The network, created by researchers at Google's X laboratory and intended to simulate the working of the human brain, "learned" to recognize cats among 10 million random digital images found in YouTube videos, The New York Times reported Tuesday.
In the Google research, the machine was given no help in identifying features, scientists said.
"We never told it during the training, 'This is a cat,'" said Google fellow Jeff Dean, who helped Google design the software that lets the neural network easily break programs into many tasks that can be computed simultaneously.
"It basically invented the concept of a cat."
The neural network result appeared to support theories developed by biologists suggesting individual neurons are trained inside the brain to detect significant objects, researchers said.
However, researchers said, despite the immense computing capacity of their network it was still dwarfed by the number of connections found in the human brain.
"It is worth noting that our network is still tiny compared to the human visual cortex, which is a million times larger in terms of the number of neurons and synapses," they said.
Still, they said, the research suggests existing machine learning algorithms can improve considerably as the machines are given access to large amounts of data.