Neural Network
A computing system of interconnected nodes inspired by biological brains, trained to recognise patterns.
Definition
An artificial neural network consists of layers of interconnected nodes (neurons) that process information using weighted connections. Each neuron applies a non-linear activation function (ReLU, sigmoid) to a weighted sum of its inputs, enabling the network to model complex non-linear relationships.
Networks are organised in layers: an input layer, one or more hidden layers, and an output layer. Deeper networks (more hidden layers) can represent increasingly complex functions. Training adjusts connection weights using backpropagation and gradient descent to minimise prediction error on training data.
Key architectures include feedforward networks (fully connected), convolutional networks (CNNs for images), recurrent networks (RNNs/LSTMs for sequences), and transformers (attention-based, dominating language and vision).
Examples
- Image classifier
- Speech recogniser
- GPT-4 internal architecture
Want a deeper dive?
Read our full explainer with use cases, how-it-works, and FAQs.
Neural Networks concept guideRelated Terms
Deep Learning
A subset of machine learning using neural networks with many layers to learn complex hierarchical representations.
Transformer
A neural network architecture using self-attention to process sequences in parallel — the foundation of all modern LLMs.
Convolutional Neural Network (CNN)
A neural network architecture using convolutional filters to extract spatial features from images or sequences.
Recurrent Neural Network (RNN)
A neural architecture that processes sequences by maintaining a hidden state across time steps.
Backpropagation
The algorithm for computing gradients in neural networks by propagating errors backwards through layers.