##GD
small number of model updates
accurate
each epoch may be expensive
easy to parallelize
##SGD
Requires lots of model updates
Not as accurate, but often good enough
A log of progress in one pass for big data
Not trivial to parallelize
##GD
small number of model updates
accurate
each epoch may be expensive
easy to parallelize
##SGD
Requires lots of model updates
Not as accurate, but often good enough
A log of progress in one pass for big data
Not trivial to parallelize
本文标题:GD vs SGD
本文链接:https://www.haomeiwen.com/subject/aoosqxtx.html
网友评论