hopfield networks

• There are recurrent connections between nodes
• The state of each node at time step \(t+1\) is determined by the neighbors' states at time \(t\) plus a bias
• The network converges to a fixed point
• A trained network, given a partial pattern, will converge to the complete pattern