FontRecognizer.ipynb

{
  "nbformat": 4,
  "nbformat_minor": 0,
  "metadata": {
    "colab": {
      "name": "FontRecognizer.ipynb",
      "version": "0.3.2",
      "provenance": [],
      "collapsed_sections": []
    },
    "kernelspec": {
      "name": "python3",
      "display_name": "Python 3"
    },
    "accelerator": "GPU"
  },
  "cells": [
    {
      "metadata": {
        "id": "Es8q77vE6tzK",
        "colab_type": "code",
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 53
        },
        "outputId": "11713445-945f-4489-caf9-044207052923"
      },
      "cell_type": "code",
      "source": [
        "!mkdir -p drive\n",
        "!google-drive-ocamlfuse drive"
      ],
      "execution_count": 1,
      "outputs": [
        {
          "output_type": "stream",
          "text": [
            "fuse: mountpoint is not empty\r\n",
            "fuse: if you are sure this is safe, use the 'nonempty' mount option\r\n"
          ],
          "name": "stdout"
        }
      ]
    },
    {
      "metadata": {
        "id": "w4D3o0NC63qB",
        "colab_type": "code",
        "colab": {}
      },
      "cell_type": "code",
      "source": [
        "import os\n",
        "os.getcwd()\n",
        "os.chdir(\"drive/Colab_Work/Mclain\")"
      ],
      "execution_count": 0,
      "outputs": []
    },
    {
      "metadata": {
        "id": "JqxHDMRt6pJm",
        "colab_type": "code",
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 971
        },
        "outputId": "6b9683dd-c1e1-4e29-dd64-0adc04c71a5f"
      },
      "cell_type": "code",
      "source": [
        "import numpy as np\n",
        "import pandas as pd\n",
        "from NeuralNetwork import NeuralNetwork\n",
        "\n",
        "df_train = pd.read_csv('devnagri_train.csv', header=None)\n",
        "df_test = pd.read_csv('devnagri_test_public.csv', header=None)\n",
        "\n",
        "X = np.array(df_train.iloc[:, 1:]).T\n",
        "X = X / 255\n",
        "X_test = np.array(df_test.iloc[:, 1:]).T\n",
        "X_test = X_test / 255\n",
        "print(X.shape, X_test.shape)\n",
        "\n",
        "Y = np.array(df_train.iloc[:, :1]).T\n",
        "Y_test = np.array(df_test.iloc[:, :1]).T\n",
        "# One-hot encode the target\n",
        "num_classes = len(np.unique(Y))\n",
        "print(num_classes)\n",
        "Y = np.eye(num_classes)[Y.reshape(-1)].T\n",
        "Y_test = np.eye(num_classes)[Y_test.reshape(-1)].T\n",
        "print(Y.shape, Y_test.shape)\n",
        "\n",
        "num_inputs = X.shape[0]\n",
        "num_outputs = Y.shape[0]  # = num_classes\n",
        "hidden_archi = [128, 64]\n",
        "\n",
        "font_nn = NeuralNetwork(input_size = num_inputs,\n",
        "                        output_size = num_outputs,\n",
        "                        hidden_layer_sizes = hidden_archi,\n",
        "                        activation = \"relu\")\n",
        "\n",
        "font_nn.fit(X, Y, 0.01, 500, True, 32)\n",
        "\n",
        "Y_pred = font_nn.predict(X)"
      ],
      "execution_count": 5,
      "outputs": [
        {
          "output_type": "stream",
          "text": [
            "(1024, 78200) (1024, 6900)\n",
            "46\n",
            "(46, 78200) (46, 6900)\n",
            "Iteration: 0 Cost: 5.267337731478517\n",
            "Iteration: 10 Cost: 4.48399292918797\n",
            "Iteration: 20 Cost: 4.467870241707459\n",
            "Iteration: 30 Cost: 4.490718680557457\n",
            "Iteration: 40 Cost: 4.477540353434241\n",
            "Iteration: 50 Cost: 4.48153470656579\n",
            "Iteration: 60 Cost: 4.480117654114061\n",
            "Iteration: 70 Cost: 4.490402710697337\n",
            "Iteration: 80 Cost: 4.480025309564043\n",
            "Iteration: 90 Cost: 4.492013571306273\n",
            "Iteration: 100 Cost: 4.488901026468319\n",
            "Iteration: 110 Cost: 4.503883120365828\n",
            "Iteration: 120 Cost: 4.475502057444009\n",
            "Iteration: 130 Cost: 4.4807009513640566\n",
            "Iteration: 140 Cost: 4.482584430992461\n",
            "Iteration: 150 Cost: 4.474785601768024\n",
            "Iteration: 160 Cost: 4.48314485296304\n",
            "Iteration: 170 Cost: 4.491389021047398\n",
            "Iteration: 180 Cost: 4.466951726828516\n",
            "Iteration: 190 Cost: 4.475218179886051\n",
            "Iteration: 200 Cost: 4.481622615306707\n",
            "Iteration: 210 Cost: 4.472416983812134\n",
            "Iteration: 220 Cost: 4.484011069664665\n",
            "Iteration: 230 Cost: 4.478216898254977\n",
            "Iteration: 240 Cost: 4.491872638509461\n",
            "Iteration: 250 Cost: 4.481833877063746\n",
            "Iteration: 260 Cost: 4.483115596155344\n",
            "Iteration: 270 Cost: 4.483138538956885\n",
            "Iteration: 280 Cost: 4.479264818212033\n",
            "Iteration: 290 Cost: 4.479042271584366\n",
            "Iteration: 300 Cost: 4.475502057444009\n",
            "Iteration: 310 Cost: 4.480086458438926\n",
            "Iteration: 320 Cost: 4.4849421708204975\n",
            "Iteration: 330 Cost: 4.485943866610354\n",
            "Iteration: 340 Cost: 4.454454745002078\n",
            "Iteration: 350 Cost: 4.4892013025369035\n",
            "Iteration: 360 Cost: 4.471896912779246\n",
            "Iteration: 370 Cost: 4.474259955648323\n",
            "Iteration: 380 Cost: 4.488871108047748\n",
            "Iteration: 390 Cost: 4.475974541106307\n",
            "Iteration: 400 Cost: 4.47906673081048\n",
            "Iteration: 410 Cost: 4.478374173078139\n",
            "Iteration: 420 Cost: 4.469549025449136\n",
            "Iteration: 430 Cost: 4.482826620689021\n",
            "Iteration: 440 Cost: 4.491749204872773\n",
            "Iteration: 450 Cost: 4.49802241159481\n",
            "Iteration: 460 Cost: 4.482599167754104\n",
            "Iteration: 470 Cost: 4.495455099923397\n",
            "Iteration: 480 Cost: 4.4751842110737385\n",
            "Iteration: 490 Cost: 4.487634442042649\n"
          ],
          "name": "stdout"
        }
      ]
    },
    {
      "metadata": {
        "id": "5tYwCKTB7arP",
        "colab_type": "code",
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 917
        },
        "outputId": "acaf9aca-1e23-4024-e5c9-e3dbedb61753"
      },
      "cell_type": "code",
      "source": [
        "font_nn.costs"
      ],
      "execution_count": 6,
      "outputs": [
        {
          "output_type": "execute_result",
          "data": {
            "text/plain": [
              "[5.267337731478517,\n",
              " 4.48399292918797,\n",
              " 4.467870241707459,\n",
              " 4.490718680557457,\n",
              " 4.477540353434241,\n",
              " 4.48153470656579,\n",
              " 4.480117654114061,\n",
              " 4.490402710697337,\n",
              " 4.480025309564043,\n",
              " 4.492013571306273,\n",
              " 4.488901026468319,\n",
              " 4.503883120365828,\n",
              " 4.475502057444009,\n",
              " 4.4807009513640566,\n",
              " 4.482584430992461,\n",
              " 4.474785601768024,\n",
              " 4.48314485296304,\n",
              " 4.491389021047398,\n",
              " 4.466951726828516,\n",
              " 4.475218179886051,\n",
              " 4.481622615306707,\n",
              " 4.472416983812134,\n",
              " 4.484011069664665,\n",
              " 4.478216898254977,\n",
              " 4.491872638509461,\n",
              " 4.481833877063746,\n",
              " 4.483115596155344,\n",
              " 4.483138538956885,\n",
              " 4.479264818212033,\n",
              " 4.479042271584366,\n",
              " 4.475502057444009,\n",
              " 4.480086458438926,\n",
              " 4.4849421708204975,\n",
              " 4.485943866610354,\n",
              " 4.454454745002078,\n",
              " 4.4892013025369035,\n",
              " 4.471896912779246,\n",
              " 4.474259955648323,\n",
              " 4.488871108047748,\n",
              " 4.475974541106307,\n",
              " 4.47906673081048,\n",
              " 4.478374173078139,\n",
              " 4.469549025449136,\n",
              " 4.482826620689021,\n",
              " 4.491749204872773,\n",
              " 4.49802241159481,\n",
              " 4.482599167754104,\n",
              " 4.495455099923397,\n",
              " 4.4751842110737385,\n",
              " 4.487634442042649]"
            ]
          },
          "metadata": {
            "tags": []
          },
          "execution_count": 6
        }
      ]
    },
    {
      "metadata": {
        "id": "Dma2wKB-SnF_",
        "colab_type": "code",
        "colab": {}
      },
      "cell_type": "code",
      "source": [
        "import matplotlib.pyplot as plt"
      ],
      "execution_count": 0,
      "outputs": []
    },
    {
      "metadata": {
        "id": "48yJltwmSnJ6",
        "colab_type": "code",
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 361
        },
        "outputId": "bd2723ef-031d-498f-b97c-1ccd4d3dce41"
      },
      "cell_type": "code",
      "source": [
        "plt.plot(range(0, 500, 10), font_nn.costs)\n",
        "plt.xlabel(\"No. of Epochs\")\n",
        "plt.ylabel(\"Mean Squared Error\")\n",
        "plt.ylim((4, 5.5))\n",
        "plt.show()"
      ],
      "execution_count": 11,
      "outputs": [
        {
          "output_type": "display_data",
          "data": {
            "image/png": "iVBORw0KGgoAAAANSUhEUgAAAe8AAAFYCAYAAAB6RnQAAAAABHNCSVQICAgIfAhkiAAAAAlwSFlz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\n",
            "text/plain": [
              "<matplotlib.figure.Figure at 0x7ff039afb6d8>"
            ]
          },
          "metadata": {
            "tags": []
          }
        }
      ]
    },
    {
      "metadata": {
        "id": "l6hDEa-SSnOv",
        "colab_type": "code",
        "colab": {}
      },
      "cell_type": "code",
      "source": [
        ""
      ],
      "execution_count": 0,
      "outputs": []
    },
    {
      "metadata": {
        "id": "5n1O_-UySnNU",
        "colab_type": "code",
        "colab": {}
      },
      "cell_type": "code",
      "source": [
        ""
      ],
      "execution_count": 0,
      "outputs": []
    }
  ]
}