Reducing Computational and Statistical Complexity in Machine Learning Through Cardinality Sparsity
share
High-dimensional data has become ubiquitous across the sciences but causes computational and statistical challenges. A common approach for dealing with these challenges is sparsity. In this paper, we introduce a new concept of sparsity, called cardinality sparsity. Broadly speaking, we call a tensor sparse if it contains only a small number of unique values. We show that cardinality sparsity can improve deep learning and tensor regression both statistically and computationally. On the way, we generalize recent statistical theories in those fields.
READ FULL TEXT