Frank Rosenblatt's Perceptron, an early trainable neural network for pattern recognition, was announced in July 1958, with the U.S. Navy touting it as the embryo of a machine that could one day walk, talk, and reproduce. The Mark I Perceptron learned to classify simple visual patterns. It laid conceptual groundwork for neural networks, though Minsky and Papert's later critique stalled the approach for years.