What is a Neural Network?
A neural network is a computing model loosely inspired by the brain, made of layers of connected nodes that learn to map inputs to outputs.
A neural network is a computing system loosely inspired by the brain. It's made of layers of simple units called neurons, connected by weighted links. By adjusting those weights during training, the network learns to turn inputs (like pixels) into outputs (like "cat").
How It Works:
- Input layer: Receives the raw data (numbers, pixels, tokens)
- Hidden layers: Each neuron multiplies inputs by weights, adds a bias, and applies an activation function
- Output layer: Produces the prediction
- Training: Compare output to the target, then use backpropagation to nudge weights toward better answers
Key Concepts:
- Weights: The learned strength of each connection
- Activation function: Adds non-linearity so the network can model complex patterns
- Loss function: Measures how wrong the prediction is
- Backpropagation: The algorithm that spreads error backward to update weights
Why They Matter:
Neural networks can approximate almost any function given enough data and capacity, which is why they're behind image recognition, translation, and generative AI.
FAQ
Is a neural network actually like a brain?
Only loosely. The "neuron" metaphor is inspiration, not a biological model. Real brains are far more complex, and artificial networks are just math optimized on data.
What is a "deep" neural network?
It simply means a network with many hidden layers. More layers let the network learn more abstract, hierarchical features.