FedNCF.ipynb

{
 "cells": [
  {
   "cell_type": "code",
   "execution_count": 2,
   "metadata": {},
   "outputs": [],
   "source": [
    "import pandas as pd\n",
    "import torch\n",
    "from sklearn.utils import shuffle\n",
    "import torch.nn as nn"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 3,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Running on the GPU\n"
     ]
    }
   ],
   "source": [
    "import torch\n",
    "if torch.cuda.is_available():\n",
    "    device = torch.device(\"cuda\")  # you can continue going on here, like cuda:1 cuda:2....etc.\n",
    "    print(\"Running on the GPU\")\n",
    "else:\n",
    "    device = torch.device(\"cpu\")\n",
    "    print(\"Running on the CPU\")"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 4,
   "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>userId</th>\n",
       "      <th>movieId</th>\n",
       "      <th>rating</th>\n",
       "      <th>TimeStamp</th>\n",
       "    </tr>\n",
       "  </thead>\n",
       "  <tbody>\n",
       "    <tr>\n",
       "      <th>0</th>\n",
       "      <td>1</td>\n",
       "      <td>1193</td>\n",
       "      <td>5</td>\n",
       "      <td>978300760</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>1</th>\n",
       "      <td>1</td>\n",
       "      <td>661</td>\n",
       "      <td>3</td>\n",
       "      <td>978302109</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>2</th>\n",
       "      <td>1</td>\n",
       "      <td>914</td>\n",
       "      <td>3</td>\n",
       "      <td>978301968</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>3</th>\n",
       "      <td>1</td>\n",
       "      <td>3408</td>\n",
       "      <td>4</td>\n",
       "      <td>978300275</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>4</th>\n",
       "      <td>1</td>\n",
       "      <td>2355</td>\n",
       "      <td>5</td>\n",
       "      <td>978824291</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>...</th>\n",
       "      <td>...</td>\n",
       "      <td>...</td>\n",
       "      <td>...</td>\n",
       "      <td>...</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>1000204</th>\n",
       "      <td>6040</td>\n",
       "      <td>1091</td>\n",
       "      <td>1</td>\n",
       "      <td>956716541</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>1000205</th>\n",
       "      <td>6040</td>\n",
       "      <td>1094</td>\n",
       "      <td>5</td>\n",
       "      <td>956704887</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>1000206</th>\n",
       "      <td>6040</td>\n",
       "      <td>562</td>\n",
       "      <td>5</td>\n",
       "      <td>956704746</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>1000207</th>\n",
       "      <td>6040</td>\n",
       "      <td>1096</td>\n",
       "      <td>4</td>\n",
       "      <td>956715648</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>1000208</th>\n",
       "      <td>6040</td>\n",
       "      <td>1097</td>\n",
       "      <td>4</td>\n",
       "      <td>956715569</td>\n",
       "    </tr>\n",
       "  </tbody>\n",
       "</table>\n",
       "<p>1000209 rows × 4 columns</p>\n",
       "</div>"
      ],
      "text/plain": [
       "         userId  movieId  rating  TimeStamp\n",
       "0             1     1193       5  978300760\n",
       "1             1      661       3  978302109\n",
       "2             1      914       3  978301968\n",
       "3             1     3408       4  978300275\n",
       "4             1     2355       5  978824291\n",
       "...         ...      ...     ...        ...\n",
       "1000204    6040     1091       1  956716541\n",
       "1000205    6040     1094       5  956704887\n",
       "1000206    6040      562       5  956704746\n",
       "1000207    6040     1096       4  956715648\n",
       "1000208    6040     1097       4  956715569\n",
       "\n",
       "[1000209 rows x 4 columns]"
      ]
     },
     "execution_count": 4,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "import pandas as pd\n",
    "rnames = ['userId', 'movieId', 'rating', \"TimeStamp\"]\n",
    "ratings = pd.read_table(\"ml-1m.inter\", header=0, names=rnames, engine='python')\n",
    "ratings"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 5,
   "metadata": {},
   "outputs": [],
   "source": [
    "ratio = 0.8\n",
    "ratings = shuffle(ratings)\n",
    "# ratings\n",
    "train = ratings.copy()\n",
    "test = ratings.copy()\n",
    "train.iloc[int(ratio*len(ratings)):,2] = None\n",
    "# train.iloc[int(ratio*len(ratings)):,'rating'] = 0\n",
    "test.iloc[:int(ratio*len(ratings)),2] = None"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 6,
   "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>userId</th>\n",
       "      <th>movieId</th>\n",
       "      <th>rating</th>\n",
       "      <th>TimeStamp</th>\n",
       "    </tr>\n",
       "  </thead>\n",
       "  <tbody>\n",
       "    <tr>\n",
       "      <th>count</th>\n",
       "      <td>1.000209e+06</td>\n",
       "      <td>1.000209e+06</td>\n",
       "      <td>1.000209e+06</td>\n",
       "      <td>1.000209e+06</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>mean</th>\n",
       "      <td>3.024512e+03</td>\n",
       "      <td>1.865540e+03</td>\n",
       "      <td>3.581564e+00</td>\n",
       "      <td>9.722437e+08</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>std</th>\n",
       "      <td>1.728413e+03</td>\n",
       "      <td>1.096041e+03</td>\n",
       "      <td>1.117102e+00</td>\n",
       "      <td>1.215256e+07</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>min</th>\n",
       "      <td>1.000000e+00</td>\n",
       "      <td>1.000000e+00</td>\n",
       "      <td>1.000000e+00</td>\n",
       "      <td>9.567039e+08</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>25%</th>\n",
       "      <td>1.506000e+03</td>\n",
       "      <td>1.030000e+03</td>\n",
       "      <td>3.000000e+00</td>\n",
       "      <td>9.653026e+08</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>50%</th>\n",
       "      <td>3.070000e+03</td>\n",
       "      <td>1.835000e+03</td>\n",
       "      <td>4.000000e+00</td>\n",
       "      <td>9.730180e+08</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>75%</th>\n",
       "      <td>4.476000e+03</td>\n",
       "      <td>2.770000e+03</td>\n",
       "      <td>4.000000e+00</td>\n",
       "      <td>9.752209e+08</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>max</th>\n",
       "      <td>6.040000e+03</td>\n",
       "      <td>3.952000e+03</td>\n",
       "      <td>5.000000e+00</td>\n",
       "      <td>1.046455e+09</td>\n",
       "    </tr>\n",
       "  </tbody>\n",
       "</table>\n",
       "</div>"
      ],
      "text/plain": [
       "             userId       movieId        rating     TimeStamp\n",
       "count  1.000209e+06  1.000209e+06  1.000209e+06  1.000209e+06\n",
       "mean   3.024512e+03  1.865540e+03  3.581564e+00  9.722437e+08\n",
       "std    1.728413e+03  1.096041e+03  1.117102e+00  1.215256e+07\n",
       "min    1.000000e+00  1.000000e+00  1.000000e+00  9.567039e+08\n",
       "25%    1.506000e+03  1.030000e+03  3.000000e+00  9.653026e+08\n",
       "50%    3.070000e+03  1.835000e+03  4.000000e+00  9.730180e+08\n",
       "75%    4.476000e+03  2.770000e+03  4.000000e+00  9.752209e+08\n",
       "max    6.040000e+03  3.952000e+03  5.000000e+00  1.046455e+09"
      ]
     },
     "execution_count": 6,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "ratings.describe()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 7,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "movieId  1     2     3     4     5     6     7     8     9     10    ...   \n",
      "userId                                                               ...   \n",
      "1         1.0   NaN   NaN   NaN   NaN   NaN   NaN   NaN   NaN   NaN  ...  \\\n",
      "2         NaN   NaN   NaN   NaN   NaN   NaN   NaN   NaN   NaN   NaN  ...   \n",
      "3         NaN   NaN   NaN   NaN   NaN   NaN   NaN   NaN   NaN   NaN  ...   \n",
      "4         NaN   NaN   NaN   NaN   NaN   NaN   NaN   NaN   NaN   NaN  ...   \n",
      "5         NaN   NaN   NaN   NaN   NaN  0.25   NaN   NaN   NaN   NaN  ...   \n",
      "...       ...   ...   ...   ...   ...   ...   ...   ...   ...   ...  ...   \n",
      "6036      NaN   NaN   NaN  0.25   NaN  0.50   NaN   NaN   NaN   NaN  ...   \n",
      "6037      NaN   NaN   NaN   NaN   NaN   NaN   NaN   NaN   NaN   NaN  ...   \n",
      "6038      NaN   NaN   NaN   NaN   NaN   NaN   NaN   NaN   NaN   NaN  ...   \n",
      "6039      NaN   NaN   NaN   NaN   NaN   NaN   NaN   NaN   NaN   NaN  ...   \n",
      "6040      0.5   NaN   NaN   NaN   NaN   NaN   NaN   NaN   NaN   NaN  ...   \n",
      "\n",
      "movieId  3943  3944  3945  3946  3947  3948  3949  3950  3951  3952  \n",
      "userId                                                               \n",
      "1         NaN   NaN   NaN   NaN   NaN   NaN   NaN   NaN   NaN   NaN  \n",
      "2         NaN   NaN   NaN   NaN   NaN   NaN   NaN   NaN   NaN   NaN  \n",
      "3         NaN   NaN   NaN   NaN   NaN   NaN   NaN   NaN   NaN   NaN  \n",
      "4         NaN   NaN   NaN   NaN   NaN   NaN   NaN   NaN   NaN   NaN  \n",
      "5         NaN   NaN   NaN   NaN   NaN   NaN   NaN   NaN   NaN   NaN  \n",
      "...       ...   ...   ...   ...   ...   ...   ...   ...   ...   ...  \n",
      "6036      NaN   NaN   NaN   NaN   NaN   NaN   NaN   NaN   NaN   NaN  \n",
      "6037      NaN   NaN   NaN   NaN   NaN   NaN   NaN   NaN   NaN   NaN  \n",
      "6038      NaN   NaN   NaN   NaN   NaN   NaN   NaN   NaN   NaN   NaN  \n",
      "6039      NaN   NaN   NaN   NaN   NaN   NaN   NaN   NaN   NaN   NaN  \n",
      "6040      NaN   NaN   NaN   NaN   NaN   NaN   NaN   NaN   NaN   NaN  \n",
      "\n",
      "[6040 rows x 3706 columns]\n",
      "800167\n"
     ]
    }
   ],
   "source": [
    "rating_matrix = train.pivot(index='userId', columns='movieId', values='rating')\n",
    "n_users, n_movies = rating_matrix.shape\n",
    "# Scaling ratings to between 0 and 1, this helps our model by constraining predictions\n",
    "min_rating, max_rating = train['rating'].min(), train['rating'].max()\n",
    "rating_matrix = (rating_matrix - min_rating) / (max_rating - min_rating)\n",
    "print(rating_matrix)\n",
    "print(n_users*n_movies-rating_matrix.isnull().values.sum())"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 8,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "tensor([[ 1.0000, -1.0000, -1.0000,  ..., -1.0000, -1.0000, -1.0000],\n",
      "        [-1.0000, -1.0000, -1.0000,  ..., -1.0000, -1.0000, -1.0000],\n",
      "        [-1.0000, -1.0000, -1.0000,  ..., -1.0000, -1.0000, -1.0000],\n",
      "        ...,\n",
      "        [-1.0000, -1.0000, -1.0000,  ..., -1.0000, -1.0000, -1.0000],\n",
      "        [-1.0000, -1.0000, -1.0000,  ..., -1.0000, -1.0000, -1.0000],\n",
      "        [ 0.5000, -1.0000, -1.0000,  ..., -1.0000, -1.0000, -1.0000]],\n",
      "       device='cuda:0')\n"
     ]
    }
   ],
   "source": [
    "# Replacing missing ratings with -1 so we can filter them out later\n",
    "\n",
    "rating_matrix[rating_matrix.isnull()] = -1\n",
    "rating_matrix = torch.FloatTensor(rating_matrix.values).to(device)\n",
    "print(rating_matrix)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 9,
   "metadata": {},
   "outputs": [],
   "source": [
    "# class GMF(torch.nn.Module):\n",
    "#     def __init__(self, feature_size):\n",
    "#         super().__init__()\n",
    "#         self.fc = nn.Linear(feature_size, feature_size)\n",
    "#         self.out = nn.Linear(feature_size, 1)\n",
    "#         self.ReLU = nn.ReLU()\n",
    "\n",
    "#     def forward(self, matrix, u_features, v_features):\n",
    "#         x = torch.einsum('ik,jk->ijk',[u_features, v_features])\n",
    "#         x = self.fc(x)\n",
    "#         x = self.ReLU(x)\n",
    "#         x = self.out(x)\n",
    "#         predicted_rating = torch.sigmoid(x)\n",
    "#         # non_zero_mask = (matrix != -1).type(torch.FloatTensor)\n",
    "#         # predicted_rating = torch.sigmoid(torch.mm(u_features, v_features.t()))\n",
    "        \n",
    "#         return predicted_rating"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 10,
   "metadata": {},
   "outputs": [],
   "source": [
    "# A = torch.tensor([[1,2,3]])\n",
    "# B = torch.tensor([[1,2,3]])\n",
    "# print(torch.einsum('ik,jk->ijk',[A, B]))"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 14,
   "metadata": {},
   "outputs": [],
   "source": [
    "class Fed_NCF(nn.Module):\n",
    "    def __init__(self,user_num, item_num, factor_num):\n",
    "        super(Fed_NCF,self).__init__()\n",
    "        self.embed_user_GMF = nn.Embedding(user_num,factor_num)\n",
    "        self.embed_item_GMF = nn.Embedding(item_num,factor_num)\n",
    "        self.hidden_layer = nn.Linear(factor_num,factor_num)\n",
    "        self.predict_layer = nn.Linear(factor_num,1)\n",
    "        self._init_weight_()\n",
    "        self.ReLU = nn.ReLU()\n",
    "    def _init_weight_(self):\n",
    "        nn.init.normal_(self.embed_item_GMF.weight,std=0.01)\n",
    "        nn.init.normal_(self.embed_user_GMF.weight,std=0.01)\n",
    "\n",
    "    def forward(self,user,item):\n",
    "        embed_user_GMF = self.embed_user_GMF(user)\n",
    "        embed_item_GMF = self.embed_item_GMF(item)\n",
    "        output_GMF = torch.einsum('ik,jk->ijk',[embed_user_GMF, embed_item_GMF])\n",
    "        # output_GMF = self.hidden_layer(output_GMF)\n",
    "        # output_GMF = self.ReLU(output_GMF)\n",
    "        prediction = torch.sigmoid(self.predict_layer(output_GMF))\n",
    "\n",
    "        return torch.squeeze(prediction)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 15,
   "metadata": {},
   "outputs": [],
   "source": [
    "class GMFLoss(torch.nn.Module):\n",
    "    def __init__(self, lam_u=0.3, lam_v=0.3):\n",
    "        super().__init__()\n",
    "        self.lam_u = lam_u\n",
    "        self.lam_v = lam_v\n",
    "    \n",
    "    def forward(self, matrix, predicted):\n",
    "        non_zero_mask = (matrix != -1).type(torch.FloatTensor).to(device)\n",
    "        diff = (matrix - predicted)**2\n",
    "        prediction_error = torch.sum(diff*non_zero_mask)\n",
    "        \n",
    "        return prediction_error"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 17,
   "metadata": {},
   "outputs": [],
   "source": [
    "num_epoch = 100\n",
    "latent_vectors = 20\n",
    "gmf = Fed_NCF(n_users, n_movies, latent_vectors).to(device)\n",
    "gmfloss = GMFLoss(lam_u=0.1, lam_v=0.1)\n",
    "user = torch.LongTensor(range(n_users)).to(device)\n",
    "item = torch.LongTensor(range(n_movies)).to(device)\n",
    "\n",
    "optimizer = torch.optim.Adam(gmf.parameters(), lr=0.01)\n",
    "for step, epoch in enumerate(range(num_epoch)):\n",
    "    optimizer.zero_grad()\n",
    "    # pred = gmf(torch.ones(n_users).long().to(device),torch.ones(n_movies).long().to(device))\n",
    "    pred = gmf(user,item)\n",
    "    loss = gmfloss(rating_matrix, pred)\n",
    "    loss.backward()\n",
    "    optimizer.step()\n",
    "    if step % 10 == 0:\n",
    "        print(f\"Step {step}, {loss:.3f}\")"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "tensor([[-1., -1., -1.,  ..., -1., -1., -1.],\n",
       "        [-1., -1., -1.,  ..., -1., -1., -1.],\n",
       "        [-1., -1., -1.,  ..., -1., -1., -1.],\n",
       "        ...,\n",
       "        [-1., -1., -1.,  ..., -1., -1., -1.],\n",
       "        [-1., -1., -1.,  ..., -1., -1., -1.],\n",
       "        [-1., -1., -1.,  ..., -1., -1., -1.]], device='cuda:0')"
      ]
     },
     "execution_count": 30,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "test_rating_matrix = test.pivot(index='userId', columns='movieId', values='rating')\n",
    "\n",
    "test_rating_matrix[test_rating_matrix.isnull()] = -1\n",
    "test_rating_matrix = torch.FloatTensor(test_rating_matrix.values).to(device)\n",
    "test_rating_matrix"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {},
   "outputs": [],
   "source": [
    "## test FedNCF\n",
    "non_zero_mask = (test_rating_matrix != -1).type(torch.FloatTensor).to(device)\n",
    "num = torch.sum(non_zero_mask)\n",
    "user = torch.LongTensor(range(n_users)).to(device)\n",
    "item = torch.LongTensor(range(n_movies)).to(device)\n",
    "predicted_ratings = gmf(user, item)\n",
    "predicted_ratings = (predicted_ratings*(max_rating - min_rating) + min_rating)*non_zero_mask\n",
    "actual_ratings = test_rating_matrix*non_zero_mask\n",
    "\n",
    "AE_diff = torch.abs(predicted_ratings - actual_ratings)\n",
    "SE_diff = (predicted_ratings - actual_ratings)**2\n",
    "\n",
    "test_MAE = torch.sum(AE_diff)/num\n",
    "test_RMSE = torch.sqrt(torch.sum(SE_diff)/num)\n",
    "print('test_MAE =', test_MAE.data.cpu().numpy())\n",
    "print('test_RMSE =', test_RMSE.data.cpu().numpy())"
   ]
  }
 ],
 "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": "3.10.11"
  }
 },
 "nbformat": 4,
 "nbformat_minor": 2
}