Architecture 2 min read

Neural Network

Also known as: ANN, Artificial Neural Network

A computing system inspired by biological neural networks, consisting of interconnected nodes (neurons) organized in layers that process information using learnable weights and activation functions.

Definition

A computing system inspired by biological neural networks, consisting of interconnected nodes (neurons) organized in layers that process information using learnable weights and activation functions.

Architecture 2 min read N

Overview

Neural networks are the computational framework that underpins modern AI. Inspired by the structure of biological brains, they consist of layers of interconnected nodes (neurons) that process and transform data. Through training, neural networks adjust the strength of connections between neurons (weights) to learn patterns in data.

Structure

Input Layer

Receives the raw data and passes it to the network. Each neuron in the input layer represents a feature of the input data.

Hidden Layers

One or more intermediate layers where the actual computation and pattern recognition occurs. Each layer transforms its input through weighted sums and non-linear activation functions.

Output Layer

Produces the final result — a classification, prediction, or generated output.

How Learning Works

Neural networks learn through a process called backpropagation:

  1. Forward Pass: Input data flows through the network to produce an output
  2. Loss Calculation: The output is compared to the desired result using a loss function
  3. Backward Pass: Gradients of the loss are computed with respect to each weight
  4. Weight Update: Weights are adjusted in the direction that reduces the loss

Activation Functions

  • ReLU: The most common, outputs zero for negative inputs and the input value for positive inputs
  • Sigmoid: Squashes output to the range (0, 1), useful for binary classification
  • Softmax: Produces a probability distribution across multiple classes
  • GELU: Used in Transformer models, a smoother variant of ReLU