arXiv:2606.11833v1 Announce Type: new Abstract: Flow matching and diffusion models enable conditional generation across domains ranging from images to proteins, with recent extensions to out-of-distribution contexts. Yet generative models of neural time series have largely remained restricted to categorical conditioning, precluding compositional and zero-shot generalization. In this work, we propose a per-timestep conditioned diffusion transformer for generating realistic fMRI brain dynamics during unseen cognitive tasks by injecting both compositional language and optional spatial priors in-c
The rapid advancements in generative AI, particularly diffusion models, are enabling increasingly sophisticated applications in complex data domains like neuroscience.
This development represents a significant step towards generating realistic and interpretable brain dynamics, which could revolutionize neuroscience research, clinical diagnostics, and AI-driven control systems for neurotechnologies.
Generative models for neural time series can now move beyond categorical conditioning to incorporate compositional language and spatial priors for out-of-distribution scenarios.
- · Neuroscience researchers
- · AI developers in medical imaging
- · Personalized medicine initiatives
- · Generative AI platforms
- · Traditional fMRI analysis methodologies that lack generative capabilities
- · Pharmaceutical companies relying solely on animal models for neurological resear
Improved understanding and synthetic generation of brain states corresponding to cognitive tasks.
Accelerated development of neuroprosthetics and brain-computer interfaces by providing realistic training data.
Potential for creating 'digital twins' of individual brains for hyper-personalized medical interventions and simulations.
This signal links to a primary source. Continuum Brief monitors and indexes it as part of the live intelligence stream — we do not republish source content.
Read at arXiv cs.LG