Recurrent neural networks (RNNs) are a key type of architecture in modern deep learning, particularly for processing sequential data such as text, speech, video, and time series data. Unlike feedforward networks, RNNs have loops that allow information to persist and be passed from one step to the next. This enables them to effectively model patterns … full description “Exploring optimization algorithms for recurrent neural networks (available)”
Project List
Implementation of biologically inspired efficient deep learning models (available)
As deep learning models continue to grow in size and complexity to tackle increasingly difficult tasks, the need for efficient and scalable models becomes ever more important. Extremely large language models like GPT-4 require massive computational resources and expensive hardware to train and run. This makes them impractical to deploy at scale in many real-world … full description “Implementation of biologically inspired efficient deep learning models (available)”
Algorithms for lifelong deep learning (available)
Continual learning [1], also known as lifelong learning, refers to the ability of an artificial intelligence system to continuously learn and adapt from new experiences over time. This is an important capability as it allows AI models to acquire new knowledge and skills as more data becomes available, without forgetting previously learned information. Continual learning … full description “Algorithms for lifelong deep learning (available)”
AI-powered COVID-19 detection via cough sounds (available)
Respiratory diseases such as COVID-19 are known to physically damage our airway and lungs, which in turn alters the produced respiratory sounds (e.g., cough, breadth). During the pandemic, cough classification has emerged as an accessible, low-cost, and environmentally friendly COVID-19 screening alternative, needing only a smartphone to collect and process cough samples. However, audio processing … full description “AI-powered COVID-19 detection via cough sounds (available)”
A commercial alternative to QKD: filling USB sticks and hard disks with random bits (completed)
Quantum key distribution (QKD) offers unconditional security according to the laws of quantum physics. A QKD system enables its users to securely set up symmetric keys for encryption by sending quantum signals, either over optical fibres or over free-space. In current commercial QKD systems, these symmetric keys are consumed in classical secure communications protocols like … full description “A commercial alternative to QKD: filling USB sticks and hard disks with random bits (completed)”
A simulation model of self-powered ankle prosthetics (completed)
Unlike conventional passive prosthetics, self-powered prosthetics provide positive net work to enhance strength and speed of limb movement. These are used not just in rehabilitation medicine but also as components of powered exoskeletons, a technology with numerous applications including military, manufacturing and construction industry, emergency and rescue. The extent of powered propulsion is typically determined … full description “A simulation model of self-powered ankle prosthetics (completed)”
Algorithmic Game Theory and Computational Social Choice (completed)
This project will be co-supervised with Eduard Eiben. We will study a topic arising from Algorithmic Game Theory and/or Computational Social Choice. The specific topic/area will be defined during an inital meeting with Eduard and myself. Some examples of such topics are (this list is by no means exhaustive):– Computation of Nash Equilibria;– Fair Division … full description “Algorithmic Game Theory and Computational Social Choice (completed)”
Attacking Large Pre-trained Programming Language Models (PLMs) via Backdoors (completed)
Project Description: Backdoors refer to a class of Machine Learning (ML) attacks where an adversary trains an ML model to intentionally misclassify any input to a specific label [1]. This is typically achieved by poisoning the training data, such that inputs are misclassified to a target label when the backdoor trigger is present. For instance, … full description “Attacking Large Pre-trained Programming Language Models (PLMs) via Backdoors (completed)”
Automatic Patch-based Exploit Generation (completed)
When a vulnerability is found, it generally goes through a responsible disclosure process. This means that the vendor will be contacted to write a fix before the vulnerability is made public. The patch will be pushed to its users and the vulnerability details become public. In theory, this is a safe process for vulnerability exposure … full description “Automatic Patch-based Exploit Generation (completed)”
Blockchain to provide Data Provenance Integrity and Privacy (completed)
Blockchain, also part of the cryptocurrencies, can be viewed as a potentially shared/semi-shared/private, the immutable ledger for recording sequence of events or history of transactions. The blockchain technology can be deployed to provide a high-degree of trust, accountability, and transparency associated with a set of transactions/events – especially log files and data provenance. Data provenance … full description “Blockchain to provide Data Provenance Integrity and Privacy (completed)”
Building a dataset of IoT device firmwares (completed)
Performing security analysis of IoT devices is often expensive as it requires purchasing the IoT device, which is not practical at scale. To avoid this some researchers perform security analysis over the apps used by those IoT devices. Although this analysis can be useful it has some limitations as only one element of the ecosystem … full description “Building a dataset of IoT device firmwares (completed)”
Building a Full Causality Chain Across an Enterprise System (completed)
Data Provenance refers to records of the inputs, entities, systems and process that influence data of interest, providing a historical record of the data and its origins. To provide a holistic view of the data provenance in an enterprise system, the provenance records of the activities carried out on a client workstation is important. Last … full description “Building a Full Causality Chain Across an Enterprise System (completed)”
Building Data Provenance from Database Log Files (completed)
Databases are an integral part of any organisations operations. They act as storage repositories for a large set of data that the respective organisation relies upon for their efficient operations. One of the commonly deployed and open source database application is MySQL. It collects and stores a large set of log files related to the … full description “Building Data Provenance from Database Log Files (completed)”
Control-flow Bending POCs (completed)
Research has shifted over the years when it comes to binary exploitation. With more accurate and practical implementations of Control-flow Integrity (CFI) [1] [2] [3], the question arises as to what attack surface is still available – and how to exploit it. In particular, what attacks can be performed when staying within the boundaries of … full description “Control-flow Bending POCs (completed)”
Control-flow graphs for Automatic-Exploit Generation (completed)
Research has shifted over the years when it comes to binary exploitation. With more accurate and practical implementations of Control-flow Integrity (CFI) [1][2], the question arises as to what attack surface is still available – and how to exploit it. In particular, what attacks can be performed when staying within the boundaries of a (perfect) … full description “Control-flow graphs for Automatic-Exploit Generation (completed)”
Data Provenance for Multi-Database Servers Enterprise Architecture (completed)
Enterprise architecture, with increasing frequencies, is based on multiple databases that split the enterprise data among itself and store them on separate database servers. Such a scheme enables an effective load balancing and management of enterprise data. However, my splitting data over multiple databases make it challenging to build a unified data provenance view of … full description “Data Provenance for Multi-Database Servers Enterprise Architecture (completed)”
Decentralised Autonomous Economy of Things – Putting rational actors back into Economy (completed)
Project Description Internet of Things (IoT) and distributed ledger technologies are part of the movement to develop the device-specific decentralised economy. Example of such a service is decentralised energy grids that couple together the smart contracts and smart-meter technology to provide traceability and verification of energy sources, efficient peer-to-peer trading of energy. One such proposal … full description “Decentralised Autonomous Economy of Things – Putting rational actors back into Economy (completed)”
Decentralised Cloud – Building an All-Purpose Execution Platform on Blockchain (completed)
Project Description Blockchain and distributed ledger technology have invigorated the decentralised architectures. The positives of a decentralised architecture are that it is distributed widely and challenging to launch a DDoS against such a network. Besides that, the immutability and strong traceability/trackability of blockchain technologies can be considered as a foundation for designing a secure, reliable, … full description “Decentralised Cloud – Building an All-Purpose Execution Platform on Blockchain (completed)”
Detecting broken security in hybrid Android apps (completed)
Many modern Android applications make use of a webview – a component providing easy access to the rendering engine and JavaScript interpreter of a full browser. The content shown by a webview can be loaded from a local resource or a remote server via HTTP and integrates seamlessly with the app. Webviews are popular with developers, … full description “Detecting broken security in hybrid Android apps (completed)”
Disassembling x86 binaries for static analysis and reverse engineering (completed)
The Jakstab static analyser for binaries automatically disassembles x86 binaries for Windows or Linux and reconstructs a control flow graph. It is particularly effective on targets that have been obfuscated with various tricks that throw off regular disassemblers such as IDA Pro. Jakstab disassembles one instruction at a time, translates it into an intermediate language, and then … full description “Disassembling x86 binaries for static analysis and reverse engineering (completed)”
Eclipse plugin for the ART parser generator (completed)
The ART parser generator is one of a new breed of compiler generation tools which provides efficient generalised parsing. This means that language designers have complete freedom to specify syntax in a way that supports downstream processes rather than having to shoehorn their ideas into the constraints imposed by current near-deterministic parser generators like Bison … full description “Eclipse plugin for the ART parser generator (completed)”
Entry into the PACE Parameterized Algorithms and Computational Experiments Challenge (completed)
Parameterized Complexity is a research field that, by its own self-description, strives to provide practical, yet theoretically well-founded ways to deal with computationally hard problems (e.g., so-called NP-hard problems). However, the vast majority of the work in the field is purely theoretical — there is a great toolbox of interesting and powerful algorithmic methods, which have been proven to have … full description “Entry into the PACE Parameterized Algorithms and Computational Experiments Challenge (completed)”
Formal semantics and C (completed)
When we want to know how a programming language construct such as Java exception handling works, we look it up in the language manual or a textbook. This is good enough for most programmers since they can test their understanding by compiling short programs and watching their effects in a debugger. However, compiler developers have … full description “Formal semantics and C (completed)”
Improving Automatic Bug Detection in JavaScript (completed)
Dynamic symbolic execution (DSE) is an effective tool for bug detection in real software. Like unit testing and fuzzing DSE executes portions of a program, exposing bugs through runtime program exceptions. In DSE, some inputs to the program under test are made “symbolic” while the rest are fixed. Whenever the symbolic execution encounters a conditional operation … full description “Improving Automatic Bug Detection in JavaScript (completed)”
Increasing the Efficiency of Low-land Natural Disaster Search and Rescue through ‘Supervised Hybrid Animal-UAV Interactions’ (completed)
Search and Rescue (SAR) is a time-critical event: after a finite period of time the chances of recovering individuals who have become lost, or trapped in the wake of a disaster, trend towards zero. As a result, management of search and rescue resources is paramount. Travel time, planning and executing search grids, are pivotal tasks … full description “Increasing the Efficiency of Low-land Natural Disaster Search and Rescue through ‘Supervised Hybrid Animal-UAV Interactions’ (completed)”
Invisible Malware using Intel SGX Enclaves (completed)
A fundamental security problem when hosting applications on cloud platforms is the increased risk of sensitive data loss (e.g. due to negligent or malicious employees of the cloud provider). An exciting approach to mitigating such attacks are new trusted execution environments (e.g. Intel SGX), recently available on commodity CPUs. Intel SGX allows users to create … full description “Invisible Malware using Intel SGX Enclaves (completed)”
Machine Learning for Crystal Structure Prediction (completed)
Crystal Structure Prediction (CSP) is one of the major problems in computational chemistry with numerous applications in real life. This is essentially the (global) minimisation of a continuous, high-dimensional, complicated function. Many heuristic methods have been proposed for CSP and recently new methods based on Machine Learning were introduced. The goal of this project is … full description “Machine Learning for Crystal Structure Prediction (completed)”
Machine Learning Library for OCaml (completed)
Frameworks for machine learning include Python’s TensorFlow [1]. These frameworks provide standard ways of specifying models that can be optimised by machine learning algorithms. OCaml [2] is a mature functional programming language with an expressive type system. DecML [3] is a prototype OCaml extension that facilitates implementing machine learning tasks, based on specifying models as … full description “Machine Learning Library for OCaml (completed)”
Mixed Nash Equilibria in Net Coordination Games (completed)
Net Coordination Games form a special class of many-player games with several applications in Theoretical Computer Sciene, Multi Agent Systems, and Semi Supervised Learning. Nash equilibria correspond to the stable outcomes and they are the prominent solution concept in games. It is known that Net Coordination Games possess a pure Nash equilibrium, but unfortunately, it … full description “Mixed Nash Equilibria in Net Coordination Games (completed)”
Penetration Testing of Autonomous Vehicle Sensor Arrays (completed)
Project Description The adoption of autonomous vehicles is a matter of when and how, rather than if it will happen. Domestic (private) and commercial (freight) autonomous vehicles are a proven concept, with most remaining questions regarding adoption hanging on the reliability and security of said vehicles. It is well-known that side-channel attacks exploiting peripheral systems, … full description “Penetration Testing of Autonomous Vehicle Sensor Arrays (completed)”
Personality Identification Using Online Social Media Networks (completed)
Project Description Personality is a combination of all the attributes—behavioural, temperamental, emotional, and mental—that characterise a unique individual. Ability to identify the personalities of people has always been of great interest to the researchers due to its importance. Owing to the increasing popularity of online social networks there is a possibility of predicting a user’s … full description “Personality Identification Using Online Social Media Networks (completed)”
Predicting Debug Symbols for Closed Source Binaries (completed)
Reverse engineering binaries, whether malicious or benign, is made more difficult by the absence of debug information. Variables and functions have had their identifiers “stripped”, so reverse engineers have to manually name them during analysis based on human understanding of the code functionality. The goal of this project is to use machine learning to predict … full description “Predicting Debug Symbols for Closed Source Binaries (completed)”
Privacy issues related to Data Provenance of Database containing End-user Data (completed)
Privacy issues related to the data stored, regarding end-users are well understood and studies. However, a large set of data can also be collected that is not directly related to the user but related to the user data – known as data provenance. Data provenance is the field of recording the history of data, from … full description “Privacy issues related to Data Provenance of Database containing End-user Data (completed)”
Real-Time Consumer Centric Data Compliance Auditing (completed)
When a user signs an end user agreement with an organisation, in return of which (s)he will give his/her data to the organisation the users do not have; a) any mechanism to assess that post-agreement the organisation abides by their own agreement and b) to see the operations carried on his/her data (that the organisation … full description “Real-Time Consumer Centric Data Compliance Auditing (completed)”