Connectionist models of computation represent a program as a set of weighted connections between nodes in a directed graph. Each node has an associated local algorithm for combining values from its input connections and passing the result along its output connections. The models are sometimes termed “electronic neural networks” because they are similar to abstract models of neural processing. The models also resemble the type of fine-grained parallelism in data flow architectures. Although still an emerging technology, electronic neural networks have many advantages over standard computational models, including fault tolerance, the ability to make graded judgments, and quick convergence to approximate solutions.