I dont think that there is a winner. Both schemes refer to different problems while in some circustances they are complementary

I'm not sure the term of ANN you mean here is the same as deep learning or not. If yes, I would like to say both conventional machine learning and deep learning have their own applicable fields at this moments. In terms of performance, it seems that deep learning wins. However, the non-interpretability (black box) of deep learning is still a big problem now.

By ANN

k-fold validation

Confusion matrix

Use of validation dataset where you know the nature of #s !

Can be measured by using random data set for testing or by using a real data set. Anyway, this data set differs from the data set that used for learning ANN.

ANN(as deep learning) might win for now.

The onlything concerns abit is the fact that it is blackbox. However, from another perspective, it is just as reliable as human in term you say you undertand a person pretty well but who could be sure so.

Artificial neural networks or connectionist systems are computing systems that are inspired by, but not necessarily identical to, the biological neural networks that constitute animal brains. Such systems "learn" to perform tasks by considering examples, generally without being programmed with any task-specific rules

It depends on covergence

Artificial Neural Networks (ANN) are computing systems of inspired the biological neural networks those are consists of animal brains. I would like to say that it depends on the "Learn" on convergence.

I used ANN (3D CNN) on my real corpus nd achieved a v good result although I so used AI techniques bt ANN r best nowadys

I think there is many different meaning to improve the performance of ANN like increase the number of hide nodes or randomly data base for each problem or using type of intelligent system in specific situations

I think they are no different. Machine learning is not about networking and structure, and it challenges most programming capabilities.