G

et, designing and constructing systems capable of carrying out processes with certain intelligence has been one of the main objectives and concerns of scientists throughout history. However, despite having tools and programming languages designed for the development of intelligent systems, there is a fundamental problem that limits the results: these systems are deployed on computers based on the philosophy of Von Neumann, and rely on a sequential description of the process of information processing.

Neural networks seek the solution of complex problems, not as a sequence of steps, but as the evolution of computing systems inspired by the human brain, and endowed by both some “intelligence”, which are not but the combination of simple elements of process (neurons) interconnected, that operating in parallel get to solve problems related to the recognition of shapes or patterns prediction, coding, control and optimization among other applications.

Application

Neural networks have a practical application in 8 different areas:

  • Deductive reasoning: the conclusion is inferred from the premises proposed.

  • Pattern recognition: Vision and hearing from a machine. Find features. Classification of samples.
  • Language Processing: Communication between people and machines using natural language.

  • Perception: Knowledge of an object through the impressions communicate senses (sensors).

  • Knowledge representation: representing knowledge in a way that facilitates the inference (conclusions) from such knowledge. Organization of data.

  • Intelligent control: consists of a series of techniques that are intended to solve intractable problems of control by classical methods.

  • Machine learning: create programs able to generalize behavior from an unstructured information supplied in the form of examples.

  • Automatic planning: production plans (i.e. planning), to get a current state to a final State.