Deep learning model
A deep learning model is a machine learning model in which the complex representations are represented in terms of relatively simple representation. For example, if a machine learning algorithm wants to detect a human face from an image, it can first detect edges in the first layer. In the next layer, the edges and other simple features can represent relatively more complex features such as the outline of eyes nose, etc. These complex features together represent higher-level features such as a human face.
Depth of a deep learning model
The depth of a deep learning model can be described in many ways-
One of the methods of determining the depth of a neural network or a machine learning model is based on the number of sequential instructions to be performed to evaluate the architecture. In this method, the depth of the model is determined by the longest chain in the calculation path for the model. For the example of the above deep learning model to detect the human face each layer will need several calculations. For example, there are three layers and each layer needs n calculation then. The depth of the model will be 3n. The number of calculation steps depends on the programming language used to code the model, And as a result this type of depth of model is dependent on the programming language used and hence, not preferred by many
In another method the depth of the deep learning model layers are assumed to be no of layer is used to built the model. For the above example three layers are used to determine the face. So, in this case, the depth of the model will be considered as three. I personally prefer this method to determine the depth of the graph.
For more complex problems, no of building blocks or no of separate blocks can be also considered as depth.