On application of kernel PCA for generating stimulus features for fMRI during continuous music listening

Exploiting nonlinear relationships among acoustic descriptors can lead to the novel high-level stimulus features, which can in turn reveal new brain structures involved in music processing.