Explainable machine learning for memory-related decoding via TabNet and non-linear features | IEEE Conference Publication | IEEE Xplore