Pourya Shamsolmoali
Lecturer

Interests

Machine learning
Computer vision

I am looking for Full-time PhD students who are interested in computer vision and learning theory (with a focus on efficient and trustworthy machine learning). This is a fully funded, research-focused position.

Possible topics include (but are not limited to):
Multi-task Learning: Investigates how models can learn multiple tasks simultaneously, sharing knowledge effectively while maintaining strong performance on each task. The goal is to develop a theoretical framework to explain the success of multi-task learning in deep neural networks and to understand the balance between shared knowledge and task-specific challenges.

Structured Perception: Focuses on how models can learn organized representations for complex visual tasks, such as scene understanding, 3D vision, or aerial image analysis.

Fine-tuning of modern machine learning models: Aims to theoretically understand the learning of fine-tuning or devise expeditious and resource-efficient inference and fine-tuning methods for multi task or multi modal learning, enabling their deployment within constrained computational resources.