radiologists with a powerful tool for early diagnosis of diseases. However, it requires substantial expertise and is also time consuming to analyse CT images to identify and segment tumours. This project involves implementing some deep neural network based architectures and carry out experiments to test their strengths and weaknesses. The project will have a co-supervisor from PETIC (Wales Research and Diagnostic PET Im...[more] | |||||
| Yixuan Liu | Using homomorphic encryption to protect privacy in the IoT | Shancang Li | George Theodorakopoulos | 2425-CMT400 | |
| This project aims to develop a solution to use homomorphic encryption to protect sensitive information in the IoT environment. Basically, this project needs to consider the privacy of data owner, data user, and server and protect all participants privacy without compromising the functionalities of all parts.[more] | |||||
| Yuchuan Chu | CT Image Segmentation and Classification | Yukun Lai | Yuhua Li | 2425-CMT400 | |
| We have seen significant progress in medical imaging, which provides doctors/radiologists with a powerful tool for early diagnosis of diseases. However, it requires substantial expertise and is also time consuming to analyse CT images to identify and segment tumours. This project involves implementing some deep neural network based architectures and carry out experiments to test their strengths and weaknesses. The project will have a co-supervisor from PETIC (Wales Research and Diagnostic PET Im...[more] | |||||
| Yue Zeng | Enhancing User Experience on Medical Forums using NLP and Knowledge Engineering | Alia I Abdelmoty | Fernando Alva Manchego | 2425-CMT400 | |
| This project will consider users' posts on Medical Forums (OCD forum as an example). Users of these forums share stories, experiences and ask questions. A tool that can automatically identify relevant tags for the users' posts can enhance search and retrieval. This project will develop a tool to annotate a user's post with tags/tag phrases that are relevant to OCD. A basic solution is to identify tags using concepts that exist in the text of the post using NLP techinques (named entity recognit...[more] | |||||
| Yuliana Karaivanova | Creating A Framework For Improving Character Consistency In AI Generated Movie Scripts | Liam Turner | Yazmin Ibanez Garcia | 2425-CM3203 | |
| AI-generated scripts often struggle with character consistency, where characters’ personalities, motivations, and speech styles shift unpredictably. This project aims to develop a framework to improve character consistency in AI-generated scripts and explore methods to improve it using structured prompting techniques. Project Objectives: 1. To develop the Character Consistency Modelling Framework (CCMF) Create a multi-dimensional framework for modelling consistent fictional characters. •Syn...[more] | |||||
| Yuv Raikhy | Solving the Traveling Tournament Problem | Nico Potyka | Yipeng Qin | 2425-CM3203 | |
| The Traveling Tournament Problem [1] is an interesting combinatorial optimization problem that occurs in the planning of schedules for sports leagues and tournaments. Roughly speaking, the problem is to schedule games such that every team meets every other team twice (a double round-robin tournament), the length of home stands (consecutive home games) and road trips (consecutive away games) meets constraints and the overall travel distance is minimized. The goal of this project is to do some li...[more] | |||||
| Zaid Javed Khan | [Industry] Your Dental Health Assistant: Chatbot Revolution | Fernando Loizides | Alia I Abdelmoty | 2425-CMT405 | |
| [Care4teeth Limited] To develop an innovative chatbot to assist users with dental concerns and promote optimal oral health via an accessible web and app interface. This chatbot will provide timely and reliable information on common dental problems, preventive care, and the latest health guidelines from current NHS government guidelines and evidence-based dental resources. As a first point of contact for dental inquiries, the chatbot will enable users to engage in informative dialogue about symp...[more] | |||||
| Zheyuan Li | A Review Summarization Framework Integrating ABSA Driven Clustering and Hierarchical Methods | Fernando Alva Manchego | Martin Caminada | 2425-CMT400 | |
| E-commerce reviews are often huge in number, scattered in opinion, and full of repetition. To handle this, I design a layered framework that first clusters reviews with Aspect-Based Sentiment Analysis (ABSA) and then generates summaries in two stages. A weak-reference evaluation is added by rewriting official references from dataset AmaSum with an Large Language Model, so score bias from structural mismatch is reduced. In this project, I'll put into practice: Aspect clustering that maintains op...[more] | |||||
| Zhiqiang Song | AI-Generated Image Quality Assessment | Wei Zhou | Carolina Fuentes Toro | 2425-CMT404 | |
| Recent years have witnessed the fast development of Artificial Intelligence (AI) technology, large model-driven generative AI has become an emerging topic. In this project, we would like to evaluate the quality of AI generated images. It would be better to send me your CV when contacted.[more] | |||||