A computing model that simulates the working of neurons in human and animal brains. Rather than relying on carefully-crafted code as traditional computers do, neural networks are “trained” to identify, understand, and act on information through an iterative process of strengthening seemingly arbitrary connections between nodes.
Neural networks are most useful when creating a clearly-defined, logic-based algorithms to execute a task that is difficult or impossible. This includes complex activities—such as identifying images, playing strategic games, or interacting with customers—that can’t be boiled down to a set of repeatable rules. Because these networks allow computers to perform tasks that are less logic-oriented, they may underpin the automation of interactive or creative digital services..