enp 0.1.0

Implementation of deep learning, machine learning and math algorithms in NumPy and pure Python

enp (stands for Everything in NumPy) provides implementation of deep learning, machine learning, and math algorithms in NumPy and pure Python.

Installation

You can install the library using pip:

pip install enp

Description

The library consists of 3 main parts:

• deep learning

  • examples of usage are in examples/nn

  • impementation is in enp/nn

  • loosely follows PyTorch conventions

• machine learning

  • examples of usage are in examples/ml

  • impementation is in enp/ml

  • loosly follows scikit-learn conventions

• linear algebra

  • examples of usage are in tests/test_linear_algebra.py and tests/test_linear_algebra_additional.py

  • impementation is in enp/linalg

Deep Learning

Example of building, prediction, and training a simple neural network consisting of linear layers:

from enp.nn import *

model = Model(learning_rate=0.0075)
input_layer = InputLayer(layer_dim=12288)
linear_1 = Linear(layer_dim=7, prev_layer=input_layer, activation='relu')
linear_2 = Linear(layer_dim=1, prev_layer=linear_1, activation='sigmoid')
cost_layer = BCELoss(layer_dim=1, prev_layer=linear_2)
model.layers = [input_layer, linear_1, linear_2, cost_layer]
model.train(x=train_x, y=train_y, num_iterations=2500)
test_y_pred = model.predict(test_x)

Machine Learning

Example of building, prediction, and training a linear regression:

from enp.ml import LinearRegression
import numpy as np

X = np.array([[1, 1], [1, 2], [2, 2], [2, 3]])
y = np.dot(X, np.array([1, 2])) + 3
reg = LinearRegression().fit(X, y)
prediction = reg.predict(np.array([[3, 5]]))

Linear Algebra

Example of solving a system of linear equations Ax=b:

from enp.linalg import solve_system_of_linear_equations
import numpy as np

a = np.array([[0, -2],
              [4, 7]])
b = np.array([1, 2])
x = solve_system_of_linear_equations(a, b)