Neural Network
The calculation model consisting of layer nodes, inspired by neuron connections in the human brain.
Artificial nervous network is a calculation model that consists of layer nodes (nöron) designed by inspiration from the form of connecting neurons in the human brain. A typical network consists of a layer of input that takes raw data, a layer of output that performs intermediate operations or produces more hidden layers and result. Each node adds a deviation (bias) value and passes this total from a non-linear activation function (such as ReLU, sigmoid). The linear feature is the critical element that the network can learn not only linear relations, complex and non-linear nucleus.
When training a nervous network, the difference between the output generated by the network is measured with a loss function, then the resemblance algorithm calculates that error should be updated in the direction and how much the weight of each connection is restored throughout the network. This process has thousands of times repeated with optimization algorithms such as gradyan landing, making the network generates increasingly hit predictions. From a simple single-layer sensor (perceptron) to billions of parameters deep networks, all modern artificial intelligence systems - large language models without recognizing the image - the basic structure stone is installed on this simple but strong idea.
When training a nervous network, the difference between the output generated by the network is measured with a loss function, then the resemblance algorithm calculates that error should be updated in the direction and how much the weight of each connection is restored throughout the network. This process has thousands of times repeated with optimization algorithms such as gradyan landing, making the network generates increasingly hit predictions. From a simple single-layer sensor (perceptron) to billions of parameters deep networks, all modern artificial intelligence systems - large language models without recognizing the image - the basic structure stone is installed on this simple but strong idea.