# Neural network A neural network is a computational model composed of interconnected processing units whose weighted connections are trained to approximate functions, forming the basis of modern deep learning and artificial intelligence systems. A neural network is a computational model composed of interconnected layers of simple processing units, loosely inspired by biological neurons, whose connection strengths, or weights, are adjusted through training to approximate a desired function. Neural networks are the core computational structure behind modern [deep-learning](https://www.wikiprompt.org/wiki/deep-learning) and, by extension, most contemporary [artificial-intelligence](https://www.wikiprompt.org/wiki/artificial-intelligence) systems. ## History The theoretical base for contemporary neural networks was independently proposed by Alexander Bain in 1873 and William James in 1890, both positing that human thought emerged from interactions among large numbers of neurons. In 1949, Donald Hebb described Hebbian learning, the idea that neural networks can change and learn over time by strengthening a synapse every time a signal travels along it. The earliest trainable artificial neural model was the [perceptron](https://www.wikiprompt.org/wiki/perceptron), introduced by [frank-rosenblatt](https://www.wikiprompt.org/wiki/frank-rosenblatt) in 1958, which could learn to classify simple patterns using a single layer of adjustable weights. Its limitations, documented in a 1969 critique by [marvin-minsky](https://www.wikiprompt.org/wiki/marvin-minsky) and Seymour Papert, contributed to a broader loss of research funding known as an [ai-winter](https://www.wikiprompt.org/wiki/ai-winter). Interest in multi-layer networks revived in the 1980s with the popularization of [backpropagation](https://www.wikiprompt.org/wiki/backpropagation), an algorithm for efficiently computing how each weight in a network should change to reduce error, which allowed networks with hidden layers to be trained in practice. Neural networks moved from a niche technique to the dominant approach in AI following the 2012 success of [alexnet](https://www.wikiprompt.org/wiki/alexnet), which demonstrated that networks with many layers, trained on large datasets using graphics processing units, could dramatically outperform prior methods on real-world tasks. ## Structure A neural network is organized into layers of units, or neurons, each of which computes a weighted sum of its inputs and passes the result through a nonlinear activation function before sending it to the next layer. The "signal" input to each neuron is a number, specifically a linear combination of the outputs of the connected neurons in the previous layer. The behavior of the network depends on the strengths (or weights) of the connections between neurons. A network is trained by modifying these weights through [empirical-risk-minimization](https://www.wikiprompt.org/wiki/empirical-risk-minimization) or backpropagation in order to fit some preexisting dataset. Networks with many such layers between input and output are described as "deep," giving deep learning its name. During training, a [loss-function](https://www.wikiprompt.org/wiki/loss-function) measures the gap between the network's output and the desired result, and [gradient-descent](https://www.wikiprompt.org/wiki/gradient-descent) iteratively adjusts weights to reduce that gap, with backpropagation supplying the necessary gradients efficiently across all layers at once. ## Architectural families Different network architectures suit different kinds of data. [convolutional-neural-network](https://www.wikiprompt.org/wiki/convolutional-neural-network)s use shared, spatially local filters and are well suited to images. [recurrent-neural-network](https://www.wikiprompt.org/wiki/recurrent-neural-network)s, including the [lstm](https://www.wikiprompt.org/wiki/lstm) variant, process sequences step by step and were historically used for text and speech before being displaced by the [transformer](https://www.wikiprompt.org/wiki/transformer) architecture, which processes entire sequences in parallel using attention. Generative architectures such as the [generative-adversarial-network](https://www.wikiprompt.org/wiki/generative-adversarial-network) and [diffusion-model](https://www.wikiprompt.org/wiki/diffusion-model) apply neural networks to the task of producing new data rather than only classifying or predicting it. ## Limitations and interpretability Neural networks are often criticized as "black boxes": their internal representations are difficult for humans to inspect or explain, even when their outputs are accurate. This has motivated a research subfield sometimes called [interpretability](https://www.wikiprompt.org/wiki/interpretability), which attempts to reverse-engineer what individual components of a trained network represent. Despite this opacity, neural networks remain the dominant tool across vision, language, audio, and increasingly robotics and scientific applications, largely because their performance continues to improve predictably with more data, more parameters, and more compute. ## Recent developments In the 2020s, neural networks have continued to scale in size and capability, particularly in the domain of [generative-ai](https://www.wikiprompt.org/wiki/generative-ai). Large language models, built on the transformer architecture, have demonstrated emergent abilities in reasoning, translation, and code generation, while diffusion models have become the standard for high-quality image and video synthesis. Research into neural network interpretability has also advanced, with techniques such as mechanistic interpretability aiming to reverse-engineer the internal computations of trained networks. These developments have solidified neural networks as a foundational technology across numerous fields, including scientific discovery and autonomous systems. ## See also * [connectionism](https://www.wikiprompt.org/wiki/connectionism) * [machine-learning](https://www.wikiprompt.org/wiki/machine-learning) * [history-of-artificial-neural-networks](https://www.wikiprompt.org/wiki/history-of-artificial-neural-networks) --- Source: https://www.wikiprompt.org/wiki/neural-network License: CC BY-SA 4.0 (https://creativecommons.org/licenses/by-sa/4.0/) Last updated: 2026-11-09T12:24:48.525181+00:00