On permutation-invariant neural networks

Masanari Kimura, Ryotaro Shimizu, Yuki Hirakawa +2

Conventional machine learning algorithms have traditionally been designed under the assumption that input data follows a vector-based format, with an emphasis on vector-centric par…