Created
August 15, 2020 10:59
-
-
Save arghyadeep99/aeee03bcb9f8303cd8f3ca1a438fe675 to your computer and use it in GitHub Desktop.
Gradient Descent from Scratch.ipynb
This file contains bidirectional Unicode text that may be interpreted or compiled differently than what appears below. To review, open the file in an editor that reveals hidden Unicode characters.
Learn more about bidirectional Unicode characters
{ | |
"nbformat": 4, | |
"nbformat_minor": 0, | |
"metadata": { | |
"colab": { | |
"name": "Gradient Descent from Scratch.ipynb", | |
"provenance": [], | |
"collapsed_sections": [], | |
"authorship_tag": "ABX9TyP/LvgkKOxKcl+NTdbKf6PB", | |
"include_colab_link": true | |
}, | |
"kernelspec": { | |
"name": "python3", | |
"display_name": "Python 3" | |
} | |
}, | |
"cells": [ | |
{ | |
"cell_type": "markdown", | |
"metadata": { | |
"id": "view-in-github", | |
"colab_type": "text" | |
}, | |
"source": [ | |
"<a href=\"https://colab.research.google.com/gist/arghyadeep99/aeee03bcb9f8303cd8f3ca1a438fe675/gradient-descent-from-scratch.ipynb\" target=\"_parent\"><img src=\"https://colab.research.google.com/assets/colab-badge.svg\" alt=\"Open In Colab\"/></a>" | |
] | |
}, | |
{ | |
"cell_type": "code", | |
"metadata": { | |
"id": "CxJJ6cXRamL9", | |
"colab_type": "code", | |
"colab": {} | |
}, | |
"source": [ | |
"import numpy as np\n", | |
"\n", | |
"class NeuralNetwork:\n", | |
" def __init__(self):\n", | |
" np.random.seed(42)\n", | |
" self.Wij = np.random.rand(4,5)\n", | |
" self.Wjk = np.random.rand(5,1)\n", | |
"\n", | |
" def sigmoid(self, x, w):\n", | |
" z = np.dot(x,w)\n", | |
" return 1/(1 + np.exp(-z))\n", | |
" \n", | |
" def diff_sigmoid(self, x, w):\n", | |
" return self.sigmoid(x,w)*(1-self.sigmoid(x,w))\n", | |
" \n", | |
" def gradient_descent(self,x,y, epochs):\n", | |
" for i in range(epochs):\n", | |
" Xi = x #Input layer\n", | |
" Xj = self.sigmoid(Xi, self.Wij) #Hidden layer\n", | |
" Y_cap = self.sigmoid(Xj, self.Wjk) #Final layer\n", | |
" grad_Wjk = np.dot(Xj.T, (y-Y_cap)*self.diff_sigmoid(Xj, self.Wjk))\n", | |
" grad_Wij = np.dot(Xi.T, np.dot((y-Y_cap)* self.diff_sigmoid(Xj, self.Wjk), self.Wjk.T) * self.diff_sigmoid(Xi, self.Wij))\n", | |
"\n", | |
" self.Wjk += grad_Wjk\n", | |
" self.Wij += grad_Wij\n", | |
"\n", | |
" print('The final prediction from neural network are: ')\n", | |
" print(Y_cap)" | |
], | |
"execution_count": null, | |
"outputs": [] | |
}, | |
{ | |
"cell_type": "code", | |
"metadata": { | |
"id": "6gYCR_7B1xM6", | |
"colab_type": "code", | |
"colab": { | |
"base_uri": "https://localhost:8080/", | |
"height": 306 | |
}, | |
"outputId": "ed9ace23-2948-4cb0-c425-acf75d6fdec3" | |
}, | |
"source": [ | |
"if __name__ == '__main__':\n", | |
" NN = NeuralNetwork()\n", | |
" print('Random starting input to hidden weights: ')\n", | |
" print(NN.Wij)\n", | |
" print('Random starting hidden to output weights: ')\n", | |
" print(NN.Wjk)\n", | |
" X = np.array([[0, 0, 1, 1], [1, 1, 0, 1], [1, 0, 1, 1], [0, 1, 0, 1], [0, 0, 0, 1]])\n", | |
" y = np.array([[0, 1, 1, 0, 1]]).T\n", | |
" neural_network.gradient_descent(X, y, 10000)" | |
], | |
"execution_count": null, | |
"outputs": [ | |
{ | |
"output_type": "stream", | |
"text": [ | |
"Random starting input to hidden weights: \n", | |
"[[0.37454012 0.95071431 0.73199394 0.59865848 0.15601864]\n", | |
" [0.15599452 0.05808361 0.86617615 0.60111501 0.70807258]\n", | |
" [0.02058449 0.96990985 0.83244264 0.21233911 0.18182497]\n", | |
" [0.18340451 0.30424224 0.52475643 0.43194502 0.29122914]]\n", | |
"Random starting hidden to output weights: \n", | |
"[[0.61185289]\n", | |
" [0.13949386]\n", | |
" [0.29214465]\n", | |
" [0.36636184]\n", | |
" [0.45606998]]\n", | |
"The final prediction from neural network are: \n", | |
"[[0.00589178]\n", | |
" [0.99658221]\n", | |
" [0.99669157]\n", | |
" [0.00609915]\n", | |
" [0.99361295]]\n" | |
], | |
"name": "stdout" | |
} | |
] | |
} | |
] | |
} |
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment