Using the BSP cost model to optimise parallel neural network training

Abstract
No abstract available