3 ms·
One hot vectors. You build a dictionary of all words + one catch all for unknown words. Each word then has a position in a sparse vector. So for example : Yes
by wodenokoto 11y ago
One hot vectors. You build a dictionary of all words + one catch all for unknown words. Each word then has a position in a sparse vector.
So for example :
Yes = (0,0,1,0,0,...)
No = (0,0,0,1,0,....)
Convolutional and recurrent nets can handle inputs of arbitrary lengths.