An artificial neural network (ANN) (English: artificial neural network (ANN), or also called a simulated neural network (SNN), or more commonly just called a neural network (NN)), is a network of group processing unit small modeled based on neural networks human. ANN is an adaptive system that can change its structure to solve problems based on external or internal information that flows through the network. Therefore, it is adaptive, JST also often called adaptive network.
Currently the field of artificial intelligence in its efforts to imitate human intelligence, has not made an approach in physical form but from the other side. First conducted a study on the basic theory of the mechanism of the intelligence process. This field is called Cognitive Science. From this basic theory made a model to be simulated on a computer, and in its development is further known various artificial intelligence systems, one of which is a neural network. Compared with other disciplines, neural networks are relatively new. Some literature considers that the concept of artificial neural networks began in papers Waffen McCulloch and Walter Pitts in 1943. In the paper they try to formulate mathematical models of brain cells. The method developed by the nervous system biology, is a step forward in the computer industry.
Model on the JST is basically a function of the mathematical model that defines the function f: X → Y The term "network" in JST refers to the interconnection of multiple neurons are placed on different layers. In general, the coating on the JST is divided into three parts: Input layer (input layer) is composed of neurons that receive input data from the variable X. All the neurons in this layer can be connected to neurons in the hidden layer or directly to the outer layer if the network does not use a hidden layer. Hidden layer (hidden layer) is composed of neurons that receive data from the input layer. Outer layer (output layer) is composed of neurons that receive data from the hidden layer or directly from the input layer which symbolizes luarannya value calculation results of X into a value Y. Mathematically, the neuron is a function that accepts input from the previous layer g i (x) (layer-i). This function is generally cultivate a vector and then transformed into a scalar value through the composition of nonlinear weighted sum, where f (x) = K (Σ iwigi (x))}, K is a special function that is often called the activation function and w is a load or weight.
Post a Comment