Objeck Documentation

All Bundles

Bundle Core machine learning types: the seedable Random generator, Matrix2D linear algebra and activations, matrix reference holders, the NeuralNetwork and CSV-backed MatrixReader. Compile with -lib ml.

Matrix2D

2D matrix operations

Operations

Add
AverageColumn
AverageRow
Concatenate
Divide
DotProduct
DotSigmoid
EuclideanDistance
GetColumn
GetRow
HadamardDivide
HadamardProduct
Identity
Inverse
LeakyReLU
Multiple
Random
RandomNormal
ReLU
Sigmoid
Softmax
Solve
Split
Subtract
SumColumn
SumRow
Tanh
ToBool
ToFloat
ToInt
Transpose

Add # native

Adds a constant to a matrix

function : native : Add(c:Float, m:Float[,]) ~ Float[,]

Parameters

Name	Type	Description
c	Float	constant
m	Float	matrix

Return

Type	Description
Float	updated matrix

Add # native

Adds a constant to a matrix

function : native : Add(m:Float[,], c:Float) ~ Float[,]

Parameters

Name	Type	Description
m	Float	matrix
c	Float	constant

Return

Type	Description
Float	updated matrix

Add # native

Adds two matrices

function : native : Add(m1:Float[,], m2:Float[,]) ~ Float[,]

Parameters

Name	Type	Description
m1	Float	left matrix
m2	Float	right matrix

Return

Type	Description
Float	updated matrix

AverageColumn # native

Average of the given column

function : native : AverageColumn(c:Int, matrix:Float[,]) ~ Float

Parameters

Name	Type	Description
c	Int	column index
matrix	Float	matrix

Return

Type	Description
Float	average of the column

AverageRow # native

Average of the given row

function : native : AverageRow(r:Int, matrix:Float[,]) ~ Float

Parameters

Name	Type	Description
r	Int	row index
matrix	Float	matrix

Return

Type	Description
Float	average of the row

Concatenate # function

Concatenates two matrix

function : Concatenate(a:Float[,], b:Float[,], is_row:Bool) ~ Float[,]

Parameters

Name	Type	Description
a	Float	left matrix
b	Float	right matrix
is_row	Bool	true concatenate by rows, false for columns

Return

Type	Description
Float	concatenated matrix

Divide # native

Divides a constant by a matrix

function : native : Divide(c:Float, m:Float[,]) ~ Float[,]

Parameters

Name	Type	Description
c	Float	constant
m	Float	matrix

Return

Type	Description
Float	updated matrix

Divide # native

Divides a constant by a matrix

function : native : Divide(m:Float[,], c:Float) ~ Float[,]

Parameters

Name	Type	Description
m	Float	matrix
c	Float	constant

Return

Type	Description
Float	updated matrix

DotProduct # native

Calculates the dot product.

function : native : DotProduct(m1:Float[,], m2:Float[,]) ~ Float[,]

Parameters

Name	Type	Description
m1	Float	left matrix
m2	Float	right matrix

Return

Type	Description
Float	updated matrix

DotSigmoid # native

Calculates the Dot Product applying while applying the Sigmoid function to all elements

function : native : DotSigmoid(a:Float[,], b:Float[,]) ~ Float[,]

Parameters

Name	Type	Description
a	Float	matrix
b	Float	matrix

Return

Type	Description
Float	updated matrix

EuclideanDistance # native

Calculates the Euclidean Distance between the two vectors

function : native : EuclideanDistance(m1:Float[,]) ~ Float

Parameters

Name	Type	Description
m1	Float	data matrix

Return

Type	Description
Float	Euclidean Distance between the two vectors

Example

# Two 3D points: [1,4], [2,5], [3,6] paired as columns
points := [[1.0, 4.0], [2.0, 5.0], [3.0, 6.0]];
dist := Matrix2D->EuclideanDistance(points);
"Distance: {$dist}"->PrintLine();

GetColumn # native

Get the column from a matrix

function : native : GetColumn(c:Int, matrix:Float[,]) ~ Float[]

Parameters

Name	Type	Description
c	Int	column index
matrix	Float	matrix

Return

Type	Description
Float	column from matrix

GetRow # native

Get the row from a matrix

function : native : GetRow(r:Int, matrix:Float[,]) ~ Float[]

Parameters

Name	Type	Description
r	Int	row index
matrix	Float	matrix

Return

Type	Description
Float	row from matrix

HadamardDivide # native

Divides two matrices using the Hadamard rule

function : native : HadamardDivide(m1:Float[,], m2:Float[,]) ~ Float[,]

Parameters

Name	Type	Description
m1	Float	left matrix
m2	Float	right matrix

Return

Type	Description
Float	updated matrix

HadamardProduct # native

Multiplies two matrices using the Hadamard rule

function : native : HadamardProduct(m1:Float[,], m2:Float[,]) ~ Float[,]

Parameters

Name	Type	Description
m1	Float	left matrix
m2	Float	right matrix

Return

Type	Description
Float	updated matrix

Identity # native

Creates an identity a matrix

function : native : Identity(s:Int) ~ Float[,]

Parameters

Name	Type	Description
s	Int	size of matrix

Return

Type	Description
Float	row from matrix

Inverse # native

Inverse of matrix

function : native : Inverse(a:Float[,]) ~ Float[,]

Parameters

Name	Type	Description
a	Float	matrix

Return

Type	Description
Float	inverted matrix

LeakyReLU # native

Leaky ReLU activation function

function : native : LeakyReLU(x:Float, alpha:Float) ~ Float

Parameters

Name	Type	Description
x	Float	input value
alpha	Float	slope for negative values (default 0.01)

Return

Type	Description
Float	Leaky ReLU value

LeakyReLU # native

Applies the Leaky ReLU function to all elements

function : native : LeakyReLU(b:Float[,], alpha:Float) ~ Float[,]

Parameters

Name	Type	Description
b	Float	matrix
alpha	Float	slope for negative values

Return

Type	Description
Float	updated matrix

Multiple # native

Multiplies a constant by a matrix

function : native : Multiple(c:Float, m:Float[,]) ~ Float[,]

Parameters

Name	Type	Description
c	Float	constant
m	Float	matrix

Return

Type	Description
Float	updated matrix

Multiple # native

Multiplies a constant by a matrix

function : native : Multiple(m:Float[,], c:Float) ~ Float[,]

Parameters

Name	Type	Description
m	Float	matrix
c	Float	constant

Return

Type	Description
Float	updated matrix

Random # function

Generates a random 2D array of values from 0.0 to 1.0

function : Random(rows:Int, cols:Int) ~ Float[,]

Parameters

Name	Type	Description
rows	Int	rows
cols	Int	columns

Return

Type	Description
Float	updated matrix

RandomNormal # function

Generates a random normal distribution of values

function : RandomNormal(mean:Float, variance:Float, rows:Int, cols:Int) ~ Float[,]

Parameters

Name	Type	Description
mean	Float	center of values
variance	Float	variance in values
rows	Int	rows
cols	Int	columns

Return

Type	Description
Float	updated matrix

RandomNormal # function

Generates a random normal value

function : RandomNormal(mean:Float, variance:Float) ~ Float

Parameters

Name	Type	Description
mean	Float	center of values
variance	Float	variance in values

Return

Type	Description
Float	updated matrix

ReLU # native

ReLU (Rectified Linear Unit) activation function

function : native : ReLU(x:Float) ~ Float

Parameters

Name	Type	Description
x	Float	input value

Return

Type	Description
Float	ReLU value (max(0, x))

Example

"ReLU(-1.5)={$Matrix2D->ReLU(-1.5)}"->PrintLine();
"ReLU(3.2)={$Matrix2D->ReLU(3.2)}"->PrintLine();

ReLU # native

Applies the ReLU function to all elements

function : native : ReLU(b:Float[,]) ~ Float[,]

Parameters

Name	Type	Description
b	Float	matrix

Return

Type	Description
Float	updated matrix

Sigmoid # native

Sigmoid 'S' function

function : native : Sigmoid(x:Float) ~ Float

Parameters

Name	Type	Description
x	Float	input value

Return

Type	Description
Float	Sigmoid value

Example

val := Matrix2D->Sigmoid(2.0);
"Sigmoid(2.0)={$val}"->PrintLine();

Sigmoid # native

Applies the Sigmoid function to all elements

function : native : Sigmoid(b:Float[,]) ~ Float[,]

Parameters

Name	Type	Description
b	Float	matrix

Return

Type	Description
Float	updated matrix

Softmax # native

Softmax activation function for classification

function : native : Softmax(b:Float[,]) ~ Float[,]

Parameters

Name	Type	Description
b	Float	matrix (single row or column vector)

Return

Type	Description
Float	normalized probability distribution

Example

logits := [[1.0], [2.0], [3.0]];
probs := Matrix2D->Softmax(logits);
each(i : probs->Size()[0]) {
  probs[i,0]->PrintLine();
};

Solve # native

Solves a unit of equations

function : native : Solve(m1:Float[,], m2:Float[,]) ~ Float[]

Parameters

Name	Type	Description
m1	Float	data matrix
m2	Float	result matrix

Return

Type	Description
Float	least squares

Example

A := [[1.0, 0.0], [0.0, 1.0], [1.0, 1.0]];
b := [[1.0], [2.0], [3.0]];
coeffs := Matrix2D->Solve(A, b);
each(c in coeffs) {
  c->PrintLine();
};

Split # native

Splits a matrix

function : native : Split(b:Float[,], offset:Int, count:Int, is_row:Bool) ~ Float[,]

Parameters

Name	Type	Description
b	Float	matrix
offset	Int	offset index
count	Int	number of rows to split
is_row	Bool	true for row split, false for column

Return

Type	Description
Float	copied matrix

Subtract # native

Subtracts a constant from a matrix

function : native : Subtract(c:Float, m:Float[,]) ~ Float[,]

Parameters

Name	Type	Description
c	Float	constant
m	Float	matrix

Return

Type	Description
Float	updated matrix

Subtract # native

Adds a constant to a matrix

function : native : Subtract(m:Float[,], c:Float) ~ Float[,]

Parameters

Name	Type	Description
m	Float	matrix
c	Float	constant

Return

Type	Description
Float	updated matrix

Subtract # native

Subtracts two matrices

function : native : Subtract(m1:Float[,], m2:Float[,]) ~ Float[,]

Parameters

Name	Type	Description
m1	Float	left matrix
m2	Float	right matrix

Return

Type	Description
Float	updated matrix

SumColumn # native

Sum of the given column

function : native : SumColumn(c:Int, matrix:Float[,]) ~ Float

Parameters

Name	Type	Description
c	Int	column index
matrix	Float	matrix

Return

Type	Description
Float	sum of the column

SumRow # native

Sum of the given row

function : native : SumRow(r:Int, matrix:Float[,]) ~ Float

Parameters

Name	Type	Description
r	Int	row index
matrix	Float	matrix

Return

Type	Description
Float	sum of the row

Tanh # native

Tanh (Hyperbolic Tangent) activation function

function : native : Tanh(x:Float) ~ Float

Parameters

Name	Type	Description
x	Float	input value

Return

Type	Description
Float	Tanh value

Example

"Tanh(0.5)={$Matrix2D->Tanh(0.5)}"->PrintLine();

Tanh # native

Applies the Tanh function to all elements

function : native : Tanh(b:Float[,]) ~ Float[,]

Parameters

Name	Type	Description
b	Float	matrix

Return

Type	Description
Float	updated matrix

ToBool # function

Parsers a boolean value

function : ToBool(m:Float[,]) ~ Bool

Parameters

Name	Type	Description
m	Float	matrix

Return

Type	Description
Bool	boolean value

ToFloat # function

Parsers a decimal value

function : ToFloat(m:Float[,]) ~ Float

Parameters

Name	Type	Description
m	Float	matrix

Return

Type	Description
Float	decimal value

ToInt # function

Parsers an integer value

function : ToInt(m:Float[,]) ~ Int

Parameters

Name	Type	Description
m	Float	matrix

Return

Type	Description
Int	integer value

Transpose # native

Transpose of matrix

function : native : Transpose(a:Float[,]) ~ Float[,]

Parameters

Name	Type	Description
a	Float	matrix

Return

Type	Description
Float	transposed matrix