Multiplying Matrices Without Multiplying
This addresses a critical bottleneck for machine learning practitioners by offering a faster alternative to existing matrix multiplication methods, potentially reducing computational costs and hardware investment needs.
The paper tackles the problem of efficiently approximating matrix multiplication, a fundamental and compute-intensive operation in machine learning, by introducing a learning-based algorithm that often runs 100× faster than exact products and 10× faster than current approximate methods, with zero multiply-adds in cases where one matrix is known ahead of time.
Multiplying matrices is among the most fundamental and compute-intensive operations in machine learning. Consequently, there has been significant work on efficiently approximating matrix multiplies. We introduce a learning-based algorithm for this task that greatly outperforms existing methods. Experiments using hundreds of matrices from diverse domains show that it often runs $100\times$ faster than exact matrix products and $10\times$ faster than current approximate methods. In the common case that one matrix is known ahead of time, our method also has the interesting property that it requires zero multiply-adds. These results suggest that a mixture of hashing, averaging, and byte shuffling$-$the core operations of our method$-$could be a more promising building block for machine learning than the sparsified, factorized, and/or scalar quantized matrix products that have recently been the focus of substantial research and hardware investment.