machine learning - Why do Recurrent Neural networks need dataset iterators to prepare data -


i have problem why recurrent neural networks need dataset iterators prepare data. can please explain reason.

in recurrent neural network complete dataset sequence, every sample neural net must provide output.

they commonly used in time series.

imagine 4 sensors in person (in hands , legs) , have predict state of person (walking, running, falling, ...)

data1 (x1,x2,x3,x4) -> running data2 (x1,x2,x3,x4) -> running data3 (x1,x2,x3,x4) -> running data4 (x1,x2,x3,x4) -> falling data5 (x1,x2,x3,x4) -> falling data6 (x1,x2,x3,x4) -> in floor data7 (x1,x2,x3,x4) -> in floor .... 

it hard predict label see single data, if see sequence (current , past data) task easier.

the iterator provide order see data.


Comments

Popular posts from this blog

sql - can we replace full join with union of left and right join? why not? -

javascript - Parallax scrolling and fixed footer code causing width issues -

iOS: Performance of reloading UIImage(name:...) -