Definition
A type of neural network architecture that uses two or more identical subnetworks with shared weights to learn a similarity function between two inputs.
In short: A type of neural network architecture that uses two or more identical subnetworks with shared weights to learn a similarity function between two inputs.