mirror of
https://github.com/TheAlgorithms/Python.git
synced 2024-11-30 16:31:08 +00:00
bc8df6de31
* [pre-commit.ci] pre-commit autoupdate updates: - [github.com/astral-sh/ruff-pre-commit: v0.2.2 → v0.3.2](https://github.com/astral-sh/ruff-pre-commit/compare/v0.2.2...v0.3.2) - [github.com/pre-commit/mirrors-mypy: v1.8.0 → v1.9.0](https://github.com/pre-commit/mirrors-mypy/compare/v1.8.0...v1.9.0) * [pre-commit.ci] auto fixes from pre-commit.com hooks for more information, see https://pre-commit.ci --------- Co-authored-by: pre-commit-ci[bot] <66853113+pre-commit-ci[bot]@users.noreply.github.com>
49 lines
1.6 KiB
Python
49 lines
1.6 KiB
Python
"""
|
|
This script implements the Soboleva Modified Hyperbolic Tangent function.
|
|
|
|
The function applies the Soboleva Modified Hyperbolic Tangent function
|
|
to each element of the vector.
|
|
|
|
More details about the activation function can be found on:
|
|
https://en.wikipedia.org/wiki/Soboleva_modified_hyperbolic_tangent
|
|
"""
|
|
|
|
import numpy as np
|
|
|
|
|
|
def soboleva_modified_hyperbolic_tangent(
|
|
vector: np.ndarray, a_value: float, b_value: float, c_value: float, d_value: float
|
|
) -> np.ndarray:
|
|
"""
|
|
Implements the Soboleva Modified Hyperbolic Tangent function
|
|
|
|
Parameters:
|
|
vector (ndarray): A vector that consists of numeric values
|
|
a_value (float): parameter a of the equation
|
|
b_value (float): parameter b of the equation
|
|
c_value (float): parameter c of the equation
|
|
d_value (float): parameter d of the equation
|
|
|
|
Returns:
|
|
vector (ndarray): Input array after applying SMHT function
|
|
|
|
>>> vector = np.array([5.4, -2.4, 6.3, -5.23, 3.27, 0.56])
|
|
>>> soboleva_modified_hyperbolic_tangent(vector, 0.2, 0.4, 0.6, 0.8)
|
|
array([ 0.11075085, -0.28236685, 0.07861169, -0.1180085 , 0.22999056,
|
|
0.1566043 ])
|
|
"""
|
|
|
|
# Separate the numerator and denominator for simplicity
|
|
# Calculate the numerator and denominator element-wise
|
|
numerator = np.exp(a_value * vector) - np.exp(-b_value * vector)
|
|
denominator = np.exp(c_value * vector) + np.exp(-d_value * vector)
|
|
|
|
# Calculate and return the final result element-wise
|
|
return numerator / denominator
|
|
|
|
|
|
if __name__ == "__main__":
|
|
import doctest
|
|
|
|
doctest.testmod()
|