Unsupervised Feature Learning Improves Prediction of Human Brain Activity in Response to Natural Images
The SC2 model was more invariant than the GWP2 model and its invariance increased from V1 to V3. (A) Mean prediction across the voxels that survived the threshold of 0.1 in the case of (b). Error bars show 1 SEM across the voxels (bootstrapping method). (B) Identification accuracy. Error bars show 1 SEM across the images in the validation set (bootstrapping method).