Decoding fMRI Data into Captions using Prefix Language Modeling
This work addresses brain decoding for neuroscience applications, but it is incremental as it builds on existing methods with modifications for efficiency and accuracy.
The paper tackles the problem of decoding fMRI data into image captions by predicting DINOv2 image embeddings from fMRI signals and using them as prefixes for GPT-2, reducing computational requirements, and explores 3D CNN mapping instead of Linear Regression for better voxel positional handling.
With the advancements in Large Language and Latent Diffusion models, brain decoding has achieved remarkable results in recent years. The works on the NSD dataset, with stimuli images from the COCO dataset, leverage the embeddings from the CLIP model for image reconstruction and GIT for captioning. However, the current captioning approach introduces the challenge of potential data contamination given that the GIT model was trained on the COCO dataset. In this work, we present an alternative method for decoding brain signals into image captions by predicting a DINOv2 model's embedding of an image from the corresponding fMRI signal and then providing its [CLS] token as the prefix to the GPT-2 language model which decreases computational requirements considerably. Additionally, instead of commonly used Linear Regression, we explore 3D Convolutional Neural Network mapping of fMRI signals to image embedding space for better accounting positional information of voxels.