Facto-CNN: Memory-Efficient CNN Training with Low-rank Tensor Factorization and Lossy Tensor Compression.

Seungtae Lee, Jonghwan Ko, Seokin Hong

Browse the full ACML paper archive.