ann.ipynb

{
 "cells": [
  {
   "cell_type": "code",
   "execution_count": 1,
   "metadata": {},
   "outputs": [],
   "source": [
    "import torch\n",
    "import torch.nn as nn\n",
    "import torch.nn.functional as F\n",
    "\n",
    "model_path = 'ann_best.pth'\n",
    "\n",
    "img_x_size = int(335)\n",
    "img_y_size = int(218)\n",
    "num_channels = 1\n",
    "input_size = img_x_size * img_y_size * num_channels\n",
    "dropout_p = 0.3\n",
    "\n",
    "if not torch.backends.mps.is_available():\n",
    "    device = torch.device('cpu')\n",
    "    if not torch.backends.mps.is_built():\n",
    "        print(\"MPS not available because the current PyTorch install was not \"\n",
    "              \"built with MPS enabled.\")\n",
    "    else:\n",
    "        print(\"MPS not available because the current MacOS version is not 12.3+ \"\n",
    "              \"and/or you do not have an MPS-enabled device on this machine.\")\n",
    "\n",
    "else:\n",
    "    device = torch.device(\"mps\")\n",
    "\n",
    "class Net(nn.Module):\n",
    "\n",
    "    def __init__(self):\n",
    "        super(Net, self).__init__()\n",
    "\n",
    "        fc1_in = input_size\n",
    "        fc1_out = 1000\n",
    "        self.fc1 = nn.Linear(fc1_in, fc1_out)\n",
    "        # nn.init.normal_(self.fc1.weight, mean=0, std=1/math.sqrt(fc1_in))\n",
    "        # nn.init.normal_(self.fc1.bias, mean=0, std=1)\n",
    "        # self.dropout1 = nn.Dropout(p=dropout_p)\n",
    "\n",
    "        self.bn1 = nn.BatchNorm1d(fc1_out) # Batch normalization for fc1\n",
    "        # TODO: weight initialization for batchnorm layer?\n",
    "\n",
    "        fc2_in = fc1_out\n",
    "        fc2_out = 500\n",
    "        self.fc2 = nn.Linear(fc2_in, fc2_out)\n",
    "        # nn.init.normal_(self.fc2.weight, mean=0, std=1/math.sqrt(fc2_in))\n",
    "        # nn.init.normal_(self.fc2.bias, mean=0, std=1)\n",
    "        self.dropout2 = nn.Dropout(p=dropout_p)\n",
    "\n",
    "        self.bn2 = nn.BatchNorm1d(fc2_out)   # Batch normalization for fc2\n",
    "\n",
    "        fc3_in = fc2_out\n",
    "        fc3_out = 100\n",
    "        self.fc3 = nn.Linear(fc3_in, fc3_out)  # Output layer with 10 units for 10 genres\n",
    "        # nn.init.normal_(self.fc3.weight, mean=0, std=1/math.sqrt(fc3_in))\n",
    "        # nn.init.normal_(self.fc3.bias, mean=0, std=1)\n",
    "        # self.dropout3 = nn.Dropout(p=dropout_p)\n",
    "        \n",
    "        self.bn3 = nn.BatchNorm1d(fc3_out)   # Batch normalization for fc2\n",
    "\n",
    "        fc4_in = fc3_out\n",
    "        fc4_out = 10\n",
    "        self.fc4 = nn.Linear(fc4_in, fc4_out)  # Output layer with 10 units for 10 genres\n",
    "        # nn.init.normal_(self.fc4.weight, mean=0, std=1/math.sqrt(fc4_in))\n",
    "        # nn.init.normal_(self.fc4.bias, mean=0, std=1)\n",
    "\n",
    "    def forward(self, x):\n",
    "        x = x.view(-1, input_size)\n",
    "\n",
    "        # Apply sigmoid activation at each layer\n",
    "        x = self.fc1(x)\n",
    "        x = self.bn1(x)  # Apply batch normalization to fc1\n",
    "        x = F.relu(x)\n",
    "        # x = self.dropout1(x)  # Apply dropout after the first layer\n",
    "        \n",
    "        x = self.fc2(x)\n",
    "        x = self.bn2(x)  # Apply batch normalization to fc2\n",
    "        x = F.relu(x)\n",
    "        x = self.dropout2(x)  # Apply dropout after the second layer\n",
    "        \n",
    "        x = self.fc3(x)\n",
    "        x = self.bn3(x)  # Apply batch normalization to fc3\n",
    "        x = F.relu(x)\n",
    "        # x = self.dropout3(x)  # Apply dropout after the second layer\n",
    "\n",
    "        x = F.relu(self.fc4(x))\n",
    "\n",
    "        # x = F.softmax(x, dim=1)\n",
    "\n",
    "        return x"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 2,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "73585810"
      ]
     },
     "execution_count": 2,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "net = Net()\n",
    "\n",
    "# Calculate the number of trainable parameters\n",
    "total_params = sum(p.numel() for p in net.parameters() if p.requires_grad)\n",
    "total_params"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 3,
   "metadata": {},
   "outputs": [
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "/Users/steven/Library/Python/3.9/lib/python/site-packages/urllib3/__init__.py:34: NotOpenSSLWarning: urllib3 v2.0 only supports OpenSSL 1.1.1+, currently the 'ssl' module is compiled with 'LibreSSL 2.8.3'. See: https://github.com/urllib3/urllib3/issues/3020\n",
      "  warnings.warn(\n"
     ]
    }
   ],
   "source": [
    "from util import load_data_from_split\n",
    "split_songs = False\n",
    "train_loader, val_loader, test_loader = load_data_from_split(resize_images=False, batch_size=10, split_songs=split_songs)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 81,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Epoch 1 | LR: 0.00010 -> 0.00019 | Train Loss: 2.178 | Val Loss: 1.968 | Val Acc: 32.00%\n",
      "Epoch 2 | LR: 0.00019 -> 0.00028 | Train Loss: 1.848 | Val Loss: 1.861 | Val Acc: 34.00%\n",
      "Epoch 3 | LR: 0.00028 -> 0.00037 | Train Loss: 1.592 | Val Loss: 1.808 | Val Acc: 44.00%\n",
      "Epoch 4 | LR: 0.00037 -> 0.00046 | Train Loss: 1.381 | Val Loss: 1.751 | Val Acc: 49.00%\n",
      "Epoch 5 | LR: 0.00046 -> 0.00055 | Train Loss: 1.176 | Val Loss: 1.705 | Val Acc: 49.00%\n",
      "Epoch 6 | LR: 0.00055 -> 0.00064 | Train Loss: 0.956 | Val Loss: 1.665 | Val Acc: 52.00%\n",
      "Epoch 7 | LR: 0.00064 -> 0.00073 | Train Loss: 0.723 | Val Loss: 1.637 | Val Acc: 50.00%\n",
      "Epoch 8 | LR: 0.00073 -> 0.00082 | Train Loss: 0.573 | Val Loss: 1.576 | Val Acc: 49.00%\n",
      "Epoch 9 | LR: 0.00082 -> 0.00091 | Train Loss: 0.439 | Val Loss: 1.555 | Val Acc: 50.00%\n",
      "Epoch 10 | LR: 0.00091 -> 0.00100 | Train Loss: 0.373 | Val Loss: 1.554 | Val Acc: 50.00%\n",
      "Epoch 11 | LR: 0.00100 -> 0.00091 | Train Loss: 0.310 | Val Loss: 1.540 | Val Acc: 50.00%\n",
      "Epoch 12 | LR: 0.00091 -> 0.00082 | Train Loss: 0.302 | Val Loss: 1.527 | Val Acc: 53.00%\n",
      "Epoch 13 | LR: 0.00082 -> 0.00073 | Train Loss: 0.226 | Val Loss: 1.479 | Val Acc: 52.00%\n",
      "Epoch 14 | LR: 0.00073 -> 0.00064 | Train Loss: 0.186 | Val Loss: 1.501 | Val Acc: 52.00%\n",
      "Epoch 15 | LR: 0.00064 -> 0.00055 | Train Loss: 0.180 | Val Loss: 1.525 | Val Acc: 57.00%\n",
      "Epoch 16 | LR: 0.00055 -> 0.00046 | Train Loss: 0.157 | Val Loss: 1.463 | Val Acc: 51.00%\n",
      "Epoch 17 | LR: 0.00046 -> 0.00037 | Train Loss: 0.162 | Val Loss: 1.486 | Val Acc: 52.00%\n",
      "Epoch 18 | LR: 0.00037 -> 0.00028 | Train Loss: 0.151 | Val Loss: 1.511 | Val Acc: 53.00%\n",
      "Epoch 19 | LR: 0.00028 -> 0.00019 | Train Loss: 0.122 | Val Loss: 1.479 | Val Acc: 53.00%\n",
      "Epoch 20 | LR: 0.00019 -> 0.00010 | Train Loss: 0.137 | Val Loss: 1.494 | Val Acc: 53.00%\n",
      "Training finished\n"
     ]
    }
   ],
   "source": [
    "import torch.nn as nn\n",
    "import torch.optim as optim\n",
    "from torch.optim import lr_scheduler\n",
    "from util import train_loop\n",
    "\n",
    "net = Net()\n",
    "num_epochs = 20\n",
    "criterion = nn.CrossEntropyLoss() # loss function\n",
    "\n",
    "optimizer = optim.SGD(net.parameters(), lr=5e-3, momentum=0, weight_decay=1e-5)\n",
    "\n",
    "scheduler = lr_scheduler.CyclicLR(optimizer, base_lr=1e-5, max_lr=1e-3, step_size_up=(num_epochs/2)*len(train_loader), step_size_down=(num_epochs/2)*len(train_loader), mode='triangular')\n",
    "\n",
    "training_losses, validation_losses = train_loop(\n",
    "    net=net,\n",
    "    device=device,\n",
    "    train_loader=train_loader,\n",
    "    val_loader=val_loader,\n",
    "    criterion=criterion,\n",
    "    model_path=model_path,\n",
    "    split_songs=split_songs,\n",
    "    num_epochs=num_epochs,\n",
    "    optimizer=optimizer,\n",
    "    scheduler=scheduler\n",
    ")"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 83,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Test Accuracy | Idv: 51.00%\n"
     ]
    }
   ],
   "source": [
    "from util import test_model, test_model_aggregate\n",
    "\n",
    "net = Net()\n",
    "net.to(device)\n",
    "\n",
    "# Load the best saved model weights\n",
    "net.load_state_dict(torch.load(f\"models/{model_path}\"))\n",
    "\n",
    "test_accuracy, test_loss = test_model(net, test_loader, device, criterion)\n",
    "\n",
    "results_str = f\"Test Accuracy | Idv: {test_accuracy:.2f}%\"\n",
    "if(split_songs):\n",
    "    test_accuracy_agg = test_model_aggregate(net, test_loader, device)\n",
    "    results_str += f\", Agg: {test_accuracy_agg:.2f}%\"\n",
    "\n",
    "print(results_str)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 84,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "[2.1776107996702194, 1.848007820546627, 1.5917977705597877, 1.380607570707798, 1.1760029457509518, 0.9559479869902134, 0.723469116166234, 0.573155352845788, 0.4388303326442838, 0.3727551707997918, 0.3104449223726988, 0.3021510051563382, 0.2257928928360343, 0.1864972621202469, 0.17957747234031557, 0.15733995707705617, 0.1623209336772561, 0.15098564508371054, 0.12175698657520115, 0.13727872404269875] [1.9678052067756653, 1.8605483531951905, 1.8079592823982238, 1.7508163094520568, 1.7045459985733031, 1.665364956855774, 1.6366252779960633, 1.575569486618042, 1.5547997593879699, 1.5540575742721559, 1.5402992129325868, 1.527490383386612, 1.4793010920286178, 1.5011501073837281, 1.5249832391738891, 1.4631983876228332, 1.4856610894203186, 1.5110902070999146, 1.4786349892616273, 1.4942681670188904]\n"
     ]
    },
    {
     "data": {
      "image/png": "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",
      "text/plain": [
       "<Figure size 1000x500 with 1 Axes>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    }
   ],
   "source": [
    "from util import plot_loss\n",
    "print(training_losses, validation_losses)\n",
    "plot_loss(training_losses, validation_losses)"
   ]
  }
 ],
 "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": "3.9.6"
  },
  "orig_nbformat": 4
 },
 "nbformat": 4,
 "nbformat_minor": 2
}