What do we do?

You can join the group of any of the 13 academics below. Each academic has offered a number of topics they are recruiting into.

Dima Damen: Professor of Computer Vision with research interests in egocentric vision, video understanding and fine-grained action understanding.

• Egocentric Videos in 3D
• Hand-Held Object Reconstruction
• Long-Term Memory in Egocentric Videos
• Audio-visual Action Recognition

Peter Flach: Professor of Artificial intelligence with research interests in evaluation and improvement of machine learning models, mining highly structured data, and human-centred AI.

• Classifier calibration
• Uncertainty representation and propagation
• Knowledge-intensive AI
• Explainability and interpretability

Majid Mirmehdi: Professor of Computer Vision with research interests in human and animal behaviour understanding and medical image/volume analysis

• Human action understanding and assessment, e.g. in healthcare for action quality scores
• Animal action analysis and understanding using camera-trap, drone or other footage
• Segmentation, Classification, and Prediction in medical images and volumes

Raul Santos-Rodriguez: Professor in Data Science and AI with research interests in the foundations of (human-centric) machine learning and its applications to healthcare and climate science.

• Explainability
• Evaluation
• Visual perception
• Multimodal perception

Michael Wray: Assistant Professor of Computer Vision with research interests in video understanding and Natural Language Processing.

• (Fine-grained) Vision-Language Retrieval
• Understanding Biases in Vision Language Models
• Multi-modal Video Question Answering
• Compositionality for Video-Language models

Zahraa S. Abdallah: Senior Lecturer in Data science and Machine Learning with research interest in time series and its healthcare applications and learning from multiple modalities.

• Time series analysis (classification, clustering, explainability) and dimensionality reduction
• Adaptive and active learning
• Validation of time series methods
• Multimodal learning

Telmo Silva Filho: Senior Lecturer in Data Science with research interests in evaluation of machine learning models, explainability, latent-variable models, and medical applications of computer vision.

• Explainable evaluation, i.e. when/why is a model expected to fail?
• Latent-variable models for evaluation
• Embeddings with built-in interpretability

Tilo Burghardt: Professor in Computer Science with research interests in animal biometrics, imageomics and applications of computer vision to animal welfare, farming, and conservation

• Deep learning for the detection of animal species, individuals, and morphological traits
• Recognition of animal poses, behaviours, and social configurations
• Integration of methods in computer vision, taxonomics, genetics, and ecology
• Autonomous visual navigation of conservation drones and related robotic platforms

Guosheng Hu: Senior Lecturer in Computer Science with research interests in computer vision and model acceleration.

• Accelerating Foundation Models (e.g., Large Language Models) through Quantisation, Pruning, Knowledge Distillation, etc.
• Applications of Foundation Models
• Multimodal Learning
• Face and Body Analysis

Wei-Hong Li: Lecturer in Computer Vision and Machine Learning interested in universal models, data-efficient learning, 3D-aware modeling, and generative models.

• Universal representation learning across tasks, domains and modalities
• 3D-aware Modeling and Motion
• Learning from partially annotated/paired data
• Multi-Modal Generative Modelling

Xiang Li: Lecturer in Computer Vision with research interests in multimodal large language models, 3D vision, and AI for Earth Observation.

• Multimodal Large Language Models for perception, reasoning, and spatial intelligence
• 3D perception from point clouds, monocular images, and videos
• Deep learning and foundation models for remote sensing image understanding

Mengyue Yang: Lecturer in Artificial Intelligence with research interests in causality, world models, and reinforcement learning.

• Foundation World Models (generative AI, multimodal foundation models, video understanding, RL for Game AI & Embodied AI)
• Causality
• Large Language Model Reasoning
• AI for Science (AI agents & causal techniques for chemistry, economics, and social science)

Nan Lu: Lecturer in AI with research interests in trustworthy machine learning from imperfect data, with applications in healthcare and neuroscience.

• Robust, explainable, and fair machine learning systems
• Weakly supervised learning (e.g., under corrupted labels, distribution shifts, selection bias)
• Multi-modal learning with noisy or uncertain cross-modal alignment
• Reinforcement learning from unreliable reward signals

What do we offer?

Applicants will be selected based on their academic excellence, previous experiences and provided references. We have a limited number of 15 positions throughout the program annually

You are expected to be in Bristol for the majority of your visit's duration. During your research visit, you will be supervised by your selected faculty member, and integrated within their research group. You will have access to our excellent resource of GPUs (Isambard AI) to run any experiments, a university of bristol email account with access to our PGR resources and a desk allocated for the duration of your visit.

Additionally, you will have opportunities to present your work and attend talks by the various MaVi researchers. A social program over weekends will give you the opportunity to explore Bristol and its surroundings, guided by one of our friendly PGR students. Note that any expenses for the social activities are not covered by the program. Note that you need to bring your own laptop – machines will not be provided as part of the program.