Matrix Compression via Randomized Low Rank and Low Precision Factorization

Neural Information Processing Systems 

Matrices are exceptionally useful in various fields of study as they provide a convenient framework to organize and manipulate data in a structured manner.