Neural networks are information-processing systems that mimic the composition of the nervous system. In the most basic form of a neural network, there is a set of input neurons that receive information according to some feature. Each neuron in this set is either activated or not-activated. For each activated neuron, a signal is sent to the neurons in the next layer that the active neuron is connected to. To determine whether or not a neuron fires, each signal that neuron receives from previous neurons is summed up and measured by an activation function. If the value is appropriate for the neuron to activate, then it does so and the signal is passed forward along the network. ![[neural-network-transparent.png]] Basic neural networks like this are only *feedforward*, that is, their signal only travels in sequence from the input layer to the output layer. In order to change the behavior of the neural network, there are various procedures to manipulate the strength of connections between the neurons by *training* the network. Once the network has been trained, it can be implemented and used like an algorithm. There are also neural networks capable of *feedback*, known as [[Recurrent Networks]]. --- #AI #Psychology/Learning #Psychology/Cognition #Philosophy/Mind #2021/8 *August 10, 2021*