Kind of. Deep learning is about more than matrix operations, and matrix operations are useful for applications other than deep learning, so I believe the distin

Kind of. Deep learning is about more than matrix operations, and matrix operations are useful for applications other than deep learning, so I believe the distinction is worth making. Just like with programming languages, which may all be used for the same application, it's all about what you make easy to do, and what you make difficult. I'm saying the TF's intention is to make many things beyond DL easy, although people think of it chiefly as a DL library atm.