description: How to use the Random model network in Neataptic authors: Thomas Wagenaar keywords: recurrent, feed-forward, gates, neural-network, random, architecture A random network is similar to a liquid network. This network will start of with a given pool of nodes, and will then create random connections between them. This network is really only useful for the initialization of the population for a genetic algorithm. ```javascript new architect.Random(input_size, hidden_size, output_size, options); ``` * `input_size` : amount of input nodes * `hidden_size` : amount of nodes inbetween input and output * `output_size` : amount of output nodes Options: * `connections` : amount of connections (default is `2 * hidden_size`, should always be bigger than `hidden_size`!) * `backconnections` : amount of recurrent connections (default is `0`) * `selfconnections` : amount of selfconnections (default is `0`) * `gates` : amount of gates (default is `0`) For example: ```javascript var network = architect.Random(1, 20, 2, { connections: 40, gates: 4, selfconnections: 4 }); drawGraph(network.graph(1000, 800), '.svg'); ``` will produce: