AI Paper Atlas

Research map / Vision-language and multimodal / Human and embodied modeling

Decoding Visual Neural Representations by Multimodal Learning of Brain-Visual-Linguistic Features

Changde Du, Kaicheng Fu, Jinpeng Li, Huiguang He

TPAMI 2023 · Vision-language and multimodal › Human and embodied modeling · approach: fMRI visual decoding · 151 citations (OpenAlex)

Read the paperCheck your idea against 137,217 papers

AI-extracted. The measured effects below were read by AI from the paper's full text (no summary yet). Numbers and claims can be wrong or missing context, so check the paper before you cite or build on it.

Measured effects reported in the paper

What was changedWhereMeasuredEffect
add textual features to visual featuresBraVL · DIR-Wiki+GOD-Wiki avgtop-5 accuracy (%): 55.51 -> 60.33Improved
add textual features to visual featuresBraVL · DIR-Wiki+GOD-Wiki avgtop-1 accuracy (%): 20.08 -> 22.43Improved
use BraVL instead of MoPoE-VAEmultimodal VAE decoding · DIR-Wiki+GOD-Wiki avgtop-5 accuracy (%): 58.70 -> 60.33Improved
use Wiki articles instead of class names as text input (ablation)BraVL (ALBERT) · GOD-Wikitop-5 accuracy (%): 45.47 -> 57.04Improved
use Wiki articles instead of class names as text input (ablation)BraVL (ALBERT) · DIR-Wikitop-5 accuracy (%): 70.58 -> 74.69Improved
use Wiki articles instead of class names as text input (ablation)BraVL (GPT-Neo) · GOD-Wikitop-5 accuracy (%): 36.69 -> 53.54Improved
add intra-modality MI term L_intra (ablation)BraVL · GOD-Wikitop-5 accuracy (%): 51.44 -> 53.54Improved
add inter-modality MI term L_inter (ablation)BraVL · GOD-Wikitop-5 accuracy (%): 51.00 -> 53.54Improved
add both MI terms to the ELBO (ablation)BraVL · DIR-Wikitop-5 accuracy (%): 69.83 -> 72.39Improved
use MoPoE instead of PoE joint posterior (ablation)BraVL · GOD-Wikitop-5 accuracy (%): 39.56 -> 53.54Improved
use MoPoE instead of MoE joint posterior (ablation)BraVL · DIR-Wikitop-5 accuracy (%): 72.33 -> 72.39No clear effect
increase latent dimension beyond 32BraVL · DIR-Wiki+GOD-WikiaccuracyHurt

Closest papers by meaning

Writing a paper in this area? scime draft is a free LaTeX editor with AI included, real-time co-authors and a check of your references and venue rules before you submit. It also imports OpenAI Prism projects.

Open scime draft

Get the next scime Dispatch and a monthly field update

Free. We send a confirmation link first; every email has a one-click unsubscribe. Privacy