M.J. Flynn
Proceedings of the IEEE
Efficient SIMD implementation for accelerating convolutional neural network