Artificial Neural Networks

Artificial neural networks (ANNs) are computing systems inspired by biological neural networks — layers of interconnected nodes that process information and learn from data. The concept was introduced in 1943 by McCulloch and Pitts; the perceptron (1958) was the first trainable neural network. Decades of limited progress followed until Geoffrey Hinton, Yann LeCun, and Yoshua Bengio developed deep learning techniques in the 2000s (2018 Turing Award). Deep neural networks — with many hidden layers — can learn to recognize images, understand speech, translate languages, generate text, and play games at superhuman levels. The 2012 ImageNet competition victory by AlexNet (a deep CNN) is often cited as the moment deep learning became dominant. GPT models, image generators (DALL-E, Stable Diffusion), and protein structure predictors (AlphaFold) are all deep neural networks.