Fast Algorithms for Convolutional Neural Networks
Date:
Graduate seminar at Department of Computer Science and Automation, IISc, Bengaluru, India
This graduate seminar reviews fast convolution algorithms for neural networks, with an emphasis on Winograd minimal filtering and FFT-based convolution. It develops the forward and backward transformations and compares their arithmetic complexity with direct convolution.
The talk also examines the practical GPU tradeoffs behind these algorithms, including transform overhead, numerical precision, workspace requirements, data layout, and the effects of filter and batch size.