Deep Learning (DL) and Machine Learning (ML) are both a subsystem of Artificial Intelligence (AI)

Deep learning is a subset of machine learning comprising artificial neural networks. The term deep refers to the structure of artificial neural networks that include several of the so-called input, output, and hidden layers. Each layer includes sections that convert the input data into information that the neighbour layer uses.

In deep learning, the algorithm can learn how to make an accurate prediction through its own data processing, thanks to the artificial neural network structure. That sounds advantageous.

Machine learning is also a subset of artificial intelligence that uses algorithms to use the experience to improve specific tasks. It uses generally the following procedure:

• Data is introduced by the user into an algorithm
• A model is developed and trained by user support
• Testing and using the model to make predictions and use the information

Briefly, Deep learning is broadly used in processing data like text and images. For complex engineering solutions where numerical output values are based on multiphysics interactions where complex relations such as environment, processes, materials influence the response and behaviour of materials, components and systems, advanced machine learning (ML) are most of the time suitable to process data based on experience. A broad variety of algorithms such as generalized linear models, decision trees or Support Vector Machines (SVM) respond more practically. It should be noticed that finding solutions for complex engineering problems is case-sensitive, thus the choise of learning/training/predicting depends heavily on each individual problem and desired goals.

In brief, deep learning architectures perform well in cases such as:

• Image classification
• Audio classification
• Translation of texts
• Text generation and classification

Machine learning in the classical sense generally relies on patterns, the relation of parameters in the data rather than being explicitly coded by computer programmes. Therefore, the user data for predicting the solution for engineering problems comprise most of the time data that is specific to the problem and gathered through experimental or hybrid methods before being used in accelerated AI methods. It comprises deep learning, as well.