One of the images that i recently noticed on the internet that compares machine learning and Deep learning is this:

enter image description here

What I understood from it is, deep learning performance is not limited when we increase the amount of data for learning "training" ..

So, is this feature restricted to the specific structure of a model!? for example, if I have a Deep neural network with a 10 layers and each layer contains 8 hidden units! if i continue increase the amount of data, Do the performance will still increase, or i have to build another complex one more number of layers and units?


closed as unclear what you're asking by Xi'an, kjetil b halvorsen, Michael R. Chernick, Peter Flom - Reinstate Monica Feb 16 '18 at 13:24

Please clarify your specific problem or add additional details to highlight exactly what you need. As it's currently written, it’s hard to tell exactly what you're asking. See the How to Ask page for help clarifying this question. If this question can be reworded to fit the rules in the help center, please edit the question.


This is an over simplified and misleading plot. Ultimately performance is determined by the quality of the predictors and the signal to noise ratio. Adding more rows of data (more examples) may help with SNR but can't improve the quality of the predictors, whether you take a deep learning or alternative approach. The unbounded trend in the plot suggests perfect prediction is attainable if only one had enough data. The truth is that in many contexts deep learning will perform no better than other approaches. It may perform better, particularly when you add more columns (predictors), but this is certainly not guaranteed.


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