hi all, i'm quite new to NN, specifically deep convolutional networks. I was wondering if anybody could shed some insight on how I can get a rough estimate as to how much processing power I would require for a given neural network size. I realize it would depend on the NN architecture.
Take for example a deep convolution network with an input layer of 500x500 and 3-4 subsequent convolution and subsampling layers. How do I measure how long it would take to run an epoch?
[link][4 comments]