Home / Maths for AI / Symbols / Probability and statistics

Normal (Gaussian) distribution

Say it: "normal with mean mu and variance sigma squared"

What it means

The bell curve, centred at μ\mu with spread σ\sigma. About 68% of values lie within one σ\sigma of the mean and 95% within two. X∼N(0,1)X \sim \mathcal{N}(0, 1) reads "X is distributed as a standard normal".

Example

X∼N(0,1)  ⇒  P(−1.96<X<1.96)≈0.95X \sim \mathcal{N}(0, 1) \;\Rightarrow\; P(-1.96 < X < 1.96) \approx 0.95

In code

import numpy as np
x = np.random.default_rng(0).normal(0, 1, 100000)
print(round(np.mean(np.abs(x) < 1.96), 3))

Output

0.95

Where it appears in machine learning

Weight initialisation, noise in diffusion models, and the errors assumed by linear regression.

Related symbols

Learn it properly in the free lesson Probability & Distributions, with practice questions that check your answers.