'deeplearning4science' 태그의 글 목록

Arxiv Link Contributions: propose a novel architecture which possibly be better MLP-alternative, having surely better interpretability and probably better accuracy. the new architecture has opened up the possibility of learning activation function itself instead of the traditional approach of learning weights.Background: Kolmogorov Arnold Representation theoremMachine learning에서, MLP는 일반적으로 univ..