You signed in with another tab or window. Reload to refresh your session.You signed out in another tab or window. Reload to refresh your session.You switched accounts on another tab or window. Reload to refresh your session.Dismiss alert
Good idea but perhaps 'multidimensional [numeric] array' would be better since matrices are themselves tensors and are '2-dimensional arrays'. In a broad sense, tensors do not have to be "numeric", but in the context of KG embeddings, we can restrict them to be numeric. Also, obviously, tensors do not have to multidimensional since vectors are tensors but are 1-dimensional, and scalars are also tensors that are 0-dimensional, but at this place in the book, it is still a little informal and "multidimensional" can be understood as "of arbitrary dimension".
Section 5.2.1 says that:
This may suggest that a tensor is formally a field in the mathematical sense, which it is not.