helloMNIST.ipynb

{
 "cells": [
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "# Hello, MNIST!"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 1,
   "metadata": {},
   "outputs": [],
   "source": [
    "import torch\n",
    "import fastai\n",
    "from fastai import *\n",
    "from fastai.vision import *\n",
    "import numpy as np\n",
    "import matplotlib\n",
    "import matplotlib.pyplot as plt\n",
    "from sklearn.model_selection import train_test_split"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 2,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "'1.0.29.dev0'"
      ]
     },
     "execution_count": 2,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "fastai.__version__"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 3,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "True"
      ]
     },
     "execution_count": 3,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "torch.manual_seed(17);\n",
    "use_gpu = torch.cuda.is_available();use_gpu"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "## Obtendo os dados"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 4,
   "metadata": {},
   "outputs": [],
   "source": [
    "from sklearn.datasets import fetch_mldata\n",
    "\n",
    "mnist = fetch_mldata('MNIST original')\n",
    "X, y = mnist[\"data\"], mnist[\"target\"]\n",
    "X_train, X_valid, y_train, y_valid = train_test_split(X, y, test_size=0.2, random_state=42)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 5,
   "metadata": {},
   "outputs": [],
   "source": [
    "# Normalization\n",
    "mean = X_train.mean()\n",
    "std = X_train.std()\n",
    "X_train = (X_train-mean)/std\n",
    "X_valid = (X_valid-mean)/std\n",
    "\n",
    "# Numpy to Torch Tensor\n",
    "\n",
    "X_train,y_train,X_valid,y_valid = map(torch.tensor, (X_train.astype(np.float32),y_train.astype(np.long),X_valid.astype(np.float32),y_valid.astype(np.long)))"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 6,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "(torch.Size([56000, 784]), tensor(0), tensor(9))"
      ]
     },
     "execution_count": 6,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "X_train.shape, y_train.min(), y_train.max()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 7,
   "metadata": {},
   "outputs": [],
   "source": [
    "bs=64\n",
    "train_ds = TensorDataset(X_train, y_train)\n",
    "valid_ds = TensorDataset(X_valid, y_valid)\n",
    "data = DataBunch.create(train_ds, valid_ds, bs=bs)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 8,
   "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": [
    "plt.imshow(X_train[:800], cmap=matplotlib.cm.binary)\n",
    "plt.axis('off')\n",
    "plt.show()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 9,
   "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"
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "tensor(4)\n"
     ]
    }
   ],
   "source": [
    "i = 100\n",
    "plt.imshow(X_train[i, :].reshape(28, 28), cmap = matplotlib.cm.binary)\n",
    "plt.axis(\"off\")\n",
    "plt.show()\n",
    "print (y_train[i])"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "## Criando o modelo"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 10,
   "metadata": {},
   "outputs": [],
   "source": [
    "import torch.nn as nn\n",
    "\n",
    "model = nn.Sequential(\n",
    "    nn.Linear(28*28, 16),\n",
    "    nn.ReLU(),\n",
    "    nn.Linear(16, 16),\n",
    "    nn.ReLU(),\n",
    "    nn.Linear(16, 10)\n",
    ").cuda()"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "## Função de Perda"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 11,
   "metadata": {},
   "outputs": [],
   "source": [
    "loss_func=nn.CrossEntropyLoss()"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "## Otimizador: Stochastic Gradient Descent (SGD)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 12,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "(torch.Size([64, 784]), torch.Size([64]))"
      ]
     },
     "execution_count": 12,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "lr = 1e-2\n",
    "x,y = next(iter(data.train_dl))\n",
    "x.shape,y.shape"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 13,
   "metadata": {},
   "outputs": [],
   "source": [
    "def update(x,y,lr):\n",
    "    y_hat = model(x)\n",
    "    loss = loss_func(y_hat, y)\n",
    "    loss.backward()\n",
    "    with torch.no_grad():\n",
    "        for p in model.parameters():\n",
    "            p.sub_(lr * p.grad)\n",
    "            p.grad.zero_()\n",
    "    return loss.item()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 14,
   "metadata": {},
   "outputs": [],
   "source": [
    "losses = [update(x,y,lr) for x,y in data.train_dl]"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 15,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "[<matplotlib.lines.Line2D at 0x7ffa88d75470>]"
      ]
     },
     "execution_count": 15,
     "metadata": {},
     "output_type": "execute_result"
    },
    {
     "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": [
    "plt.plot(losses)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 16,
   "metadata": {},
   "outputs": [],
   "source": [
    "learn = Learner(data=data, model=model, loss_func=loss_func, metrics=[accuracy])"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 17,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "application/vnd.jupyter.widget-view+json": {
       "model_id": "",
       "version_major": 2,
       "version_minor": 0
      },
      "text/plain": [
       "VBox(children=(HBox(children=(IntProgress(value=0, max=10), HTML(value='0.00% [0/10 00:00<00:00]'))), HTML(val…"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Total time: 00:28\n",
      "epoch  train_loss  valid_loss  accuracy\n",
      "1      0.258577    0.238942    0.928429  (00:02)\n",
      "2      0.232224    0.253103    0.926857  (00:02)\n",
      "3      0.228010    0.209510    0.939071  (00:02)\n",
      "4      0.210804    0.235379    0.935000  (00:02)\n",
      "5      0.203141    0.226332    0.940214  (00:02)\n",
      "6      0.203531    0.232411    0.940143  (00:02)\n",
      "7      0.195709    0.242415    0.933714  (00:02)\n",
      "8      0.211786    0.216327    0.941929  (00:02)\n",
      "9      0.189413    0.186680    0.949286  (00:02)\n",
      "10     0.177739    0.209640    0.943929  (00:02)\n",
      "\n"
     ]
    }
   ],
   "source": [
    "learn.fit(epochs=10, lr=1e-2)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 18,
   "metadata": {},
   "outputs": [],
   "source": [
    "net = nn.Sequential(\n",
    "    nn.Linear(28*28, 500),\n",
    "    nn.ReLU(),\n",
    "    nn.Linear(500, 300),\n",
    "    nn.ReLU(),\n",
    "    nn.Linear(300, 100),\n",
    "    nn.ReLU(),\n",
    "    nn.Linear(100, 16),\n",
    "    nn.ReLU(),\n",
    "    nn.Linear(16, 10)\n",
    ").cuda()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 19,
   "metadata": {},
   "outputs": [],
   "source": [
    "learn = Learner(data=data, model=net, loss_func=loss_func, metrics=[accuracy])"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 20,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "application/vnd.jupyter.widget-view+json": {
       "model_id": "",
       "version_major": 2,
       "version_minor": 0
      },
      "text/plain": [
       "VBox(children=(HBox(children=(IntProgress(value=0, max=10), HTML(value='0.00% [0/10 00:00<00:00]'))), HTML(val…"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Total time: 00:34\n",
      "epoch  train_loss  valid_loss  accuracy\n",
      "1      0.210793    0.202065    0.942929  (00:03)\n",
      "2      0.201581    0.158427    0.957786  (00:03)\n",
      "3      0.247223    0.295954    0.931500  (00:03)\n",
      "4      0.201971    0.229460    0.943286  (00:03)\n",
      "5      0.167386    0.165112    0.963286  (00:03)\n",
      "6      0.111749    0.145264    0.970429  (00:03)\n",
      "7      0.079752    0.142233    0.973214  (00:03)\n",
      "8      0.050061    0.120662    0.976857  (00:03)\n",
      "9      0.034403    0.112151    0.980714  (00:03)\n",
      "10     0.017906    0.125989    0.981286  (00:03)\n",
      "\n"
     ]
    }
   ],
   "source": [
    "learn.fit_one_cycle(10, 1e-2)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 21,
   "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": [
    "learn.recorder.plot_lr()"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "Observação:  Se ainda não estiver convencido de que entende como NN funcionam, tente acompanhar o passo a passo em [jcjohnson/pytorch-examples](http://github.com/jcjohnson/pytorch-examples)"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "# Fim"
   ]
  }
 ],
 "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": "3.7.0"
  }
 },
 "nbformat": 4,
 "nbformat_minor": 2
}