Normal distribution
Also: Gaussian distribution, Gaussian, bell curve
The continuous distribution with density (1/(σ√(2π)))·e^(−(x−μ)²/(2σ²)). By the Central Limit Theorem, averages of many independent effects tend toward it, which is why Gaussian noise is such a common assumption.
f(x) = (1 / (σ√(2π))) · e^(−(x − μ)² / (2σ²))
Example
For N(μ = 0, σ = 1), about 68% of draws fall in [−1, 1] and 95% in [−2, 2]. Assuming Gaussian noise on targets turns maximum likelihood into minimizing squared error.