Exploring deep neural networks
Deep learning is defined as a class of machine learning algorithms with certain characteristics. These models use multiple, hidden, nonlinear cascade layers to perform feature extraction and transformation jobs. Each level takes in the outputs from the previous level. These algorithms can be supervised, to deal with classification problems, or unsupervised, to deal with pattern analysis. The latter is based on multiple hierarchical layers of data characteristics and representations. In this way, the features of the higher layers are obtained from those of the lower layers, thus forming a hierarchy. Moreover, they learn multiple levels of representation corresponding to various levels of abstraction until they form a hierarchy of concepts.
The composition of each layer depends on the problem that needs to be solved. Deep learning techniques mainly adopt multiple hidden levels of an ANN but also sets of propositional formulas. The ANNs adopted have...
