Year:

Projects from 2026

Student	Title	Supervisor	Moderator	Module
Abdullah Sarker	Assessing Rail Network Vulnerability Through Spatial Network Analysis	Crispin Cooper	Victor Romero Cano	2526-CM3203
This project will model a railway network as a weighted graph using OSM data. Centrality measures can be used to identify important stations and railway corridors. Disruptions can be simulated by removing important nodes or edges in order to gauge consequent network fragmentation and changes to connectivity. GIS will be used to visualise the distribution of vulnerable points and deduce patterns with regards to urban geography and travel demand. The project will combine a thorough evaluation of v...[more]
Adrian Rivera	Designing an Interactive Mobile Application to Promote Healthy Eating Practices Among Young Adults Aged 18-25	Nervo Verdezoto Dias	Mohammad Taher Pilehvar	2526-CM3203
Prerequisites: -Basic knowledge of Human-Computer Interaction and user-centered and/or participatory design methods -Experience developing mobile and/or web applications in several platforms (e.g., Android, HTML5 & Javascript, etc.) -Desirable experience with Spanish (not mandatory) Digital health technologies are becoming increasingly prevalent to support self-care practices aiming to enhance people’s health and wellbeing. However, their potential to promote healthy eating practices amon...[more]
Aisha Afzal	Assessing the accuracy of MR tractography in capturing intra-operative brain shift.	Stefano Zappala	Dr Soumya Barathi	2526-CM3203
Tractography is a valuable tool to capture the distribution of white matter tracts in the brain. Its impact in neurosurgery planning is clear, where the least damage to important nerve tract is critical. The technique, although promising and non-invasive, it depends on quality of data and robustness of the modelling: its use in a clinical setting is still limited. The simplest form of change in anatomy is given by brain shift, the physiological small redistribution of the brain tissues given ...[more]
Albert Collins	Machine Learning and AI-Based Prediction and Optimisation of Vehicle Performance Metrics Using Simulated Telemetry Data	Walter Colombo	Alexia Zoumpoulaki	2526-CM3203
1. Introduction and Background Data-driven optimisation is central to modern motorsport and automotive engineering. Teams like Mercedes-AMG Petronas Formula One use telemetry, simulation, and predictive algorithms to refine setups, boost reliability, and maximise performance. The rise of high-fidelity simulation tools and open datasets now enables the exploration of similar methods through machine learning. This dissertation proposes a system that predicts vehicle performance metrics—such as...[more]
Angelos Kiousopoulos	Underwater Image Enhancement	Wei Zhou	Yazmin Ibanez Garcia	2526-CM3203
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]
Anthony Jace	Software to predict the winner of a Formula 1 race	Fernando Alva Manchego	Theodoros Spyridopoulos	2526-CM3203
A software that tries to predict the winner of a F1 race based on driver form and skill, car speed as well as taking into account factors such as weather, track type and whether a car has done well in tracks similar to the one being predicted.[more]
Arnav Jain	Identifying Demographic Bias in Automated CV Screening Using Large Language Models	Nedjma Ousidhoum	Federico Liberatore	2526-CM3203
In this work, the student will study the motivation for using natural language processing to assist with different forms of bias identification. They will then choose the task and the domain they would like to focus on (e.g., social media, forum data, LLM-generated data, …). In one of the topics, the student will come up with a new framework to tackle the problem based on work in progress.[more]
Arrian Ghassemy Tabatabai	Feasibility and Analysis of AI agents in Agile Software Development	Nitesh Kumar	Sylwia Polberg	2526-CM3203
This project explores how Large Language Model (LLM) based agents can participate as autonomous software developers within an Agile and DevOps workflow. The vision is a development model where a single human acts as Product Owner and reviewer, while multiple AI agents implement small user stories, raise merge requests, and interact with continuous integration (CI) pipelines. The human focuses on defining and reviewing work rather than writing code directly. This project contributes to emerging ...[more]
Arsalaan Ahmad	Do Saliency Benchmarks Reflect Human-Like Attention? A Structural Analysis Using CAT2000	Hantao Liu	Yuhua Li	2526-CM3203
This project investigates whether strong benchmark performance in visual saliency modelling is accompanied by behavioural fidelity to human gaze. Using the CAT2000 eye-tracking dataset, a failure-oriented evaluation framework is applied to assess state-of-the-art saliency models using both standard benchmark metrics and additional structural measures of spatial dispersion, attention fragmentation, and centre dependence. The analysis identifies systematic mismatches between benchmark success and ...[more]
Artin Abdoullahi Mahavi	Developing a web app to support dietitians in caring for women with gestational diabetes	Nervo Verdezoto Dias	Alexia Zoumpoulaki	2526-CM3203
Gestational diabetes is a complex condition that pregnant women can face, which involves continuous dietary management. This can be difficult for dietitians, as they have to create personalised meal plans for each patient, depending on their needs, with extremely stringent criteria to ensure it is effective and safe (McIntyre et al. 2019). This is extremely time-consuming, and relies on traditional methods. NUTRIA is an ongoing project developed by Fraunhofer AICOS that aims to solve this issue,...[more]
Ayoub Chahid	Participatory development of resource booking and inventory management systems for university laboratories	Victor Romero Cano	Wei Zhou	2526-CM3203
This project aims to design and implement a resource booking and inventory management system tailored to the unique needs of university laboratories, with the Human Centred Computing and Robotics (HCCR) laboratory as use case. Robotics labs often handle a diverse range of components, including robots, sensors and specialised tools, making efficient booking and inventory tracking a challenging task. The student should: - use existing implementations of a Resource Booking System for the Human Ce...[more]
Benjamin Jones	Quantitative Analysis of Sheepdog Trial Performance using Computer Vision	Paul L Rosin	Yukun Lai	2526-CM3203
Straight Huntaway sheepdog trials currently rely on subjective human analysis, lacking automated frameworks for performance evaluation. This research proposes an end-to-end computer vision pipeline designed to quantify trial performance from recorded trial footage. The system employs a YOLOv26 detector to extract precise coordinate data for both sheepdog and sheep from video sequences, which are subsequently engineered into interactional behavioural metrics. Performance is evaluated via an ensem...[more]
Callum Smith	Hierarchical Phishing Email Detection and Attack Type Classification	Neetesh Saxena	Irena Spasic	2526-CM3203
Use the existing dataset(s) and apply DL techniques to classify data into benign and malicious. Also do a multi-classification using either DL or ruled based analysis (kind of semi auto, you can run Python scripts to do this part as well) to perform deep analysis about the email sources, contents, nature of engineering techniques, formation of email messages, crafts etc.. Dataset: https://github.com/rokibulroni/Phishing-Email-Dataset.git https://research.utwente.nl/en/datasets/phishing-valid...[more]
Calum Best	Evaluation of LLMs Against Traditional Supervised ML for Financial Trend Prediction	Hiroyuki Kido	Hantao Liu	2526-CM3203
This study aims to evaluate the performance of zero-shot large language models (LLMs) against traditional supervised machine learning models for next-day market direction prediction using the S&P 500 index. Three LLMs (GPT-4o-mini, Claude Haiku 4.5 and Qwen3.5) were compared against two supervised ML models (Logistic Regression and Random Forest) using the same test dataset of 199 rows of S&P 500 features. These were compared against a majority-class baseline of 55.78%, with the supervis...[more]
Cameryn Luis	Something's Phishy: A Behavioural Telemetry Study of Cybersecurity Decision-Making in a Web-Based Learning Game	Eirini S Anthi	Yazmin Ibanez Garcia	2526-CM3203
Phishing remains one of the leading causes of cybersecurity breaches, succeeding by exploiting how people make decisions under pressure rather than by defeating technical defences. Awareness training has been shown to reduce susceptibility, but evaluation in this field focuses overwhelmingly on outcome accuracy and tells us little about which behavioural changes explain any improvement. This project addresses that gap through Something’s Phishy, a web-based cybersecurity learning game in whic...[more]
Carys Gratton	Industry-Based Project - Cyber Security Awareness Games	Catherine Teehan	Matthew J W Morgan	2526-CM3203
This project is collaboration with an external organisation. The aim of this project is to create a set of interactive games that raise cyber security awareness in an engaging way. The desired outcome is a functional, educational tool that teaches a broad and possibly non-technical audience, to identify cyber threats such as phishing, social engineering, malware and insider threats. The benefit is both a resource to promote cyber resilience in the wider community and provide a way to gamify and ...[more]
Chidi Anagu	Evaluating AI Image Detection Methods on Out-Of-Distribution Data	Bailin Deng	Surya Thottam Valappil	2526-CM3203
In this project, you will develop a method to detect images generated with AI.[more]
Ching Ng	Explainable Loan Approval System	Alexia Zoumpoulaki	Bailin Deng	2526-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. 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]
Ciaran Flynn	Technocamps Islands	Catherine Teehan	Carla Perez Almendros	2526-CM3203
Using Minecraft for Education, create islands that will allow users to explore core computational thinking theories and methods through play. The islands should include a range of interactive activities and supporting educational materials that will engage and encourage children and young people to explore computer science as an option at GCSE and beyond. The islands will need to align with the new 2022 Welsh curriculum for science and technology. This includes all elements of CS and can be ad...[more]
Connor O'Hagan	Optimal Urban Park Placement Using Spatial Analysis	Padraig Corcoran	Shancang Li	2526-CM3203
This report explains the development of a GIS-based decision-support tool that identifies candidate car park sites for potential conversion into new urban parks. It uses openly available spatial and statistical datasets, the tool combines network-based accessibility modelling with multi-criteria scoring based on demand, underserved demand, distance from existing green spaces, site size and flood suitability. Case studies in Cardiff and Swansea show that the tool can be used to identify locations...[more]
Corey Taylor	A Multi-layered Framework for Detecting AI-Generated Phishing Attacks	Amir Javed	Mohammad Taher Pilehvar	2526-CM3203
Phishing attacks are increasingly sophisticated due to the widespread availability of generative artificial intelligence (AI), which enables attackers to produce highly personalised phishing emails that are more difficult to detect using traditional rule-based and stylistic detection methods. This creates significant challenges for automated detection systems, particularly in maintaining robustness across shifts in time periods and domains. Prior work has explored both the use of stylometric fea...[more]
Daniel Davis	Development of Model-Predictive-Control (MPC) Systems for autonomous mobile robots	Victor Romero Cano	Richard Booth	2526-CM3203
This project explores the implementation of Model Predictive Control (MPC) for autonomous navigation and motion control in mobile robots. MPC is a powerful control strategy that optimises future control actions based on a predictive model of the robot's dynamics and constraints. The student will select one robot platform (listed below), develop an MPC framework tailored to its mobility and sensing capabilities. The project will involve modelling the robot’s kinematics, designing the MPC contro...[more]
Daniel Tannous	Detecting Market Anomalies Using Graph-Based Machine Learning Across Financial Markets	Walter Colombo	Irena Spasic	2526-CM3203
This project aims to develop a graph-based machine learning framework for detecting anomalous behaviour in financial markets, with a primary focus on the US stock market and optional extension to cryptocurrency markets. The project will model market activity as a graph, where nodes represent financial assets (e.g., US equities or cryptocurrencies) and edges capture relationships such as abnormal correlations, co-movements, or unusual trading activity. By analysing the evolving structure of thes...[more]
Darrshan Mahendran	Teaching Allocation system	Surya Thottam Valappil	Liam Turner	2526-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]
Darsh Chanduka	Investigating performance improvements of just-in-time compilers on emulated processors	Ramalakshmi Vaidhiyanathan	Hantao Liu	2526-CM3203
Just-In-Time (JIT) compilers dynamically translate code into machine instructions at runtime, enabling faster execution compared to interpreted approaches. In processor emulation, target machine instructions are often interpreted by the host system, which introduces significant performance overhead. This project aims to bridge the performance gap between interpreted emulation and native execution by investigating how JIT compilation techniques can improve the efficiency of emulated processors. ...[more]
Darsh Kanjani	Cross-Dataset Generalisation of Flow-Based IoT Device Identification: An Empirical Three-Tier Evaluation	George Theodorakopoulos	Stephen Miles	2526-CM3203
Passive identification of IoT devices from network traffic is a valuable capability for asset visibility, segmentation, and security monitoring, but many machine-learning classifiers are evaluated only within a single dataset or collection environment. This project investigates whether flow-based IoT device-identification models remain reliable when tested under progressively more realistic forms of dataset and environment shift. The project develops a leakage-aware NFStream pipeline to extract...[more]
David Twigger	Explainable SMS Phishing Detection using LLM-based Agents	Neetesh Saxena	Víctor Gutiérrez Basulto	2526-CM3203
Read here: https://www.usenix.org/system/files/soups2025-wang.pdf Come up with your idea on doing a similar work.[more]
Dionizy Szczerba	Lightweight machine intelligence for resource constrained devices in the Internet of Things	Yuhua Li	Steven Silva Mendoza	2526-CM3203
This project is particularly suitable for those students who wish to pursue further study in a PhD in novel, lightweight and efficient machine learning. The proliferation of the Internet of Things (IoT) has led to the constant generation of massive data from electronic devices. The techniques that make sense of data largely rely on machine learning. However, conventional machine learning, particularly deep learning, demands great computational power and consumer high energy so they are suitably...[more]
Ela Sayar	Fitness vs. Truth: Simulating the Evolution of Perceptual Interfaces	Kirill Sidorov	Shancang Li	2526-CM3203
Cognitive scientist Donald Hoffman’s Interface Theory of Perception (ITP) proposes that natural selection favours perceptual systems shaped by fitness rather than by the world’s true structure, much as the icons of a computer hide the underlying hardware yet enable the user to interact with it. In line with this, Mark, Marion and Hoffman (2010) introduce the Interface Game, a one-dimensional resource-selection task in which competing perceptual strategies are tested across many simulated gen...[more]
Ethan Hudson-Lloyd	Pepper Robot as a Social Engineering Vector: Evaluating the Security of Traditional and User-Generated Security Questions	Christopher Wallbridge	Padraig Corcoran	2526-CM3203
This pilot study aims to see if social robots such as pepper robot (PR) are feasible as social engineering attack vectors. This will be done by having the robot attempt to extract information relating to both common security questions (for example name of first pet) and user-created security questions.[more]
Fiachra Bermingham	Machine Learning-Driven Game-Theoretic Simulation of Economic Conflict Between nations	Richard Booth	Bailin Deng	2526-CM3203
Creating a system that simulates strategic economic behaviour between nations in conflict. The system will explore how countries adapt policies-such as sanctions, tariffs, or alliances using machine learning to maximize their own outcomes within dynamic, game-theoretic environment.[more]
Frank Buck	AgileAI: Autonomous Multi-Agent Software Development in a DevOps Workflow with Human Oversight	Nitesh Kumar	Jandson Santos Ribeiro Santos	2526-CM3203
This dissertation investigates whether LLM-based agents can autonomously implement, test, and validate software within an Agile DevOps workflow while maintaining effective human oversight. It presents AgileAI, a multi-agent system that integrates with GitLab's issue tracker and continuous integration / continuous deployment (CI/CD) pipeline. A human Product Owner approves user stories via labels; a planner agent groups approved stories into sprints; developer and test-generation agents implement...[more]
Geneve Purayil	Interactive Text Simplification with Complexity Visualization	Alexia Zoumpoulaki	Jing Wu	2526-CM3203
This project would develop a system that simplifies complex text while showing users what changes are being made and why. The system would identify different types of complexity in text (difficult words, complex sentences, abstract concepts) and allow users to control how much simplification they want. A key innovation would be visual feedback showing which parts of the text are complex and how they're being simplified. Users could see color-coded complexity maps and track transformations step-b...[more]
George Barnsley	Spatial Reasoning with a Robot	Christopher Wallbridge	Hiroyuki Kido	2526-CM3203
Just looking at something the way it is is not enough to understand it. True understanding comes from looking at the way things interact. A good example of this is location control in spatial relations. Why do we get IN a car but ON a bus? Because a car has more location control of us than a bus does. If a robot needs to use the correct word it somehow needs to gain a knowledge of the correct interaction. One way would be to HYPOTHETICALLY consider the movement of one object in relation to anot...[more]
Giuliana Emberson Lato	5GHz Machine Learning Device Positioning Prediction Model	Victor Romero Cano	Jing Wu	2526-CM3203
The Global Positioning System (GPS) is widely used for outdoor positioning and navigation. However, GPS relies on satellite signals that are significantly attenuated by buildings and obstacles, making it unreliable indoors. As a result, alternative approaches are required to enable accurate device positioning in scenarios with multipath interference and signal attenuation. This project aims to address the limitations of GPS by developing a machine learning (ML)-based positioning prediction mode...[more]
Guodong Liang	A GIS-based Location-Allocation System for On-Street Electric Vehicle Charging Points in Cardiff	Jing Wu	Hantao Liu	2526-CM3203
Background Electric vehicle (EV) adoption is growing rapidly, and cities need to decide where to place new public charge points so that they are accessible, fair and cost-effective. This project will focus on Cardiff as a case study and investigate how spatial data, road networks and optimisation techniques can be combined into a practical decision-support tool for planning on-street EV charging infrastructure. Aim The aim of the project is to design and implement a GIS-based location-allocatio...[more]
Harrison Boon	Cryptographic Analysis of 2D Second-Order Additive Cellular Automata for Image Encryption	Paul L Rosin	Xianfang Sun	2526-CM3203
Cellular automata (CA) are discrete computational models capable of generating complex behaviour from simple local rules, making them a natural candidate for lightweight image encryption on resource-constrained hardware. While prior CA-based encryption schemes have demonstrated strong cryptographic performance, no existing work characterises how that performance is distributed across the full keyspace. This dissertation addresses that gap by implementing a 2D second-order reversible CA-based ima...[more]
Harry Wyatt	Online detection, grouping and tracking of human subjects for robotics application	Victor Romero Cano	Hantao Liu	2526-CM3203
Future social robots must exhibit social awareness to safely and successfully coexist in human-shared spaces, such as being able to act based on human presence and actions. In order to do this, they must be able to perceive humans and their social structures. This dissertation proposes and implements a two-tiered pipeline for human perception and tracking, and also for group detection. The solution utilises SAM 3 for human detection, ByteTrack for tracking and an LMM to perform socially-aware de...[more]
Heledd Jubb	Evaluating Urban Accessibility Through Personalised Scoring	Padraig Corcoran	Yazmin Ibanez Garcia	2526-CM3203
Accessibility — the ease with which individuals can reach needed or desired activities — is a relative measure that varies according to a person's age, health, wealth and preferences. I developed a tool to evaluate accessibility in urban areas by modeling street networks as graphs, computing metrics like shortest path distances to key amenities (e.g., schools, hospitals, parks) and considering a user's preference of amenities and maximum walking distance. I used OpenStreetMap (OSM), Ordnan...[more]
Hritika Kaila	Refusal Taxonomy and Over-Correction in LLMs	Nedjma Ousidhoum	Alexia Zoumpoulaki	2526-CM3203
While significant work has been done on technical over-correction, the human element, specifically how social class, role, and education and user intent influence the frequency of refusals, remains relatively under-explored. It has been demonstrated that lower SES (socioeconomic status) users tend to use more 'concrete' language and anthropomorphise AI more frequently, i.e., using greetings. On the other hand higher SES users utilise higher levels of abstraction. This project will investigate...[more]
Irisha Kotwani	Reference checker	Sylwia Polberg	Natasha Edwards	2526-CM3203
The purpose of this project is to create a tool that will be able to scan PDF files for references and then proceed to check whether they exist or have been hallucinated.[more]
Ismael Shaukat-Lopez	Design and Simulation of a Blockchain with Suspicion-Based Anomaly Detection and Response Mechanisms	Shancang Li	Yazmin Ibanez Garcia	2526-CM3203
Traditional blockchain systems validate transactions as either valid or invalid but lack any mechanism to monitor suspicious behavioural patterns. This means that nodes can act maliciously whilst remaining technically compliant and undetected. This project addresses that limitation by designing and implementing a decentralised blockchain simulation in Python using a UTXO transaction model and Proof of Work consensus. This is extended with a suspicion scoring system that continuously monitors nod...[more]
Isra Ismail	Implementation of functional updates for a Cardiff community charity website	Nervo Verdezoto Dias	Wei Zhou	2526-CM3203
This project will be in collaboration with Grange Pavilion (GP) and Cardiff University’s Community Gateway. The current website for Grange Pavilion (https://grangepavilion.wales/) was initially designed by Cardiff University’s web team along with GP staff, however, some content is now out of date and updates/improvements to its functionality is required as the charity moves into its next phase. We will require a student to: - Conduct a short evaluation of the current website and suggest p...[more]
Jack Brocklehurst-Read	Code Quest	Catherine Teehan	Stefano Zappala	2526-CM3203
Create a game called Code Quest using either Twine, Scratch 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]
Jack Downie	AI chatbot to provide guidance and reassurance to diabetic eye screening patients	Stuart M Allen	Mohammad Taher Pilehvar	2526-CM3203
Following diabetic eye screening, patients receive formal results letters that can be hard to interpret. For example, the overall outcome may be that no changes have been seen, some changes were observed but don’t require any further action at this time, or that the changes noticed require follow up and a further appointment with ophthalmology. The letters contain very little other information so it leaves patients stressed and anxious about whether they will lose their vision. The aim of this...[more]
Jack Pope	AI Generation of Music Videos	Bailin Deng	Kirill Sidorov	2526-CM3203
Recent advances in AI has made it possible to generate images and videos from text prompts (for example, see https://github.com/yzhang2016/video-generation-survey/blob/main/video-generation.md for a list of video generation models). 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. Assuming we know the lyr...[more]
Jack Stacey	Digital Forensic Techniques for Identifying and Analysing Compromises in Industrial Network Infrastructures	Laurence Semmens	Theodoros Spyridopoulos	2526-CM3203
This project will explore how digital forensic analysis can be carried out on components within an industrial computer network, including both modern equipment and legacy devices that may lack up-to-date security features. Industrial networks are used in areas such as manufacturing and energy, and typically include devices like PLCs, HMIs, robot controllers, industrial switches, and older controllers. The student will investigate what types of forensic evidence these devices can produce and how ...[more]
Jameel Ruhomaun	Machine Learning for Robustness Assessment of Quantum Controls	Frank C Langbein	Carla Perez Almendros	2526-CM3203
Quantum computers depend on moving information through tiny, noisy systems where small errors can compound rapidly. One way to build such a device is to arrange qubits in a ring and steer information from one site to another by tuning a fixed magnetic field at each site; each design is called a controller. The most useful way to score a controller is not its noise-free behaviour but how well it holds up when the system is being disturbed by random environmental noise. The standard score, the fir...[more]
James Crake-Merani	Reconstruct the Bidirectional Reflectance Function	Victor Romero Cano	Jandson Santos Ribeiro Santos	2526-CM3203
Create a forward simulation which allows the introduction of errors. If enough time, try to integrate it with off the shelf photogrammetry to get a 3D model. Try different approaches to the reconstruction, and data representation, comparing these approaches based on questions such as the sensitivity to positional, and image noise.[more]
James Holt	Forensic Analysis and Evaluation of Smart Home Hubs and Sensors	Laurence Semmens	Kirill Sidorov	2526-CM3203
Project Description This project will investigate the forensic analysis and security of devices commonly found in modern smart homes, including smart home hubs, a range of peripheral IoT sensors, and smart cameras. These systems often combine open-source and closed-source software, operate on wireless protocols, and store or transmit sensitive information, making them an important area for digital forensics and security research. The student will examine how forensic artefacts are generated, st...[more]
Jason Zhou	User Perceptions of the Security and Privacy Benefits of Social Media Apps	Neetesh Saxena	Fernando Alva Manchego	2526-CM3203
This project investigates whether users who feel safest on a social media platform actually take the most risks on it, examining the gap between perceived safety and real privacy behaviour across Snapchat, Facebook and WhatsApp. Using a mixed-methods design (survey and interviews) and a Python analysis pipeline that produces five composite scores per participant, it explores how platform design features, grooming risk awareness, and technical background shape that contradiction.[more]
Jasper Wilkes	Hierarchical metric-semantic SLAM methods for architectural mapping	Victor Romero Cano	Irena Spasic	2526-CM3203
Simultaneous Localisation and Mapping (SLAM) is a fundamental problem in robotics, involving the estimation of both the pose of a mobile robot and the map of its surroundings. This representations can then be used by the robot to make and execute decisions. Recent advancements address the SLAM problem by employing a graph-based representation that integrates metric (occupancy) and semantic (classes) estimates, along with hierarchical relationships. This enhanced representation, known as a 3D Dyn...[more]
Jiatong Zhu	Learning Disentangled Representations of Content and Layout in 2D Images	Jing Wu	Carla Perez Almendros	2526-CM3203
1 Context and Motivation Most existing methods for learning representations of 2D images leverages autoencoder architectures to learn a compressed latent representation that captures the overall content of the image. In simple cases such as the MNIST dataset, the latent space is structured around the digit identity, allowing for linear decoding of digit classes. However, if the dataset no longer guarantees consistent positional placement of the digits, the latent space becomes less structured...[more]
Jia Yin	Real-Time Animal Detection on Resource-Constrained Embedded Platforms	Jonas Beuchert	Amir Javed	2526-CM3203
Background Camera traps are essential for wildlife monitoring, but transmitting raw images is energy-intensive. Running object detection locally on embedded hardware enables intelligent filtering and event-based alerts. This project investigates deploying a vision model for animal detection on a microcontroller with an integrated camera interface. Key Topics Embedded computer vision Model compression and optimization Image preprocessing on constrained hardware Firmware development in C/C++ Pe...[more]
Joel Marsh-Trim	A Budget-Friendly Mobile Application for UK University Students: An Offline-First Approach to Meal Planning	Nervo Verdezoto Dias	Liam Turner	2526-CM3203
I propose a lightweight mobile app that helps students plan affordable, healthy meals and build cooking confidence. The MVP will include a recipe cost estimator (using a small reviewed price list), a pantry & leftovers tracker with use-by reminders, simple “swap to save/healthify” suggestions (e.g., ingredient substitutions and cooking method tips), and a 3–5 day meal plan + shopping list generator. The focus is practical, privacy-respecting functionality with a clean, accessible UI. n...[more]
Jonty Spall	FreeTxt: Aspect-Based Sentiment Analysis and Summarisation	Fernando Alva Manchego	Crispin Cooper	2526-CM3203
Whenever someone receives a survey, there are normally two types of questions: (1) ones where the respondant selects an option or provides a score; and (2) ones where the respondant can write anything they'd like, such us when providing comments or suggestions. This project aims to develop tools for obtaining useful information from the latter type of responses. FreeTxt (https://freetxt.app/ is a tool that "supports the analysis and visualisation of free-text data in both English and Welsh (e.g...[more]
Joseph Franklin	Creating a Web Application for Detecting Harmful Messages	Carla Perez Almendros	Theodoros Spyridopoulos	2526-CM3203
Journalists, campaigners or social media users often write messages that can involuntarily offend others, specially when talking about underrepresented groups or vulnerable communities. There are many models that can help to identify these potentially offensive messages, but they are not commonly known or easy to use by the general user. This project consist of creating a plug-in or web app which integrates one or several already existing models to detect offensive or harmful language in free...[more]
Joseph Harris	Application of Bipartite Tournaments in Educational Assessment: A Comparative Study of Student Proficiency and Question Difficulty	Richard Booth	Eirini S Anthi	2526-CM3203
A tournament consists of a finite set of players equipped with a beating relation describing pairwise comparisons between each pair of players. Determining a ranking of the players in a tournament has applications in voting, where players represent alternatives and x beats y if a majority of voters prefer x over y, paired comparisons analysis, where players represent products and the beating relation expresses the preferences of a consumer, search engines, sports and other domains. Bipartite to...[more]
Joshua Davis	AI chatbot to provide guidance and reassurance to diabetic eye screening patients	Stuart M Allen	Hantao Liu	2526-CM3203
Following diabetic eye screening, patients receive formal results letters that can be hard to interpret. For example, the overall outcome may be that no changes have been seen, some changes were observed but don’t require any further action at this time, or that the changes noticed require follow up and a further appointment with ophthalmology. The letters contain very little other information so it leaves patients stressed and anxious about whether they will lose their vision. The aim of this...[more]
Joshua Noble	Developing a 3D Interactive Mobile Game to Promote Healthier Food Alternatives for Children Aged 7-12	Nervo Verdezoto Dias	Carolina Fuentes Toro	2526-CM3203
Project Proposal: This project aims to design and implement an interactive application which promotes healthy eating practises at home. We will use a human-centred design approach, alongside human-computer interaction techniques to create a software product which will fulfil our requirements (Roccetti, Prandi, Mirri, Salomoni, 2020). Web and mobile applications for supporting self-care and health concerns have become increasingly popular for children; there were more than 325,000 mobile health...[more]
Justin Woodham	Automatic Text Simplification of Public Administration Texts with Large Language Models	Fernando Alva Manchego	Amir Javed	2526-CM3203
Automatic Text Simplification (ATS) aims to make texts easier to read while still preserving their meaning. Large Language Models (LLMs) have become the predominant approach for this task, yet most work has focused on a few domains, such as Wikipedia, news and medical texts. The Public Administration (PA) domain, covering legal notices, council documents, and online government portals which citizens access daily, has been largely left unexplored, especially in languages other than English. This ...[more]
Kacper Kruszewski	Music player designed for people with dementia	Stuart M Allen	Dr Soumya Barathi	2526-CM3203
Dementia encompasses a variety of conditions that degrade cognitive abilities, including problem-solving and memory. This, in turn, makes everyday tasks, such as operating a music player, more difficult in comparison to an able-bodied person. This project aims to tackle this by simplifying operation of a music player for the patient but still allowing more complex operations to occur remotely. This project utilises a remote solution, providing Python files to interface with the device. Operatio...[more]
Kathleen Townend	Evaluating the Efficacy of Language Models in Automating Semantic Analysis and Categorisation of Social Work Doctoral Theses.	Stuart M Allen	Hiroyuki Kido	2526-CM3203
Large Language Models (LLMs) have demonstrated remarkable capabilities in understanding and processing complex text and have recently shown promise in tasks of extracting information from unstructured and semi-structured documents and records. This can be used to summarise and analyse literature, which is beneficial to researchers as it reduces cognitive load by allowing them to judge the relevance of theses quickly, making common themes and methods stand out. The aim of this project is to addr...[more]
Kathryn MacKenzie	User-Facing Explainable AI in Automated Loan Approval Systems	Alexia Zoumpoulaki	Mohammad Taher Pilehvar	2526-CM3203
In financial settings, like loan evaluation, it is essential that the machine learning models used, are highly accurate to avoid financial stress for lenders and applicants. More complex models are needed to analyse non-linear patterns that commonly occur in financial data. Such models are more opaque, so present a greater barrier to explainability. Lack of transparency creates issues surrounding fairness and conflict with current EU AI legislation, when used to evaluate people for access to se...[more]
Kaviya Sundarapalan	AI to support adult social care	Stuart M Allen	Jing Wu	2526-CM3203
Increasing demand and cuts in funding for social care are posing significant challenges to the sector. Artificial intelligence offers the potential to address these challenges, and build smarter, more agile, and more effective social care services that support and enhance human relationships. The Centre for Social Care and Artificial Intelligence Learning (SCALE) is a new collaboration between researchers in Computer Science, CASCADE (the Children's Social Care Research and Development Centre) a...[more]
Kelvin Mahaja	Student Social Scheduling: Designing an Automated Solution to Reduce Coordination Barriers	Nervo Verdezoto Dias	Federico Liberatore	2526-CM3203
This project will investigate how scheduling difficulties impact university students' social lives and whether an automated group availability tool could reduce these coordination barriers. Many students struggle to coordinate group social plans due to conflicting schedules, leading to frustration, wasted time, and missed opportunities for connection. What should be simple often turns into never-ending group chat messages and plans that fall through. Students want to spend time with friends, but...[more]
Kian Shah	Escape Rooms to Teach Cyber Security	Catherine Teehan	Carolina Fuentes Toro	2526-CM3203
Using a variety of security and forensic concepts, we are looking to develop an 'escape room' style crime scene suitable for teaching a range of cyber security and digital forensic techniques. The room will be designed to challenge players using cyber security techniques to help develop cyber security knowledge and skills.[more]
Kipp Batchelor	Natural Language Processing (AI) for Automated Fact Verification	Nedjma Ousidhoum	Yuhua Li	2526-CM3203
Natural Language Processing is part of multiple automated fact-checking systems and related tasks such as rumour detection. It can be helpful at different steps in the verification process such as when identifying claims, retrieving evidence, etc. In this work, the student will choose the fact verification sub-task (e.g., claim identification, evidence retrieval, etc.) they would like to focus on and study the related literature. They will then implement an NLP or an Information Retrieval (IR)...[more]
Lewis McClure	Teaching cyber security skills using Minecraft for Education	David J Humphreys	Carla Perez Almendros	2526-CM3203
Create a game called Code Quest using 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. Taken from Catherine teehan but wasn’t available f...[more]
Linus Krog	Security Vulnerability Analysis and Exploitation of IoT Devices	George Theodorakopoulos	Hiroyuki Kido	2526-CM3203
Consumer IoT devices such as IP cameras pose significant security risks, especially once they have reached end-of-life where they stop receiving firmware updates. Despite these devices being implemented within homes and businesses, they are overlooked by current security standards, as they only target devices entering the market. This project addresses that gap by penetration testing the TP-Link NC200 IP camera (firmware 2.1.13, end-of-life), using PTES as the primary methodology, with OWASP IoT...[more]
Louise Igarta	Designing an Interactive Application to Promote Healthy Ageing among Older Adults at Home	Nervo Verdezoto Dias	Nitesh Kumar	2526-CM3203
Prerequisites: -Basic knowledge of Human-Computer Interaction and user-centered and/or participatory design methods -Experience developing mobile and/or web applications in several platforms (e.g., Android, HTML5 & Javascript, etc.) -Desirable experience with Spanish (not mandatory) Digital health technologies are becoming increasingly prevalent to support and enhance the everyday experiences of older adults. However, their potential to promote healthy ageing practices has been limited. T...[more]
Lucas Von Mullen Brown	Extension of AI Town Research on Social Engineering Attacks	Eirini S Anthi	Irena Spasic	2526-CM3203
Project: Goal-Directed Planning and Navigation for Adversarial Agents in Social Engineering Simulations This project extends AI Town, an open-source multi-agent simulation built on Convex with Llama 3.1 as the underlying language model, to study how navigation and planning capabilities affect an adversarial agent's behaviour in a social engineering scenario. A single agent (Darren) is tasked with extracting a password held by a specific defender (Harry) from within a population of twelve worker...[more]
Lucy Dangerfield	Escape Rooms to Teach Cyber Security	Catherine Teehan	Carolina Fuentes Toro	2526-CM3203
Using a variety of security and forensic concepts, we are looking to develop an 'escape room' style crime scene suitable for teaching a range of cyber security and digital forensic techniques. The room will be designed to challenge players using cyber security techniques to help develop cyber security knowledge and skills.[more]
Luke Navarra	Code Quest - An Educational Adventure Game Teaching Cybersecurity To Students Under The Welsh Curriculum	Catherine Teehan	Matthew J W Morgan	2526-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]
Luke Pinson	Using ML to detect cyber attacks in IoT/ICS systems	Eirini S Anthi	George Theodorakopoulos	2526-CM3203
This project investigates the use of machine learning techniques for detecting cyber attacks in Internet of Things (IoT) and Industrial Control System (ICS) environments. It evaluates supervised learning models, including Logistic Regression, Decision Tree and Random Forest, alongside an anomaly-based Isolation Forest approach and a simple hybrid fusion strategy. The project uses the ToN-IoT and HAI datasets to compare model performance across controlled benchmark data and more realistic ICS-sty...[more]
Matteo Goldsmith-Moulin	AI Generation of Music Videos	Bailin Deng	Paul L Rosin	2526-CM3203
Recent advances in AI has made it possible to generate images and videos from text prompts (for example, see https://github.com/yzhang2016/video-generation-survey/blob/main/video-generation.md for a list of video generation models). 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. Assuming we know the lyr...[more]
Matthew Purvis	Evaluating a Virtual Environment for Simulating Lateral Movement Originating from IIoT Devices in Industrial Networks	Elaine Haigh	Dr Soumya Barathi	2526-CM3203
This project evaluates the use of virtualisation in industrial test environments by implementing a virtualised testbed for simulating lateral movement. A reproducible testbed based on the Purdue Enterprise Reference Architecture was designed and implemented using KVM, QEMU, and Open vSwitch. Meanwhile, emulators simulate the behaviour of operational devices within the testbed, and the additional inclusion of an attack platform enables end-to-end observations of isolated simulations. An evaluati...[more]
Matthew Russell	A video game to raise cybersecurity awareness	Eirini S Anthi	Amir Javed	2526-CM3203
General IoT users lack the knowledge and awareness of Cyber security to adequately defend themselves from various attacks, for example, viruses and ransomware. This project aims to provide an interactive learning environment, in the form of a game, about various aspects of Cyber security, for example how to identify phishing attacks. The importance of this project is to aid in spreading cyber security awareness across general IoT, in an entertaining and interactive manor. Furthermore, this proje...[more]
Michael Drury	Agent framework backend for Teaching assistant	Fernando Alva Manchego	Nedjma Ousidhoum	2526-CM3203
I would like to propose a project which would eventually become a web application which holds an agent framework in the backend to eventually become an aid/ assistant to help teachers. I am originally considering primarily primary school teachers but am willing to look into creating the assistant also for secondary school teachers. The application would have a collection of capabilities in order to help teachers with the non teaching work that is involved behind the scenes of teaching including...[more]
Mingyang Hu	Occlusion-aware Crowd Pose Estimation on image	Wei Zhou	Hiroyuki Kido	2526-CM3203
Scope: Occlusion among pedestrians is very common in crowds, which greatly impairs pose estimation and subsequent pedestrian behavior understanding. This task mainly aims to improve the performance of the pose estimation model in crowded scenarios by constructing a interaction based occlusion-aware pose estimation model. Objectives: - Optimize AlphaPose’s detection post-processing process by introducing an adaptive NMS strategy, reducing the overlap and missing rates of human detection boxes ...[more]
Morgan Diment	Solving Battleship via ILP	Jandson Santos Ribeiro Santos	Dr Soumya Barathi	2526-CM3203
This dissertation compares Cell-based and Ship-based Integer Linear Programming formulations for solving the NP-Complete Battleship Solitaire puzzle. A Python and Gurobi benchmark across grids up to 30×30 isolates matrix density, not search-tree depth, as the dominant performance factor.[more]
Muhammad Mirza	Detecting and Explaining Bias in Social Media Posts	Nedjma Ousidhoum	Wei Zhou	2526-CM3203
This project investigates the use of Natural Language Processing (NLP) and machine learning techniques to detect and explain social bias in short-form social media posts, specifically from platforms such as X (formerly Twitter) and LinkedIn. The study focuses on identifying gender, nationality, and profession-based bias through the development and evaluation of multiple detection systems, including rule-based methods, transformer-based neural models, and a hybrid architecture named PRISM (Probab...[more]
Nathaniel Bradshaw	Algorithmically Generating MIDI Music from a Sample	Crispin Cooper	Christopher Wallbridge	2526-CM3203
With the rapid development of artificial intelligence, multiple specialised machine learning tools have emerged, capable of generating photos, 3D models and music. These models can accurately generate new pieces of media but fall short when being used to help fuel the creative process. From the perspective of an artist, an AI-powered tool can take away from their own creative expression, especially within the realm of music, where small flairs in notes, timing and dynamics can separate individua...[more]
Noor Raza	Deploying a Calibrated Hybrid Ensemble for Explainable Online Recruitment Fraud Detection	Fernando Alva Manchego	Carla Perez Almendros	2526-CM3203
Online recruitment fraud has become one of the fastest-growing categories of cybercrime, with reported losses rising rapidly and generative AI lowering the cost of producing convincing fraudulent listings. Existing defences leave four gaps: contextual signals from outside the listing are rarely incorporated, probability calibration is largely ignored, explainability is acknowledged but not addressed, and end-to-end deployment as a usable tool remains theoretical. This dissertation presents a cal...[more]
Pinhan Chen	AI-Generated Image Quality Assessment	Wei Zhou	Fernando Alva Manchego	2526-CM3203
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]
Piyumi Thotabaduge	URL Inspection Tasks: Helping Users Detect Phishing Links in Emails	Neetesh Saxena	Alexia Zoumpoulaki	2526-CM3203
Phishing is one of the most common and effective cyber-attacks, with email being the main method used to deliver malicious links to users. These attacks often rely on deceptive URLs that appear legitimate but redirect users to fake websites designed to steal information. While spam filters exist and blacklist based systems can block known threats, they are less effective against new phishing URLs and often provide little explanation for why a message is flagged. My project investigates the auto...[more]
Rhydian Brown	Implementing a System for 'Tournament' Peer Grading	Richard Booth	Yazmin Ibanez Garcia	2526-CM3203
Peer grading is an increasingly popular way to grade students' work, especially in MOOCs (Massive Open Online Courses). The idea is that each student grades the work of a subset of the other students, and from this information a final overall ranking of the students is calculated. Intuitively we want the opinion of students that are ranked more highly to carry more weight than those students ranked lower down, since they, presumably, are better qualified to recognise good quality work. What mak...[more]
Rhys Lewis	Urban Accessibility Analysis Using Street Networks	Padraig Corcoran	Tingting Li	2526-CM3203
Develop a tool to evaluate accessibility in urban areas by modeling street networks as graphs and computing metrics like shortest path distances to key amenities (e.g., schools, hospitals, parks). Use OpenStreetMap (OSM) data to fetch networks, apply graph algorithms for betweenness centrality to identify bottlenecks, and visualize heatmaps of accessibility scores. Implement in Python with OSMnx for network retrieval, NetworkX for analysis, and Folium for interactive maps. Test on multiple citie...[more]
Rohan Khakhar	Constraint based meal planning system	Matthew J W Morgan	Bailin Deng	2526-CM3203
Using multiple algorithms on users: goals, allergies, cooking ability , budget etc. a weekly meal plan can be made tailored for the user providing real value to those who are not as knowledgeable in nutrition, keeping them healthier and motivated to cook in a cost of living crisis.[more]
Saad Ullah	Medical image processing - lesions	Paul L Rosin	Stefano Zappala	2526-CM3203
This project will use a large database of images of skin lesions and attempt to perform automatic recognition of melanomas (cancer). The first step will be to segment the lesions, i.e. locate the region of interest in the image. This will be done by using rules to combine several techniques. Next, the regions are described by their shape , colour and texture to identify whether the lesion is malignant or not. The project could be based on techniques from the following paper: H. Ganster, A. Pi...[more]
Salmah Abdullahi	Design and Evaluation of an Interactive AI-Based Public Engagement Tool to Support Non-Expert Understanding of AI in Social Care	Stuart M Allen	Carla Perez Almendros	2526-CM3203
Artificial intelligence (AI) is increasingly being explored as a means of supporting social care; however, limited public understanding of how these systems operate presents challenges for responsible implementation. This project develops an interactive web-based application designed to support non-expert understanding of AI within public-facing settings. The system enables users to explore key aspects of AI through a sequence of short, interactive activities that demonstrate how outputs are ge...[more]
Samuel Stanford	Addressing Gym Progression Inconsistency Through a Tracking Application	Fernando Alva Manchego	Christopher Wallbridge	2526-CM3203
Many gym goers struggle to maintain consistent training progression, without a clear way to track and review their workouts. Current applications are either too basic or too complex, leaving a gap for an accessible, data driven training application. The project is a cross platform mobile gym tracking application, built with React Native and Expo. This project uses offline-first with SQLite so it’s in a gym environment without internet connection. A rulebased recommendation engine analyses the ...[more]
Sanay Shah	Designing an Interactive Application to Promote Healthy Bedtime Routines and Remote Family Bonding	Nervo Verdezoto Dias	David J Humphreys	2526-CM3203
Travel and separation with the parent and child can disrupt these critical routines. Remote interactions are effective for shared activities with children however children lack the attention span to sit still and just talk to a face on the screen effectively disrupting their routine [1]. Digital communication technologies are prevalent but often fail around families separated by work [3]. This means families need synchronous and asynchronous strategies for their children's bedtime routine, not j...[more]
Sara Alrifaie	Resilience Assessment of Street Networks to Disruptions	Padraig Corcoran	Wei Zhou	2526-CM3203
Design a system to assess how street networks withstand disruptions like floods or earthquakes, by simulating edge removals and measuring connectivity loss (e.g., via giant component size or detour lengths). Source network data from free repositories like OpenStreetMap, and incorporate vulnerability layers from elevation models. Use Python with OSMNX and NetworkX for robustness algorithms and spatial visualization. Test on case studies (e.g., coastal cities prone to flooding), evaluate recovery ...[more]
Scott Sandbrook	An Exploratory NLP and Machine Learning Feasibility Study for the Personality Analysis of Fictional Characters from Screenplay Dialogue and Behaviour	Walter Colombo	Wei Zhou	2526-CM3203
This project investigates the application of NLP and machine learning to produce computational estimates of the Big Five personality traits and emotions contained within the dialogue and behavioural descriptions of fictional characters in film and television screenplays. A system was developed as an exploratory feasibility study to assist individual screenwriters in evaluating how characters resonate with audiences through their narrative portrayal. By integrating NLP models and sentiment analys...[more]
Sebastian Latham	Feasibility vs. Optimality: The Impact of Objective Function Formulation on ILP Performance in Numberlink	Jandson Santos Ribeiro Santos	Víctor Gutiérrez Basulto	2526-CM3203
Zig-Zag Numberlink is more than just a puzzle, it is an abstraction of the spatial routing challenges faced in real-world domains such as Multi-Agent Pathfinding (MAPF) and Very Large Scale Integration (VLSI) design. Traditionally, Zig-Zag Numberlink has been treated purely as a Constraint Satisfaction Problem (CSP), resulting in the use of highly optimised Boolean logic frameworks such as Boolean satisfiability (SAT) or Zero-suppressed Binary Decision Diagrams (ZDD). While these approaches outp...[more]
Shangsi Lv	AI-Generated Image Detection	Wei Zhou	Carolina Fuentes Toro	2526-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]
Sultaan Hassan	Artificial Intelligence and Semantic Risk: A Comparative Study of Word Embeddings in Codenames	Federico Liberatore	Alexia Zoumpoulaki	2526-CM3203
This dissertation looks at whether a simple, interpretable scoring function over pre-trained word embeddings can perform well at a cooperative language game, Codenames, and how much the choice of embedding matters. Three research questions are addressed: does risk-aware scoring beat a naive similarity-only baseline, does the choice of embedding affect performance, and does a scoring function calibrated for one embedding still work for others. The project builds a spymaster agent that works with...[more]
Tarrin Curtis	Balancing Cost, Latency, and Capacity: Network Optimisation for High-Frequency Trading	Richard Booth	Alexia Zoumpoulaki	2526-CM3203
High-frequency trading (HFT) firms rely on colocated servers to minimise latency when sending orders to exchanges. However, leasing cross-connects and maintaining multiple links across the globe can be costly. Firms must carefully balance trade-offs between latency, bandwidth, and financial cost when designing their network. This project formalises the network design problem as a combinatorial optimisation task: selecting which cross-connects to lease and routing expected order flows to maximise...[more]
Taylor Hughes	Social Engineering AI: Evaluating the Vulnerability of Conversational Robots	Christopher Wallbridge	Nedjma Ousidhoum	2526-CM3203
Due to the increase in AI/ Automated systems, I want to make a study explaining how social engineering (an already prolific hacking tool) will in future become one of the only ways to get into system. Manipulating humans as well as circumventing automated bots and bypassing their in-built rules to gain information you are not permitted to access. This dissertation will go over the possible vulnerabilities in conversational robots and how change to systems like customer service could provide a se...[more]
Thomas McCloy	Enabling Faster Acquisition of Quantitative Susceptibility Mapping through Deep Learning-Based Super-Resolution	Stefano Zappala	Mohammad Taher Pilehvar	CMT400-Jan-Start
Background: Quantitative Susceptibility Mapping (QSM) is a highly powerful sought-after radiological technique, due to its unique sensitivity to iron, myelin and venous blood. Wider clinical adoption is hindered by the long acquisition times required to achieve the necessary high-resolution scans that can visualise the necessary small anatomy, such as venules. Aim: To develop, train and validate a single-modality framework that is capable of super-resolving low resolution QSM scans to high-reso...[more]
Tobias Gadsby	Fuzzy Similarity Matching of different data points for various personas with a real industry use case.	Federico Liberatore	Oktay Karakus	2526-CM3203
I currently have a continuing role from my placement at Lloyds Banking Group, working in Economic Crime Prevention. Specifically I work in List Screening, which involves developing a system for fuzzy matching of customers against various Economic Crime Controls. My proposal is to compare the efficacy of fuzzy matching scenarios on large customer datasets (80M+). I will explore a variety of fuzzy matching techniques, including various lexical similarity algorithms, geographical matching, semantic...[more]
Vivian Quintero Rubiano	SICRAF: Fabrication-Origin Threats in SiC/GaN Critical National Infrastructure	Shancang Li	Paddy Slator	2425-CMT400
This project examines fabrication-stage cyber-physical threats in SiC/GaN devices critical to CNI sectors including HVDC grids, radar systems, and EV traction inverters. It proposes the SICRAF framework—a five-layer model (identification, classification via 5D taxonomy, IEC 60812 risk assessment, stress-forecasting, mitigation)—validated through five scenarios drawn from verified supply-chain incidents (2021-2025). Grounded in PRISMA literature review (n=82, 2018+), the work aligns with NIS2...[more]
William Davies	To what extent can natural phenomena be modelled as fractals using computational methods?	Oktay Karakus	Paul L Rosin	2526-CM3203
To build a cohesive set of generative models that demonstrate how fractal processes underlie and unify patterns across biological, atmospheric, and geological systems. The project integrates multiple fractal types (branching, field/noise, stochastic walks) into a comparative visual framework.[more]
William Godfrey	Attention-Based Sequence Models for Security Patch Detection Using Behavioural Data	Neetesh Saxena	Oktay Karakus	2526-CM3203
It is a common practice that once a vulnerability is detected and fixed, the developers behind the software issue a Common Vulnerabilities and Exposures or CVE record to alert the user community of the security hazard and urge them to integrate the security patch. However, some companies might not disclose their vulnerabilities and just update their repository. As a result, users are unaware of the vulnerability and may remain exposed. In this work, we aim to develop an approach for the system t...[more]
Ying Lau	A Comparative Analysis of Machine Learning and Deep Learning Techniques for Short-Term Stock Price Prediction	Xianfang Sun	Carla Perez Almendros	2526-CM3203
This project investigates the application of machine learning and deep learning to short-term stock price prediction. Specifically, it conducts a controlled comparative study of four approaches — XGBoost, a one-dimensional Convolutional Neural Network, a Long Short-Term Memory network, and a novel parallel CNN-LSTM Hybrid architecture — for next-day stock return direction prediction across five stocks: Apple, Microsoft, Tesla, NVIDIA and JPMorgan Chase, covering the period 2011 to 2026. The ...[more]
Yonathan Wondimu	2D Image Registration to Capture Brain Tissue Deformation	Stefano Zappala	David J Humphreys	2526-CM3203
Brain shift is the deformation of the brain under the effect of gravity - the whole tissue can displace similar to a gel-like material when changing the orientation of the head. It can significantly affect the outcome of neurosurgery, as targets for deep brain stimulation or biopsy are planned beforehand and leads implanted blindly with the only help of imaging. Information on the potential displacement of these targets is critical. There is lack of evidence about the early stages of such defo...[more]
Zahirah Shahrul Azman	Detecting AI-Generated Images	Bailin Deng	Frank C Langbein	2526-CM3203
The democratisation of AI image generation has enabled people to create images with just a simple text prompt. Although beneficial in particular areas such as creative processes or business marketing, it has also given way to more harmful and dangerous acts such as identity theft, the viral spread of misinformation and disinformation, and the non-consensual creation of deepfake content. The expansion from GAN-based to diffusion-based image generative models adds an additional dimension to the pr...[more]
Zainah Aljarrah	Detection and Security Analysis of Malicious Websites	Amir Javed	Alexia Zoumpoulaki	2526-CM3203
The approach uses a crawler to extract metadata and combines security-focused analysis with machine learning–based classification to identify harmful activity and present the factors contributing to detection results in a clear and understandable manner. The project aims to support efficient security decision-making by focusing on the presentation and interpretation of classification outcomes alongside detection accuracy. It aligns with current academic and industry objectives and addresses a ...[more]
Zakaria Attalah	Industry-Based Project - Cyber Security and OSINT Awareness Games	Catherine Teehan	Bailin Deng	2526-CM3203
This project is collaboration with an external organisation. The aim of this project is to create a set of interactive games that raise cyber security and OSINT awareness in an engaging way. The desired outcome is a functional, educational tool that teaches a broad and possibly non-technical audience, to identify cyber threats and OSINT risks such as phishing, social engineering, malware and insider threats. The benefit is both a resource to promote cyber resilience in the wider community and pr...[more]
Ziad Arafeh	Investigating whether news sentiment analysis improves machine learning model accuracy for the next day stock price direction prediction	Yuhua Li	Richard Booth	2526-CM3203
This project investigated whether incorporating news sentiment analysis into machine learning models improve the accuracy of next day stock price direction compared to models using technical indicators. Three machine learning models from distinct architectural families were evaluated: XGBoost, LSTM, and the Transformer. Each was trained on three actively traded stocks (Apple, Tesla, and Amazon) under two feature configurations with a baseline using only technical indicators, and an extended set ...[more]
Zi Liang Woon	Comparative Evaluation of NLP Approaches for Phishing Email Detection: Classical ML, Deep Learning, Transformers, and Semantic Hashing	Neetesh Saxena	Alexia Zoumpoulaki	2526-CM3203
Phishing remains the most prevalent cybersecurity threat facing organisations today, driven increasingly by AI-generated content that bypasses traditional detection systems. Existing research on NLP-based phishing email detection suffers from persistent methodological gaps: methods are rarely compared under consistent, controlled conditions, and critical deployment metrics such as false-positive rate and inference time are seldom reported alongside accuracy. This project conducts a comparative...[more]