Bio-Behavioural and Perceptual Approaches to Enhancing Deepfake Analysis 

Supervised by Professor Jennifer Cook and Dr Hélio Cuve together with the National Police Chief Council (NPCC) 

To apply for this project, please include ‘Cook & NPCC’ as the project descriptor in the subject heading of your email. 

Application deadline: 29 July 2024; interview on 5 August. 

Deepfakes are synthetically generated media such as audio, images, and videos created using generative AI to mimic or simulate plausible human features and behaviour (faces, bodies, movement, voices). As deepfakes become increasingly sophisticated, they pose significant challenges to society due to their potential misuse for criminal activities, including fraud, impersonation, and misinformation. Deepfake detection is currently an open research area and a moving target, with new deepfake generation tools often surpassing current detection capabilities of both humans and algorithms. Furthermore, current detection models lack explainability, making it difficult to assess their reliability and potential biases, which is crucial for deployment in real-world high-stake contexts such as policing. 

Project Description

This PhD project aims to develop explainable deepfake detection tools by identifying robust multimodal bio-behavioural “fingerprints” (e.g., face and body movement patterns, physiological cues). This project will build on the extensive expertise of the supervisors in these areas, including the development of novel algorithms to track and analyse statistics in facial expressions and body movements.

The PhD student will:

• Conduct literature reviews to synthesise relevant multidisciplinary literature (e.g. from psychology and computer science).
• Design and run lab-based and online experiments combining a range of behavioural and biometric analysis measures (e.g., eye-tracking, face-tracking, video-based biometrics).
• With the support of the supervisory team, create novel detection models that could be used as screening tools for deepfakes.
• Undertake a placement with the NPCC to test real-world use-cases relevant for policing priorities and engage with stakeholders (e.g. dissemination, seminars)

About our lab

Our lab is a vibrant and friendly research community. We have regular hybrid lab meetings, catch-up meetings, and socials. We strongly believe in supporting each other and working through problems together. The supervisory team spans the University of Birmingham, University of Bristol, and NPCC. We are dedicated to providing a broad range of training and professional development opportunities to ensure that you get the most out of your PhD. Moreover, we will support you to present your findings at national and international conferences, publish your results in academic journals, and conduct public engagement activities. 

Candidates with degrees in Psychology, Neuroscience, Computer Science, and related fields, are particularly encouraged to apply. While having an MSc is desirable, it is not a requirement. The ideal candidate will also demonstrate strong analytical and communication skills, the ability to collaborate effectively, and experience or interest in data-driven methods (e.g., machine learning, computational methods), programming for research and data analysis for cognitive science (e.g., R, Python, MATLAB).  

Eligibility

We are looking for a motivated, talented, and dedicated PhD student with a 1st class or 2:1 degree in Psychology, Neuroscience, Computer Science, or related fields. An MSc degree in a relevant area is desirable though not necessary. The ideal candidate will also demonstrate strong statistical and analytical skills, experience in conducting quantitative research, and communication skills, the ability to collaborate effectively, and experience or interest in data-driven methods (e.g., machine learning), programming for research and data analysis for cognitive science (e.g., R, Python, MATLAB). The position is open to home students and require residency within the UK of over 3 years. Candidates will be subject to police vetting. 

Prospective candidates should express their interest via email, directly to the both supervisors: Professor Jennifer Cook (J.L.Cook@bham.ac.uk) and Dr Hélio Cuve (helio.cuve@bristol.ac.uk) and notify them of the final submission.