How do machine translation, speech recognition fit in the supervised learning framework?

Course Q&A Machine Learning Specialization Supervised ML: Regression and Classification
,
1

Consider the video “Supervised learning part 1” in Course 1, ML Specialisation.

To recall:

It seems in supervised learning we are given input - output label pairs { \lbrace (x _i, y _i) \rbrace } and we want to find a good fit { f(x) } where for any { x , } { f(x) } is a real number.

Prof. Ng mentions speech recognition, machine translation as examples of supervised learning.

But note that in speech recognition, machine translation, the fit by design might give different outputs when the same input is given repeatedly.

(For these applications, a better framework seems to be : We are given input - output label pairs { \lbrace (x _i, y _i) \rbrace } and we want to find a good fit { f(x) } where for any { x , } { f(x) } is a random variable).

Could someone clarify the right framework (so to speak) to view machine translation, speech recognition, etc.?

Thank you in advance.

2

These tasks require far more complicated models than are covered in the MLS courses (which are Basic-level).

However they are still based on creating a good fit between a set of data and the model’s predicted outputs.

These methods are introduced in the Deep Learning Specialization (Intermediate-level).

3