Transition from ML to DL

Since DL is being used so widely now does it mean non-DL ML implementations will soon become outdated? Are there still use cases where non-DL ML implementations are better suited over DL implementations?

Yes as of now many other ML algorithms are widely used too, like for eg. trees, KNN, clustering etc.

Keep in mind that generally speaking an neural network could take a lot of resources compared to other ML algorithms so other might be particularly useful in some situations.

1 Like

I think not. Linear regression and logistic regression, trees, etc, will rarely require the complexity of a deep NN solution.

1 Like