ICANN 2025 Tutorial:
Brain Visual Decoding with Generative Models
Time: TBD
Location: TBD
[Slides] [Paper List] [Recording]
Overview
Computer vison and deep learning have emerged as power- ful tools in cognitive neuroscience, particularly in the domain of brain visual decoding, which is reconstructing or interpreting visual stimuli from neural activity. This tutorial provides a comprehensive overview of how computer vision models contributes to understanding the brain’s visual processing mechanisms. We first introduce the fundamental principles of visual decoding and its significance in neuroscience and artificial intelligence. Then, we explore state-of-the-art DL models, including convolutional neural networks and transformer-based architectures, high- lighting their applications in brain imaging modalities such as functional magnetic resonance imaging (fMRI) and electroencephalography (EEG). Key challenges, including data scarcity, model interpretability, and cross-subject generalization, are discussed alongside emerging solutions such as self-supervised learning and multimodal approaches. Finally, we examine future directions, emphasizing the potential to bridge the gap between biological and artificial vision systems. This tutorial synthesizes the latest advancements and challenges in the field, with the goal of inspiring further research at the intersection of neuroscience and artificial intelligence.
Speakers
University of Cambridge
University of Cambridge
Schedule
| Title | Speaker | Time (PST) |
|---|---|---|
| Preliminary Brain encoding, neuroimaging datasets, generative models, etc. |
Weihao Xia | TBD |
| Brain Visual Decoding Generativel models, brain alignment strategies, brain encoder design, case studies, etc. |
Weihao Xia | TBD |
| Challenges and Opportunities Ethical concerns, inter-subject variability, cross-subject generalization, etc. |
Weihao Xia | TBD |
| Discussion and Q&A | Weihao Xia | TBD |
About Us
University of Cambridge
Weihao Xia is a postdoctoral researcher at the Department of Computer Science and Technology, University of Cambridge, with over ten academic papers published in leading venues in computer vision and neuroscience, seven of which are as the first author. He also serves as a regular reviewer for top AI conferences and journals in these fields.
University of Cambridge
Cengiz Öztireli is a Professor of Visual Computing and Artificial Intelligence at the University of Cambridge, the director of Core Lab, and a Research Scientist and Team Lead at Google. His research interests are in computer graphics, vision, machine learning, and artificial intelligence. He has been honored with several awards including the Eurographics Best PhD Thesis Award, Fulbright Science and Technology Award, and the UKRI Future Leaders Fellowship.