Tutorials/_Project_2_auto_encoder.ipynb

{
 "cells": [
  {
   "cell_type": "code",
   "execution_count": 1,
   "metadata": {},
   "outputs": [],
   "source": [
    "import numpy as np \n",
    "import pandas as pd\n",
    "import matplotlib.pyplot as plt\n",
    "\n",
    "from miniml.ann import *"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 2,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/html": [
       "<div>\n",
       "<style scoped>\n",
       "    .dataframe tbody tr th:only-of-type {\n",
       "        vertical-align: middle;\n",
       "    }\n",
       "\n",
       "    .dataframe tbody tr th {\n",
       "        vertical-align: top;\n",
       "    }\n",
       "\n",
       "    .dataframe thead th {\n",
       "        text-align: right;\n",
       "    }\n",
       "</style>\n",
       "<table border=\"1\" class=\"dataframe\">\n",
       "  <thead>\n",
       "    <tr style=\"text-align: right;\">\n",
       "      <th></th>\n",
       "      <th>label</th>\n",
       "      <th>pixel0</th>\n",
       "      <th>pixel1</th>\n",
       "      <th>pixel2</th>\n",
       "      <th>pixel3</th>\n",
       "      <th>pixel4</th>\n",
       "      <th>pixel5</th>\n",
       "      <th>pixel6</th>\n",
       "      <th>pixel7</th>\n",
       "      <th>pixel8</th>\n",
       "      <th>...</th>\n",
       "      <th>pixel774</th>\n",
       "      <th>pixel775</th>\n",
       "      <th>pixel776</th>\n",
       "      <th>pixel777</th>\n",
       "      <th>pixel778</th>\n",
       "      <th>pixel779</th>\n",
       "      <th>pixel780</th>\n",
       "      <th>pixel781</th>\n",
       "      <th>pixel782</th>\n",
       "      <th>pixel783</th>\n",
       "    </tr>\n",
       "  </thead>\n",
       "  <tbody>\n",
       "    <tr>\n",
       "      <th>0</th>\n",
       "      <td>1</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>...</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>1</th>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>...</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>2</th>\n",
       "      <td>1</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>...</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>3</th>\n",
       "      <td>4</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>...</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>4</th>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>...</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "    </tr>\n",
       "  </tbody>\n",
       "</table>\n",
       "<p>5 rows × 785 columns</p>\n",
       "</div>"
      ],
      "text/plain": [
       "   label  pixel0  pixel1  pixel2  pixel3  pixel4  pixel5  pixel6  pixel7  \\\n",
       "0      1       0       0       0       0       0       0       0       0   \n",
       "1      0       0       0       0       0       0       0       0       0   \n",
       "2      1       0       0       0       0       0       0       0       0   \n",
       "3      4       0       0       0       0       0       0       0       0   \n",
       "4      0       0       0       0       0       0       0       0       0   \n",
       "\n",
       "   pixel8  ...  pixel774  pixel775  pixel776  pixel777  pixel778  pixel779  \\\n",
       "0       0  ...         0         0         0         0         0         0   \n",
       "1       0  ...         0         0         0         0         0         0   \n",
       "2       0  ...         0         0         0         0         0         0   \n",
       "3       0  ...         0         0         0         0         0         0   \n",
       "4       0  ...         0         0         0         0         0         0   \n",
       "\n",
       "   pixel780  pixel781  pixel782  pixel783  \n",
       "0         0         0         0         0  \n",
       "1         0         0         0         0  \n",
       "2         0         0         0         0  \n",
       "3         0         0         0         0  \n",
       "4         0         0         0         0  \n",
       "\n",
       "[5 rows x 785 columns]"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    }
   ],
   "source": [
    "train = pd.read_csv('./datasets/digit-recognizer/train.csv')\n",
    "training_data = pd.read_csv('./datasets/digit-recognizer/train.csv')\n",
    "\n",
    "display(train.head())\n",
    "train_label = train['label']\n",
    "train = train.drop(['label'], axis=1)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 16,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Epoch: 1, Loss: 1.5709\n",
      "Epoch: 2, Loss: 0.0681\n",
      "Epoch: 3, Loss: 0.0685\n",
      "Epoch: 4, Loss: 0.0688\n",
      "Epoch: 5, Loss: 0.0691\n",
      "Epoch: 6, Loss: 0.0695\n",
      "Epoch: 7, Loss: 0.0693\n",
      "Epoch: 8, Loss: 0.0690\n",
      "Epoch: 9, Loss: 0.0686\n",
      "Epoch: 10, Loss: 0.0676\n",
      "Epoch: 11, Loss: 0.0644\n",
      "Epoch: 12, Loss: 0.0632\n",
      "Epoch: 13, Loss: 0.0627\n",
      "Epoch: 14, Loss: 0.0616\n",
      "Epoch: 15, Loss: 0.0610\n",
      "Epoch: 16, Loss: 0.0601\n",
      "Epoch: 17, Loss: 0.0573\n",
      "Epoch: 18, Loss: 0.0541\n",
      "Epoch: 19, Loss: 0.0519\n",
      "Epoch: 20, Loss: 0.0508\n",
      "Epoch: 21, Loss: 0.0498\n",
      "Epoch: 22, Loss: 0.0484\n",
      "Epoch: 23, Loss: 0.0470\n",
      "Epoch: 24, Loss: 0.0453\n",
      "Epoch: 25, Loss: 0.0441\n",
      "Epoch: 26, Loss: 0.0431\n",
      "Epoch: 27, Loss: 0.0424\n",
      "Epoch: 28, Loss: 0.0410\n",
      "Epoch: 29, Loss: 0.0398\n",
      "Epoch: 30, Loss: 0.0389\n",
      "Epoch: 31, Loss: 0.0381\n",
      "Epoch: 32, Loss: 0.0372\n",
      "Epoch: 33, Loss: 0.0364\n",
      "Epoch: 34, Loss: 0.0358\n",
      "Epoch: 35, Loss: 0.0348\n",
      "Epoch: 36, Loss: 0.0340\n",
      "Epoch: 37, Loss: 0.0335\n",
      "Epoch: 38, Loss: 0.0330\n",
      "Epoch: 39, Loss: 0.0324\n",
      "Epoch: 40, Loss: 0.0319\n",
      "Epoch: 41, Loss: 0.0315\n",
      "Epoch: 42, Loss: 0.0310\n",
      "Epoch: 43, Loss: 0.0307\n",
      "Epoch: 44, Loss: 0.0303\n",
      "Epoch: 45, Loss: 0.0299\n",
      "Epoch: 46, Loss: 0.0296\n",
      "Epoch: 47, Loss: 0.0292\n",
      "Epoch: 48, Loss: 0.0290\n",
      "Epoch: 49, Loss: 0.0287\n",
      "Epoch: 50, Loss: 0.0285\n"
     ]
    }
   ],
   "source": [
    "X = train.to_numpy(dtype=np.float64)/255\n",
    "\n",
    "ann = ANN(X, X, 'regressor')\n",
    "param = ann.fit(hidden_layer=[256, 128, 20, 128, 256], \n",
    "                optimizer='Adam', lr=1e-2, epochs=50, batch_size=128, verbose=True)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 4,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<Figure size 432x288 with 1 Axes>"
      ]
     },
     "metadata": {
      "needs_background": "light"
     },
     "output_type": "display_data"
    }
   ],
   "source": [
    "losses = param[-1]\n",
    "_ = plt.plot(losses[1:])"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 20,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<Figure size 432x288 with 1 Axes>"
      ]
     },
     "metadata": {
      "needs_background": "light"
     },
     "output_type": "display_data"
    }
   ],
   "source": [
    "graph = ann.predict(X[8:12]).reshape(4,28,28)\n",
    "_ = plt.imshow(np.vstack((np.hstack(graph[:2]), np.hstack(graph[2:4]))))"
   ]
  }
 ],
 "metadata": {
  "kernelspec": {
   "display_name": "Python 3",
   "language": "python",
   "name": "python3"
  },
  "language_info": {
   "codemirror_mode": {
    "name": "ipython",
    "version": 3
   },
   "file_extension": ".py",
   "mimetype": "text/x-python",
   "name": "python",
   "nbconvert_exporter": "python",
   "pygments_lexer": "ipython3",
   "version": "3.7.10"
  }
 },
 "nbformat": 4,
 "nbformat_minor": 4
}