Maths for ML / Notation

\approx

approximately equal

Arithmetic and relationsBeginner

Say it “is approximately”

Close enough for the purpose at hand, but not exactly equal.

Example

\pi \approx 3.14

In code

np.isclose(a, b)

torch.allclose(a, b)

Where you’ll see it in ML

Numerical gradients, approximations like a Taylor expansion, and quantized weights.

Don’t confuse it with

$=$ equals Both sides have exactly the same value.
$\propto$ proportional to Equal up to a constant multiplier you're choosing to ignore.
$x \sim p(x)$ is distributed as / sampled from x is a random value drawn from the distribution p.

Lessons that use it

The derivative: slope, and sensitivity

← All 94 symbols