Cheat-sheet

Math conventions for Machine Learning

130+ essential symbols with reading, meaning, examples, and real ML usage. Search, filter by category, copy the LaTeX in one click.

165 symbols

14 categories

Bilingual EN/ES

Press / to search

★ Most used

Most used

The subset you'll see in every ML paper. Click any card to expand examples.

x \in A

Element of

Read it: "x is in A"

x

is a member of the set

A

A \subset B

Proper subset

Read it: "A is a proper subset of B"

Every element of

A

is in

B

, and

A \neq = B

A \cup B

Union

Read it: "A union B"

Elements that are in

A

, in

B

, or in both.

A \cap B

Intersection

Read it: "A intersect B"

Elements that are in both

A

and

B

\emptyset

Empty set

Read it: "the empty set"

The set with no elements.

\forall x \in A

For all

Read it: "for all x in A"

The statement holds for every

x

A

\exists x \in A

There exists

Read it: "there exists x in A"

At least one

x \in A

satisfies the condition.

N

Natural numbers

Read it: "the natural numbers"

{0, 1, 2, 3, \dots}

(or

{1, 2, \dots}

depending on convention).

R

Real numbers

Read it: "the reals"

All numbers on the continuous number line.

R^{n}

n-dim Euclidean space

Read it: "R-n"

All ordered

n

-tuples of reals.

f : X \to Y

Function notation

Read it: "f from X to Y"

f

is a function with domain

X

and codomain

Y

f \circ g

Composition

Read it: "f composed with g"

(f \circ g) (x) = f (g (x))

x \to a lim f (x)

Limit

Read it: "limit as x goes to a of f(x)"

Value

f (x)

approaches as

x

gets arbitrarily close to

a

x

Vector (bold lowercase)

Read it: "vector x" (or "bold x")

Convention: bold lowercase letters are vectors.

x \in R^{d}

x_{i}

i-th component

Read it: "x sub i"

Scalar coordinate

i

of vector

x

∥ x ∥

Norm

Read it: "norm of x"

A measure of length. Defaults to L2 unless subscripted.

∥ x ∥_{2}

L2 norm (Euclidean)

Read it: "L2 norm of x"

\sum_{i} x_{i}^{2}

— the geometric length.

⟨ x, y ⟩

Inner product

Read it: "inner product of x and y"

Scalar measure of similarity between two vectors.

x \cdot y

Dot product

Read it: "x dot y"

\sum_{i} x_{i} y_{i}

— the standard inner product on

R^{n}

x^{⊤}

Transpose (vector)

Read it: "x transpose"

Switches a column vector to a row vector.

x^{⊤} y

is the dot product.

A

Matrix (bold uppercase)

Read it: "matrix A"

Convention: bold uppercase letters are matrices.

A \in R^{m \times n}

A_{ij}

Matrix entry

Read it: "A sub i j"

Element in row

i

, column

j

A^{⊤}

Transpose

Read it: "A transpose"

Swap rows and columns:

(A^{⊤})_{ij} = A_{j i}

A^{- 1}

Inverse

Read it: "A inverse"

Matrix with

A A^{- 1} = I

— exists iff

det (A) \neq = 0

det (A)

Determinant

Read it: "determinant of A"

Scalar measuring the signed volume scaling of the linear map

A

tr (A)

Trace

Read it: "trace of A"

Sum of diagonal entries:

\sum_{i} A_{ii}

. Also equals the sum of eigenvalues.

I_{n}

Identity matrix

Read it: "identity"

n \times n

diagonal matrix with all

1

Ix = x

\frac{df}{d x}

Derivative

Read it: "d f over d x"

Instantaneous rate of change of

f

with respect to

x

\frac{\partial f}{\partial x _{i}}

Partial derivative

Read it: "partial f partial x i"

Derivative with respect to one variable, others held fixed.

\nabla f

Gradient

Read it: "gradient of f" or "nabla f"

Vector of partial derivatives — points in direction of steepest ascent.

\int_{a}^{b} f (x) d x

Definite integral

Read it: "integral from a to b of f of x dx"

Signed area between

f

and the

x

-axis on

[a, b]

i = 1 \sum n a_{i}

Sum

Read it: "sum from i=1 to n of a sub i"

a_{1} + a_{2} + \dots + a_{n}

i = 1 \prod n a_{i}

Product

Read it: "product from i=1 to n of a sub i"

a_{1} \cdot a_{2} \dots a_{n}

P (A)

Probability of event

Read it: "probability of A"

A number in

[0, 1]

measuring how likely event

A

is.

p (x)

PDF / PMF

Read it: "p of x"

Density (continuous) or mass (discrete) of a random variable at

x

p (x ∣ y)

Conditional

Read it: "p of x given y"

Density of

X

given that

Y = y

is observed.

E [X]

Expectation

Read it: "expected value of X"

Average value of

X

under its distribution.

Var (X)

Variance

Read it: "variance of X"

E [(X - E [X])^{2}]

— squared spread around the mean.

X \sim D

Distributed as

Read it: "X distributed as D"

X

has distribution

D

N (μ, σ^{2})

Gaussian / Normal

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

Symmetric bell-shaped distribution with mean

μ

and variance

σ^{2}

x arg min f (x)

Argmin

Read it: "argmin over x of f"

The

x

that minimises

f

(not the minimum value).

x arg max f (x)

Argmax

Read it: "argmax over x of f"

The

x

that maximises

f

L (θ)

Loss / objective

Read it: "loss of theta"

Function we minimise during training.

η

Learning rate

Read it: "eta"

Step size in gradient-based optimisation.

θ

Parameters

Read it: "theta"

Generic name for the parameters of a model.

w

Weights

Read it: "weights"

Linear coefficients of a model.

b

Bias

Read it: "bias"

Additive offset in

\overset{y}{^} = w^{⊤} x + b

\overset{y}{^}

Predicted output

Read it: "y hat"

Model's prediction (the hat marks 'estimated').

σ (z)

Sigmoid

Read it: "sigmoid of z"

\frac{1}{1 + e ^{- z}}

— squashes any real number into

(0, 1)

softmax (z)_{i}

Softmax

Read it: "softmax of z, i-th component"

\frac{e ^{z_{i}}}{\sum _{j} e ^{z_{j}}}

— converts logits to a probability vector.

O (f (n))

Big-O

Read it: "big O of f"

Upper bound: grows no faster than

f

α

Alpha

Read it: "alpha"

Often: prior parameter (Beta/Dirichlet), learning rate (older texts), Type-I error.

ε

Epsilon

Read it: "epsilon"

Small constant, error term, exploration rate in RL.

η

Eta

Read it: "eta"

Most common letter for the learning rate.

θ

Theta

Read it: "theta"

Generic parameter vector. Also: angle.

λ

Lambda

Read it: "lambda"

Eigenvalue, regularisation strength, rate of an exponential / Poisson.

μ

Read it: "mu"

Mean (statistics), step size in some optimisers.

π

Read it: "pi"

Mixture weights, RL policy, and the constant

\approx 3.14159

σ

Sigma

Read it: "sigma"

Standard deviation, sigmoid (yes, the same letter is reused).

Ω

Big Omega (sample space)

Read it: "big omega"

Sample space — the set of all possible outcomes.

≡ Full catalogue

Full catalogue

All notation, grouped by area. Use the chips above to jump to a category.

Sets & logic

Membership, subsets, unions, quantifiers — the language any math text starts in.

17 items

x \in A

Element of

Read it: "x is in A"

x

is a member of the set

A

x \in / A

Not an element of

Read it: "x is not in A"

x

is not a member of

A

A \subset B

Proper subset

Read it: "A is a proper subset of B"

Every element of

A

is in

B

, and

A \neq = B

A \subseteq B

Subset (or equal)

Read it: "A is a subset of B"

Every element of

A

is in

B

;

A = B

is allowed.

A \supset B

Proper superset

Read it: "A is a proper superset of B"

B

is a proper subset of

A

A \cup B

Union

Read it: "A union B"

Elements that are in

A

, in

B

, or in both.

A \cap B

Intersection

Read it: "A intersect B"

Elements that are in both

A

and

B

A ∖ B

Set difference

Read it: "A minus B"

Elements in

A

that are not in

B

A^{c}

Complement

Read it: "A complement"

Everything in the universe

Ω

that is not in

A

\emptyset

Empty set

Read it: "the empty set"

The set with no elements.

\forall x \in A

For all

Read it: "for all x in A"

The statement holds for every

x

A

\exists x \in A

There exists

Read it: "there exists x in A"

At least one

x \in A

satisfies the condition.

P \Rightarrow Q

Implies

Read it: "P implies Q"

P

is true, then

Q

is true.

P \Leftrightarrow Q

If and only if

Read it: "P iff Q"

P

and

Q

are logically equivalent.

\neg P

Logical NOT

Read it: "not P"

Negation:

P

is false.

P \land Q

Logical AND

Read it: "P and Q"

Both

P

and

Q

are true.

P \lor Q

Logical OR

Read it: "P or Q"

At least one of

P

Q

is true.

Number sets

Naturals, integers, rationals, reals — the universes where ML actually lives.

8 items

N

Natural numbers

Read it: "the natural numbers"

{0, 1, 2, 3, \dots}

(or

{1, 2, \dots}

depending on convention).

Z

Integers

Read it: "the integers"

{\dots, - 2, - 1, 0, 1, 2, \dots}

Q

Rational numbers

Read it: "the rationals"

Numbers expressible as

p / q

with

p, q \in Z

and

q \neq = 0

R

Real numbers

Read it: "the reals"

All numbers on the continuous number line.

R^{n}

n-dim Euclidean space

Read it: "R-n"

All ordered

n

-tuples of reals.

R_{> 0}

Positive reals

Read it: "strictly positive reals"

{x \in R : x > 0}

C

Complex numbers

Read it: "the complex numbers"

Numbers of the form

a + bi

with

a, b \in R

and

i^{2} = - 1

∣ x ∣

Absolute value

Read it: "absolute value of x"

Distance from

x

0

on the real line.

Functions & limits

Function notation, composition, limits, sup/inf, max/min.

10 items

f : X \to Y

Function notation

Read it: "f from X to Y"

f

is a function with domain

X

and codomain

Y

f^{- 1}

Inverse function

Read it: "f inverse"

The function that undoes

f

(when one exists).

f \circ g

Composition

Read it: "f composed with g"

(f \circ g) (x) = f (g (x))

id_{X}

Identity function

Read it: "identity on X"

id_{X} (x) = x

x \to a lim f (x)

Limit

Read it: "limit as x goes to a of f(x)"

Value

f (x)

approaches as

x

gets arbitrarily close to

a

sup S

Supremum

Read it: "supremum of S"

Least upper bound — the smallest number

\geq

everything in

S

in f S

Infimum

Read it: "infimum of S"

Greatest lower bound — largest number

\leq

everything in

S

max

Maximum

Read it: "maximum"

Largest value in a set.

min

Minimum

Read it: "minimum"

Smallest value in a set.

dom (f)

Domain

Read it: "domain of f"

Set of inputs

f

accepts.

Vectors

Bold lowercase, indices, norms, inner products, projections.

15 items

x

Vector (bold lowercase)

Read it: "vector x" (or "bold x")

Convention: bold lowercase letters are vectors.

x \in R^{d}

x_{i}

i-th component

Read it: "x sub i"

Scalar coordinate

i

of vector

x

0

Zero vector

Read it: "zero vector"

All entries equal to

0

1

All-ones vector

Read it: "vector of ones"

All entries equal to

1

e_{i}

Standard basis vector

Read it: "e sub i"

Vector with

1

in position

i

and

0

elsewhere.

∥ x ∥

Norm

Read it: "norm of x"

A measure of length. Defaults to L2 unless subscripted.

∥ x ∥_{2}

L2 norm (Euclidean)

Read it: "L2 norm of x"

\sum_{i} x_{i}^{2}

— the geometric length.

∥ x ∥_{1}

L1 norm (Manhattan)

Read it: "L1 norm of x"

\sum_{i} ∣ x_{i} ∣

— taxicab distance.

∥ x ∥_{\infty}

Infinity norm

Read it: "max norm"

max_{i} ∣ x_{i} ∣

— the largest absolute coordinate.

⟨ x, y ⟩

Inner product

Read it: "inner product of x and y"

Scalar measure of similarity between two vectors.

x \cdot y

Dot product

Read it: "x dot y"

\sum_{i} x_{i} y_{i}

— the standard inner product on

R^{n}

x^{⊤}

Transpose (vector)

Read it: "x transpose"

Switches a column vector to a row vector.

x^{⊤} y

is the dot product.

x ⊥ y

Orthogonal

Read it: "x is orthogonal to y"

⟨ x, y ⟩ = 0

. Geometrically: 90° angle.

x \otimes y

Outer product

Read it: "x outer y"

Matrix with entries

x_{i} y_{j}

. Equals

x y^{⊤}

proj_{u} (v)

Projection onto u

Read it: "projection of v onto u"

\frac{u ^{⊤} v}{u ^{⊤} u} u

— the closest scaled multiple of

u

v

Matrices

Bold uppercase, transpose, inverse, determinant, trace, rank.

15 items

A

Matrix (bold uppercase)

Read it: "matrix A"

Convention: bold uppercase letters are matrices.

A \in R^{m \times n}

A_{ij}

Matrix entry

Read it: "A sub i j"

Element in row

i

, column

j

A^{⊤}

Transpose

Read it: "A transpose"

Swap rows and columns:

(A^{⊤})_{ij} = A_{j i}

A^{- 1}

Inverse

Read it: "A inverse"

Matrix with

A A^{- 1} = I

— exists iff

det (A) \neq = 0

A^{†}

Pseudoinverse

Read it: "A dagger"

Moore–Penrose pseudoinverse — generalises

A^{- 1}

to non-square / singular matrices.

A^{*}

Conjugate transpose

Read it: "A star"

Transpose followed by complex conjugation. For real matrices, equals

A^{⊤}

det (A)

Determinant

Read it: "determinant of A"

Scalar measuring the signed volume scaling of the linear map

A

tr (A)

Trace

Read it: "trace of A"

Sum of diagonal entries:

\sum_{i} A_{ii}

. Also equals the sum of eigenvalues.

rank (A)

Rank

Read it: "rank of A"

Number of linearly independent columns (= rows). Counts the matrix's intrinsic dimension.

diag (d_{1}, \dots, d_{n})

Diagonal matrix

Read it: "diagonal matrix with entries d1 to dn"

Square matrix with the given values on the diagonal and zeros elsewhere.

I_{n}

Identity matrix

Read it: "identity"

n \times n

diagonal matrix with all

1

Ix = x

0

Zero matrix

Read it: "zero matrix"

All entries equal to

0

A \otimes B

Kronecker product

Read it: "A Kronecker B"

Block matrix with each block

A_{ij} B

A ⪰ 0

Positive semidefinite

Read it: "A is positive semidefinite"

x^{⊤} A x \geq 0

for all

x

. Covariance matrices are always PSD.

∥ A ∥_{F}

Frobenius norm

Read it: "Frobenius norm of A"

\sum_{ij} A_{ij}^{2}

— the L2 norm of the flattened matrix.

Linear algebra (advanced)

Subspaces, span, dimension, kernels, images, orthogonal complements.

8 items

col (A)

Column space

Read it: "column space of A"

Subspace spanned by the columns of

A

row (A)

Row space

Read it: "row space of A"

Subspace spanned by the rows of

A

ker (A)

Null space (kernel)

Read it: "kernel of A"

{x : Ax = 0}

im (A)

Image

Read it: "image of A"

Set of all outputs

Ax

. Same as the column space.

span {v_{1}, \dots, v_{k}}

Span

Read it: "span of v1 to vk"

All linear combinations of the given vectors.

dim (V)

Dimension

Read it: "dimension of V"

Number of vectors in any basis of the subspace

V

W^{⊥}

Orthogonal complement

Read it: "W perp"

{v : v ⊥ w for every w \in W}

U \oplus V

Direct sum

Read it: "U direct sum V"

Sum of subspaces with

U \cap V = {0}

— every vector splits uniquely.

Calculus

Derivatives, partials, gradients, Hessians, integrals.

10 items

\frac{df}{d x}

Derivative

Read it: "d f over d x"

Instantaneous rate of change of

f

with respect to

x

f^{'} (x)

Prime derivative

Read it: "f prime of x"

Same as

\frac{df}{d x} (x)

, in shorter notation.

\frac{\partial f}{\partial x _{i}}

Partial derivative

Read it: "partial f partial x i"

Derivative with respect to one variable, others held fixed.

\nabla f

Gradient

Read it: "gradient of f" or "nabla f"

Vector of partial derivatives — points in direction of steepest ascent.

\nabla^{2} f

Hessian

Read it: "nabla squared f"

Matrix of second partials — measures local curvature.

Δ f

Laplacian

Read it: "laplacian of f"

\sum_{i} \frac{\partial ^{2} f}{\partial x _{i}^{2}}

\nabla \cdot F

Divergence

Read it: "divergence of F"

Net 'flow' out of a point for vector field

F

\int_{a}^{b} f (x) d x

Definite integral

Read it: "integral from a to b of f of x dx"

Signed area between

f

and the

x

-axis on

[a, b]

\int f (x) d x

Indefinite integral

Read it: "integral of f of x dx"

Family of antiderivatives of

f

J_{f} (x)

Jacobian

Read it: "Jacobian of f at x"

Matrix of all first-order partial derivatives of a vector-valued function.

Sums, products & indexing

Big sigma and pi, factorials, binomials, floor/ceiling.

6 items

i = 1 \sum n a_{i}

Sum

Read it: "sum from i=1 to n of a sub i"

a_{1} + a_{2} + \dots + a_{n}

i = 1 \prod n a_{i}

Product

Read it: "product from i=1 to n of a sub i"

a_{1} \cdot a_{2} \dots a_{n}

(k n)

Binomial coefficient

Read it: "n choose k"

\frac{n !}{k ! ( n - k )!}

— number of

k

-subsets of an

n

-set.

n!

Factorial

Read it: "n factorial"

n \cdot (n - 1) \dots 2 \cdot 1

, with

0! = 1

⌊ x ⌋

Floor

Read it: "floor of x"

Greatest integer

\leq x

⌈ x ⌉

Ceiling

Read it: "ceiling of x"

Least integer

\geq x

Probability

Events, densities, expectation, variance, independence, divergence, entropy.

16 items

P (A)

Probability of event

Read it: "probability of A"

A number in

[0, 1]

measuring how likely event

A

is.

p (x)

PDF / PMF

Read it: "p of x"

Density (continuous) or mass (discrete) of a random variable at

x

p (x ∣ y)

Conditional

Read it: "p of x given y"

Density of

X

given that

Y = y

is observed.

p (x, y)

Joint

Read it: "p of x and y"

Joint distribution of two random variables.

E [X]

Expectation

Read it: "expected value of X"

Average value of

X

under its distribution.

Var (X)

Variance

Read it: "variance of X"

E [(X - E [X])^{2}]

— squared spread around the mean.

Cov (X, Y)

Covariance

Read it: "covariance of X and Y"

E [(X - μ_{X}) (Y - μ_{Y})]

— joint variability.

σ_{X}

Standard deviation

Read it: "sigma X"

Var (X)

— spread in the same units as

X

X \sim D

Distributed as

Read it: "X distributed as D"

X

has distribution

D

X_{i} \sim iid D

i.i.d.

Read it: "X i, i.i.d. from D"

Independent and identically distributed — fundamental ML assumption on training data.

X ⊥ Y

Independent

Read it: "X is independent of Y"

p (x, y) = p (x) p (y)

— knowing one tells you nothing about the other.

X ⊥ Y ∣ Z

Conditionally independent

Read it: "X is independent of Y given Z"

p (x, y ∣ z) = p (x ∣ z) p (y ∣ z)

D_{K L} (p ∥ q)

KL divergence

Read it: "KL of p from q"

\sum_{x} p (x) lo g \frac{p ( x )}{q ( x )}

— asymmetric 'distance' between distributions.

H (X)

Entropy

Read it: "entropy of X"

- \sum_{x} p (x) lo g p (x)

— average information content.

H (X ∣ Y)

Conditional entropy

Read it: "entropy of X given Y"

Average uncertainty in

X

given

Y

I (X; Y)

Mutual information

Read it: "mutual information of X and Y"

Reduction in entropy of

X

from observing

Y

I (X; Y) = H (X) - H (X ∣ Y)

Common distributions

Gaussian, Bernoulli, Binomial, Uniform, Exponential, Categorical, Dirichlet.

11 items

N (μ, σ^{2})

Gaussian / Normal

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

Symmetric bell-shaped distribution with mean

μ

and variance

σ^{2}

N (μ, Σ)

Multivariate Gaussian

Read it: "multivariate normal"

Vector-valued generalisation with mean

μ

and covariance

Σ

Bern (p)

Bernoulli

Read it: "Bernoulli p"

0/1 outcome with

P (X = 1) = p

. The atomic model of binary classification.

Bin (n, p)

Binomial

Read it: "binomial n p"

Number of successes in

n

independent Bernoulli(

p

) trials.

U (a, b)

Uniform

Read it: "uniform on a b"

Constant density on

[a, b]

Exp (λ)

Exponential

Read it: "exponential lambda"

Continuous waiting times with rate

λ

. Memoryless.

Pois (λ)

Poisson

Read it: "Poisson lambda"

Counts of independent events in a fixed window with mean

λ

Gamma (α, β)

Gamma

Read it: "Gamma alpha beta"

Continuous positive distribution; conjugate prior for the precision of a Gaussian.

Beta (α, β)

Beta

Read it: "Beta alpha beta"

Distribution on

[0, 1]

; conjugate prior for Bernoulli/Binomial.

Cat (π)

Categorical

Read it: "categorical with weights pi"

Discrete distribution over

K

classes with probabilities

π

Dir (α)

Dirichlet

Read it: "Dirichlet alpha"

Distribution over the simplex; conjugate prior for Categorical.

Optimization

argmin / argmax, subject-to, Lagrangians, learning rates.

7 items

x arg min f (x)

Argmin

Read it: "argmin over x of f"

The

x

that minimises

f

(not the minimum value).

x arg max f (x)

Argmax

Read it: "argmax over x of f"

The

x

that maximises

f

s.t.

Subject to

Read it: "subject to"

Introduces constraints in an optimisation problem.

L (x, λ)

Lagrangian

Read it: "Lagrangian of x and lambda"

Objective + constraints folded together via multipliers

λ

L (θ)

Loss / objective

Read it: "loss of theta"

Function we minimise during training.

η

Learning rate

Read it: "eta"

Step size in gradient-based optimisation.

λ

Regularisation strength

Read it: "lambda" (when used as regulariser)

Scalar weighting the regularisation term in a loss.

ML-specific

Parameters, weights, predictions, losses, design matrix, activations.

13 items

θ

Parameters

Read it: "theta"

Generic name for the parameters of a model.

w

Weights

Read it: "weights"

Linear coefficients of a model.

b

Bias

Read it: "bias"

Additive offset in

\overset{y}{^} = w^{⊤} x + b

\overset{y}{^}

Predicted output

Read it: "y hat"

Model's prediction (the hat marks 'estimated').

X

Design matrix

Read it: "design matrix X"

Stacked features: row

i

is the feature vector of sample

i

n

Number of samples

Read it: "n"

Training set size.

d

Feature dimension

Read it: "d"

Number of features per sample.

ε

Noise / error

Read it: "epsilon"

Random fluctuation in

y = f (x) + ε

σ (z)

Sigmoid

Read it: "sigmoid of z"

\frac{1}{1 + e ^{- z}}

— squashes any real number into

(0, 1)

softmax (z)_{i}

Softmax

Read it: "softmax of z, i-th component"

\frac{e ^{z_{i}}}{\sum _{j} e ^{z_{j}}}

— converts logits to a probability vector.

ReLU (z)

ReLU

Read it: "relu of z"

max (0, z)

— the dominant non-linearity in modern networks.

- lo g p (y ∣ x; θ)

Negative log-likelihood

Read it: "NLL"

Loss function for any probabilistic model: minimise the negative log-likelihood.

H (p, q)

Cross-entropy

Read it: "cross entropy of p and q"

- \sum_{x} p (x) lo g q (x)

— classification loss.

Asymptotic notation

Big-O, big-Omega, big-Theta, little-o.

4 items

O (f (n))

Big-O

Read it: "big O of f"

Upper bound: grows no faster than

f

Ω (f (n))

Big-Omega

Read it: "big Omega of f"

Lower bound: grows at least as fast as

f

Θ (f (n))

Big-Theta

Read it: "big Theta of f"

Tight bound: grows at the same rate as

f

o (f (n))

Little-o

Read it: "little o of f"

Strictly slower than

f

—

g (n) / f (n) \to 0

Greek alphabet

The 24 lowercase letters and their canonical conventional uses in ML.

25 items

α

Alpha

Read it: "alpha"

Often: prior parameter (Beta/Dirichlet), learning rate (older texts), Type-I error.

β

Beta

Read it: "beta"

Regression coefficients, Adam's momentum, Type-II error.

γ

Gamma

Read it: "gamma"

Discount factor in RL, RBF kernel width.

δ

Delta

Read it: "delta"

Small change, Dirac delta, Kronecker delta

δ_{ij}

ε

Epsilon

Read it: "epsilon"

Small constant, error term, exploration rate in RL.

ζ

Zeta

Read it: "zeta"

Slack variable in soft-margin SVMs.

η

Eta

Read it: "eta"

Most common letter for the learning rate.

θ

Theta

Read it: "theta"

Generic parameter vector. Also: angle.

ι

Iota

Read it: "iota"

Rare in ML; used in some indexing/embedding texts.

κ

Kappa

Read it: "kappa"

Condition number of a matrix; concentration parameter.

λ

Lambda

Read it: "lambda"

Eigenvalue, regularisation strength, rate of an exponential / Poisson.

μ

Read it: "mu"

Mean (statistics), step size in some optimisers.

ν

Read it: "nu"

Degrees of freedom (Student-

t

), or 'rate' of a hyperprior.

ξ

Read it: "xi"

Slack variable, sample point.

π

Read it: "pi"

Mixture weights, RL policy, and the constant

\approx 3.14159

ρ

Rho

Read it: "rho"

Correlation, density, momentum decay rate.

σ

Sigma

Read it: "sigma"

Standard deviation, sigmoid (yes, the same letter is reused).

τ

Tau

Read it: "tau"

Time constant, temperature in softmax sampling.

ϕ

Phi

Read it: "phi"

Basis function

ϕ (x)

, feature map of a kernel.

χ

Chi

Read it: "chi"

χ^{2}

distribution and test.

ψ

Psi

Read it: "psi"

Digamma function, alternative basis function.

ω

Omega

Read it: "omega"

Angular frequency, sample point of

Ω

Ω

Big Omega (sample space)

Read it: "big omega"

Sample space — the set of all possible outcomes.

Σ

Big Sigma (covariance / sum)

Read it: "big sigma"

Covariance matrix in stats; summation operator

\sum

Φ

Big Phi (Gaussian CDF)

Read it: "big phi"

CDF of the standard normal distribution.