Core Difference between ML and DL - an example by case study

Let’s say we ought to develop an algorithm that aims to discriminate a picture with labels: CAT, NOT CAT

If we know the features and their correlations that the algorithm should use to do the prediction, it’s ML.

If the algorithm has to learn the features first, develop their correlations and then make predictions, it’s DL.

I do not think that is correct, particularly the “learn the features first” part.

We almost always need a set of labeled training data in order to develop a model. The training data includes the features.

I think there’s a confusion.

Deep learning models are trained by using large sets of labeled data and neural network architectures that learn features directly from the data without the need for manual feature extraction.

All neural networks do this, even if a simple NN with only one hidden layer.

Deep Learning refers more to NN’s with multiple hidden layers, or advanced methods like convolutional or recurrent NN’s.