Dimensions overall score 5.0

GitHub Code Pulse

No public code linked for this paper yet.

• Current multi-modal fusion methods in brain networks are static and feed different samples into the same model with identical computation.limitation 90%
• The proposed M3D-BFS is a multi-stage dynamic fusion strategy for sample-adaptive multi-modal brain network analysis.method 95%
• M3D-BFS uses mixture-of-experts (MoEs) where modules can adaptively change as the input sample changes during inference.technical 90%
• The method is divided into 3 training stages to alleviate the issue of MoE training collapse.method 90%
• A multi-modal disentanglement loss is designed to enhance the final representations.technical 90%
• This is the first work for dynamic fusion for multi-modal brain network analysis.result 85%
• Extensive experiments on different real-world datasets demonstrate the superiority of M3D-BFS.result 80%
• Multi-modal fusion can achieve better performance than uni-modal methods in downstream tasks.technical 70%

Competitive landscape

Competitor map is still being generated for this paper. Enable generation or check back soon.

Startup potential card