1
$\begingroup$

I'm looking for a paper that does some comparisons between neural networks (deep learning) and traditional methods in order to prove that DL usually performs better with enough data. I know this is "known" fact but I'm struggling to find a good paper doing some research in this area.

Thanks!

$\endgroup$
1
  • $\begingroup$ I suggest you pick some task in NLP, go to some recent paper using LSTMs or whatnot, then go back in history using their references. Typically one will be to the first using their specific technique, and that is likely to contain benchmarks including "classical" techniques. $\endgroup$
    – Miguel
    Dec 28 '17 at 13:31
1
$\begingroup$

Here's a blog post providing a view to the history of the process with several popular references in it. In particular, consider the Natural Language Processing section and its conclusion:

On NLP tasks that have a long enough history to graph, there seems to be no clear indication that deep learning performs above trendline.

Also, worth noting from the general conclusion of the article:

Deep learning provides discontinuous jumps relative to previous machine learning or AI performance trendlines in image recognition and speech recognition; it doesn’t in strategy games or natural language processing, and machine translation and arcade games are ambiguous (machine translation because metrics differ; arcade games because there is no pre-deep-learning comparison.)

$\endgroup$

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service, privacy policy and cookie policy

Not the answer you're looking for? Browse other questions tagged or ask your own question.