Neuro-fuzzy-genetic architectures used in shape recognition

Artificial intelligence is one of the newest and most challenging research directions belonging to the field of computer science and has as its central objective the development of (artificial) systems endowed with certain properties (abilities) that traditionally they are assimilated to human intelligence (more concretely, the human brain), namely: language comprehension, learning ability, reasoning, generalization ability, ability to solve certain tasks, etc.

Having this aspect as a reference point, improving the performance of a standard neural network through direct action on its architecture basically involves the determination and optimization of those interaction mechanisms capable of leading to obtaining some structures robust connections as well as performance level. Additionally, these mechanisms can be divided into the following two major directions of action, namely: at the level of neural network connectivity and, respectively, at the level of neural training algorithms.

Having as its defining objective the study of neuro-fuzzy-genetic architectures with applicability in shape recognition from both a theoretical and applied perspective, the present paper is organized in the form of three chapters.