What describes a basic neural network where data flows in one direction from input to output without cycles?

A basic neural network that allows data to flow in one direction from input to output without cycles is characterized as a feed-forward neural network. In this type of architecture, the information moves straightforwardly through the network layers, beginning from the input layer, progressing through one or more hidden layers, and concluding at the output layer. This linear flow of information is essential because it simplifies the training process and helps in applications requiring straightforward mapping from inputs to outputs, such as classification tasks.

In contrast, a recurrent neural network, which is not the correct choice, contains cycles that allow information to be reused by feeding back outputs to previous layers. Convolutional neural networks are specialized for processing data with a grid-like topology, such as images, but this specification doesn't align with the basic definition of a network with unidirectional flow. Deep neural networks refer to the number of layers in a network (often multiple hidden layers), but this term does not specifically describe the flow directionality.

Therefore, the description of a basic architecture featuring unidirectional data flow corresponds precisely to a feed-forward neural network.