Honours and Masters project

The rapid growth of electric vehicles (EVs) is transforming the transportation systems worldwide. Both EV fleets and private EVs are emerging as a cleaner and more sustainable component of urban mobility, forming an effective way to solve environmental problems and reduce commute costs in future smart cities. Due to the complex spatiotemporal behaviors of passengers and their travel patterns, the unmanaged electric charging demand from EVs may significantly impact the existing transportation and electrical power infrastructure.

Thanks to the widespread deployment of smart meters, high volumes of residential load data have been collected and made available to both consumers and utility companies. Smart meter data open up tremendous opportunities, and various analytical techniques have been developed to analyse smart meter data using machine learning. This project will provide a new angle toward energy data analytics and aims to discover the consumption patterns, lifestyle, and behavioural changes of consumers.

Since the COVID-19 pandemic there has been an increasing shift within higher education away from traditional lectures and towards asynchronous content delivery through pre-recorded videos. This has a number of benefits: students can consume content at their own pace, videos can be reused, and production value can be increased. However, academics typically have no training or experience in video production, so pre-recorded videos are most often just a simulacrum of a standard lecture (i.e., a slideshow with voiceover).

Traditional active learning helps reduce labeling costs by selecting the most useful examples from a large pool of unlabeled data. However, in many real-world cases, such a large pool doesn't exist or is expensive to collect. This project explores a new approach using large language models to create synthetic unlabeled text data instead. Rather than just picking data to label, the model will also generate new examples that are diverse and potentially helpful for learning.

This project focuses specifically on LLM applications: chatbots used in customer support (e.g., healthcare). The goal is to investigate how user requirements (e.g., “the bot should de-escalate frustrated users”) can be systematically translated into prompt templates or prompt strategies.

This project aims to develop privacy-preserving deepfake detection techniques that enable accurate and secure identification of synthetic audio and video content without exposing sensitive user data. Traditional detection methods often require access to raw audio or visual inputs, raising significant privacy concerns, especially in scenarios involving personal or biometric data.

Mis/disinformation (also known as fake news), in the era of digital communication, poses a significant challenge to society, affecting public opinion, decision-making processes, and even democratic systems. We still know little about the features of this communication, the manipulation techniques employed, and the types of people who are more susceptible to believing this information.

This project extends upon Prof Whitty's work in this field to address one of the issues above.

This project focuses on enhancing security and privacy protection in blockchain-based systems for verifying education credentials, with the goal of combating the proliferation of fake certificates. By leveraging the immutable and decentralized nature of blockchain, the project aims to develop a secure credential verification framework that ensures the authenticity and integrity of academic records while safeguarding users' personal information.

Background and motivation

As intelligent agents make decisions, any project aiming to realize human-like AGI should model decision-making.  As we have been pursuing the WBA approach to create AGI by learning from the architecture of the entire brain, we request you to model the decision-making of the mammalian brain.

Markov Decision Processes (MDPs) are frameworks used to model decision-making in situations where outcomes are partly random and partly under the control of a decision maker. While small MDPs are inherently interpretable for people, MDPs with thousands of states are difficult to understand by humans. The K-MDP problem is the problem of finding the best MDP with, at most, K states by leveraging state abstraction approaches to aggregate states into sub-groups. The aim of this project is to measure and improve the interpretability of K-MDP approaches using state-of-the-art XAI approaches.