I was watching a Youtube video in which the problem of trying to predict the last word in a sentence was posed. The sentence was "I took my cat for a" and the last word was "walk". The lecturer in this video stated that whilst sentences (the sequence) can be of varying lengths, if we take a really large fixed window we can model the whole sentence. In essence she said that we can convert any sentence into a fixed size vector and still preserve the order of the sentence (sequence). I was then wondering why do we need RNNs if we can just use FFNNs? Also does a fixed size vector really preserve sequential order information?
Thank You for any help!
