Help for package SBOAtools

Type:

Package

Title:

Secretary Bird Optimization for Continuous Optimization and Neural Network Training

Version:

0.1.1

Description:

Provides an implementation of Secretary Bird Optimization for general-purpose continuous optimization, benchmark optimization, and training single-hidden-layer feed-forward neural network models. The implemented optimizer is based on the Secretary Bird Optimization Algorithm proposed by Fu et al. (2024) <doi:10.1007/s10462-024-10729-y>. The neural network training functionality is based on Dilber and Özdemir (2026) <doi:10.1007/s00521-026-11874-x>.

License:

MIT + file LICENSE

URL:

https://github.com/burakdilber/SBOAtools

BugReports:

https://github.com/burakdilber/SBOAtools/issues

Encoding:

UTF-8

RoxygenNote:

7.3.3

Imports:

stats, graphics

NeedsCompilation:

Packaged:

2026-05-02 15:20:24 UTC; hp

Author:

Burak Dilber [aut, cre, cph], A. Fırat Özdemir [aut, cph]

Maintainer:

Burak Dilber <burakdilber91@gmail.com>

Repository:

CRAN

Date/Publication:

2026-05-03 06:30:17 UTC

Apply min-max normalization using existing bounds

Description

Apply min-max normalization using existing bounds

Usage

apply_minmax(x, mins, maxs)

Arguments

x

Numeric vector or matrix.

mins

Column minima.

maxs

Column maxima.

Value

Scaled matrix.

Clip values to lower and upper bounds

Description

Restricts each element of a vector to the given bounds.

Usage

clip_bounds(x, lower, upper)

Arguments

x

Numeric vector.

lower

Lower bound vector.

upper

Upper bound vector.

Value

A bounded numeric vector.

Reverse min-max normalization

Description

Reverse min-max normalization

Usage

denormalize_minmax(x_scaled, mins, maxs)

Arguments

x_scaled

Scaled numeric vector or matrix.

mins

Column minima.

maxs

Column maxima.

Value

Values on the original scale.

Forward pass for a single-hidden-layer MLP

Description

Computes hidden-layer activations and output predictions.

Usage

forward_mlp(params, X, input_dim, hidden_dim, output_dim)

Arguments

params

Numeric parameter vector.

X

Input matrix.

input_dim

Number of input variables.

hidden_dim

Number of hidden neurons.

output_dim

Number of output neurons.

Value

A list containing hidden activations and predictions.

Get a built-in benchmark definition

Description

Returns a built-in benchmark definition by name.

Usage

get_benchmark(name)

Arguments

name

Benchmark name such as "F1" or "F14".

Value

A list containing the benchmark function and its metadata.

Examples

b <- get_benchmark("F1")
b$fn(rep(0, 5))

Lévy flight step generator

Description

Generates Lévy-distributed random steps.

Usage

levy_flight(d, beta = 1.5)

Arguments

d

Dimension of the step vector.

beta

Stability parameter.

Value

A numeric vector of length d.

List built-in benchmark functions

Description

Returns a summary table of the built-in F1-F23 benchmark functions.

Usage

list_benchmarks()

Value

A data frame.

Examples

list_benchmarks()

Mean absolute error

Description

Computes the mean absolute error between actual and predicted values.

Usage

mae_vec(actual, predicted)

Arguments

actual

Numeric vector or matrix of observed values.

predicted

Numeric vector or matrix of predicted values.

Value

A numeric scalar.

Mean absolute percentage error

Description

Computes the mean absolute percentage error between actual and predicted values.

Usage

mape_vec(actual, predicted, eps = 1e-08)

Arguments

actual

Numeric vector or matrix of observed values.

predicted

Numeric vector or matrix of predicted values.

eps

Small constant to avoid division by zero.

Value

A numeric scalar.

MSE fitness function for MLP training

Description

Computes mean squared error for a single-hidden-layer MLP.

Usage

mlp_mse_fitness(params, X, Y, input_dim, hidden_dim, output_dim)

Arguments

params

Numeric parameter vector.

X

Input matrix.

Y

Output matrix.

input_dim

Number of input variables.

hidden_dim

Number of hidden neurons.

output_dim

Number of output neurons.

Value

Mean squared error.

Min-max normalization

Description

Scales each column of a matrix to the interval from 0 to 1.

Usage

normalize_minmax(x)

Arguments

x

Numeric vector or matrix.

Value

A list containing scaled data and min-max values.

Plot method for SBOA objects

Description

Plots the convergence curve of an SBOA optimization result.

Usage

## S3 method for class 'sboa'
plot(x, ...)

Arguments

x

An object of class "sboa".

...

Additional graphical arguments passed to graphics::plot().

Value

No return value. Called for its side effect.

Plot method for SBOA-MLP objects

Description

Plots the convergence curve of a trained SBOA-MLP model.

Usage

## S3 method for class 'sboa_mlp'
plot(x, ...)

Arguments

x

An object of class "sboa_mlp".

...

Additional graphical arguments passed to graphics::plot().

Value

No return value. Called for its side effect.

Predict method for SBOA-MLP objects

Description

Generates predictions from a trained SBOA-MLP model.

Usage

## S3 method for class 'sboa_mlp'
predict(object, newdata, ...)

Arguments

object

An object of class "sboa_mlp".

newdata

New predictor data.

...

Additional arguments, ignored.

Value

A matrix of predicted values.

Print method for SBOA objects

Description

Prints a summary of an SBOA optimization result.

Usage

## S3 method for class 'sboa'
print(x, ...)

Arguments

x

An object of class "sboa".

...

Additional arguments, ignored.

Value

The input object, invisibly.

Print method for SBOA-MLP objects

Description

Prints a summary of a trained SBOA-MLP model.

Usage

## S3 method for class 'sboa_mlp'
print(x, ...)

Arguments

x

An object of class "sboa_mlp".

...

Additional arguments, ignored.

Value

The input object, invisibly.

Root mean squared error

Description

Computes the root mean squared error between actual and predicted values.

Usage

rmse_vec(actual, predicted)

Arguments

actual

Numeric vector or matrix of observed values.

predicted

Numeric vector or matrix of predicted values.

Value

A numeric scalar.

R-squared

Description

Computes the coefficient of determination.

Usage

rsq_vec(actual, predicted)

Arguments

actual

Numeric vector or matrix of observed values.

predicted

Numeric vector or matrix of predicted values.

Value

A numeric scalar.

Secretary Bird Optimization Algorithm

Description

General-purpose continuous optimization using the Secretary Bird Optimization Algorithm (SBOA).

Usage

sboa(
  fn,
  lower,
  upper,
  n_agents = 30,
  max_iter = 500,
  ...,
  verbose = TRUE,
  seed = NULL
)

Arguments

fn

Objective function to be minimized, or a character string naming a built-in benchmark function such as "F1".

lower

Lower bounds for decision variables.

upper

Upper bounds for decision variables.

n_agents

Number of search agents.

max_iter

Maximum number of iterations.

...

Additional arguments passed to fn.

verbose

Logical; if TRUE, progress is printed.

seed

Optional random seed.

Value

An object of class "sboa".

References

Fu, W., Wang, K., Liu, J., et al. (2024). Secretary Bird Optimization Algorithm. Artificial Intelligence Review. DOI: 10.1007/s10462-024-10729-y

Examples

sphere <- function(x) sum(x^2)

res <- sboa(
  fn = sphere,
  lower = rep(-10, 5),
  upper = rep(10, 5),
  n_agents = 10,
  max_iter = 20,
  seed = 123,
  verbose = FALSE
)

res2 <- sboa(
  fn = "F1",
  lower = rep(-100, 5),
  upper = rep(100, 5),
  n_agents = 10,
  max_iter = 20,
  seed = 123,
  verbose = FALSE
)

print(res)
print(res2)
list_benchmarks()
get_benchmark("F9")

Train a single-hidden-layer MLP using SBOA

Description

Trains a single-hidden-layer multilayer perceptron with the Secretary Bird Optimization Algorithm.

Usage

sboa_mlp(
  X_train,
  y_train,
  hidden_dim = 10,
  n_agents = 30,
  max_iter = 500,
  lower = -1,
  upper = 1,
  seed = NULL,
  verbose = TRUE
)

Arguments

X_train

Training input data.

y_train

Training output data.

hidden_dim

Number of hidden neurons.

n_agents

Number of search agents.

max_iter

Maximum number of iterations.

lower

Lower bound for parameter search.

upper

Upper bound for parameter search.

seed

Optional random seed.

verbose

Logical; if TRUE, progress is printed.

Value

An object of class "sboa_mlp".

References

Dilber, B., and Ozdemir, A. F. (2026). A novel approach to training feed-forward multi-layer perceptrons with recently proposed secretary bird optimization algorithm. Neural Computing and Applications. DOI: 10.1007/s00521-026-11874-x

Examples

set.seed(123)

X_train <- matrix(runif(40), nrow = 10, ncol = 4)
y_train <- matrix(runif(10), nrow = 10, ncol = 1)

fit <- sboa_mlp(
  X_train = X_train,
  y_train = y_train,
  hidden_dim = 3,
  n_agents = 10,
  max_iter = 20,
  lower = -1,
  upper = 1,
  seed = 123,
  verbose = FALSE
)

print(fit)
pred <- predict(fit, X_train)
head(pred)

Sigmoid activation function

Description

Computes the sigmoid transformation.

Usage

sigmoid(x)

Arguments

x

A numeric vector, matrix, or scalar.

Value

Transformed values in the interval (0, 1).

Unpack MLP parameter vector

Description

Converts a flat parameter vector into weight matrices and bias vectors for a single-hidden-layer multilayer perceptron.

Usage

unpack_mlp_params(params, input_dim, hidden_dim, output_dim)

Arguments

params

Numeric parameter vector.

input_dim

Number of input variables.

hidden_dim

Number of hidden neurons.

output_dim

Number of output neurons.

Value

A list containing W1, b1, W2, and b2.