Compression-enabled interpretability of voxel-wise encoding models
Fatemeh Kamali, Amir Abolfazl Suratgar, Mohammadbagher Menhaj, Reza Abbasi-Asl
This study shows that model compression improves the interpretability and stability of voxel-wise CNN models for brain activity prediction without sacrificing accuracy.
PLoS Computational Biology · Feb 19, 2025Peer-reviewedNeuroscience