Reconstructing Natural Scenes from fMRI Patterns using BigBiGAN

Understand

Decoding and reconstructing images from brain imaging data is a research area of high interest.

• Recent progress in deep generative neural networks has introduced new opportunities to tackle this problem.
• Here, we employ a recently proposed large-scale bi-directional generative adversarial network, called BigBiGAN, to decode and reconstruct natural scenes from fMRI patterns.
• BigBiGAN converts images into a 120-dimensional latent space which encodes class and attribute information together, and can also reconstruct images based on their latent vectors.