Music Genre Recognition v1.0.ipynb

{
  "cells": [
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "42vHVFcpHq48"
      },
      "source": [
        "# Imports"
      ]
    },
    {
      "cell_type": "code",
      "execution_count": null,
      "metadata": {
        "id": "m5ZrHVb4zs96"
      },
      "outputs": [],
      "source": [
        "import pandas as pd\n",
        "from sklearn import preprocessing\n",
        "from sklearn.decomposition import PCA\n",
        "from sklearn.utils import shuffle\n",
        "import seaborn as sns\n",
        "import matplotlib\n",
        "matplotlib.use('Agg')\n",
        "import matplotlib.pyplot as plt\n",
        "from sklearn.neighbors import KNeighborsClassifier\n",
        "from sklearn.model_selection import train_test_split\n",
        "from sklearn.metrics import confusion_matrix, accuracy_score\n",
        "import librosa\n",
        "import librosa.display\n",
        "import os\n",
        "import numpy as np\n",
        "import csv\n",
        "import torch\n",
        "import time\n",
        "import torchvision\n",
        "import torchvision.transforms as transforms\n",
        "import torch.optim as optim\n",
        "import torch.nn as nn\n",
        "import torch.nn.functional as F\n",
        "import soundfile as sf\n",
        "from torchvision import datasets\n",
        "from math import floor"
      ]
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "rDINDOztYDZT"
      },
      "source": [
        "# Mounting"
      ]
    },
    {
      "cell_type": "code",
      "execution_count": null,
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/"
        },
        "id": "253Ue8xUXGcl",
        "outputId": "88585112-7a1e-437a-9543-9d4a872c8071"
      },
      "outputs": [
        {
          "output_type": "stream",
          "name": "stdout",
          "text": [
            "Mounted at /content/drive/\n"
          ]
        }
      ],
      "source": [
        "from google.colab import drive\n",
        "drive.mount('/content/drive/')\n",
        "path = \"/content/drive/MyDrive/APS360Project/GTZAN\""
      ]
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "hEzt-RvXHexD"
      },
      "source": [
        "# Data Processing"
      ]
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "X7-uGWZyKMHa"
      },
      "source": [
        "Mel-Spectrogram Generation\n"
      ]
    },
    {
      "cell_type": "code",
      "execution_count": null,
      "metadata": {
        "id": "SeEWzahnKePJ"
      },
      "outputs": [],
      "source": [
        "%matplotlib agg\n",
        "\n",
        "audio_path =  path + \"/genres_original/\"\n",
        "MFCC_path =  path + \"/MFCC_imgs/\"\n",
        "num_genres = 0\n",
        "\n",
        "if not os.path.isdir(MFCC_path):\n",
        "  os.mkdir(MFCC_path)\n",
        "\n",
        "for genre in os.listdir(audio_path):\n",
        "  num_genres += 1\n",
        "\n",
        "  if not os.path.isdir(MFCC_path + genre):\n",
        "    os.mkdir(MFCC_path + genre)\n",
        "\n",
        "  for audio_clip in os.listdir(audio_path + genre):\n",
        "    curr_audio_path = audio_path + genre + '/' + audio_clip\n",
        "\n",
        "    y, sr = librosa.load(curr_audio_path)\n",
        "    S = librosa.feature.melspectrogram(y, sr=sr)\n",
        "    S_DB = librosa.amplitude_to_db(S, ref=np.max)\n",
        "\n",
        "    librosa.display.specshow(S_DB, sr=sr, hop_length=512,)\n",
        "\n",
        "    plt.savefig(MFCC_path + genre + '/' + audio_clip.split('.')[0] +\n",
        "                audio_clip.split('.')[1] + '.png')"
      ]
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "KxDivXF64ECE"
      },
      "source": [
        "Mel-Spectrogram Generation for 3 second splits \\\n",
        "Ran locally to save drive space"
      ]
    },
    {
      "cell_type": "code",
      "execution_count": null,
      "metadata": {
        "id": "NL-Ihskg4DzJ"
      },
      "outputs": [],
      "source": [
        "\n",
        "audio_path = path + \"/genres_original/\"\n",
        "MFCC_path = path + \"/MFCC_imgs/\"\n",
        "\n",
        "# Assign directory for the split MFCC files. The folder path should be added manually.\n",
        "split_MFCC_path = path + \"/MFCC_imgs_split/\"\n",
        "\n",
        "num_genres = 0\n",
        "\n",
        "if not os.path.isdir(MFCC_path):\n",
        "    os.mkdir(MFCC_path)\n",
        "\n",
        "for genre in os.listdir(audio_path):\n",
        "    num_genres += 1\n",
        "\n",
        "    if not os.path.isdir(MFCC_path + genre):\n",
        "        os.mkdir(MFCC_path + genre)\n",
        "\n",
        "    for audio_clip in os.listdir(audio_path + genre):\n",
        "        curr_audio_path = audio_path + genre + '/' + audio_clip\n",
        "\n",
        "        y, sr = librosa.load(curr_audio_path)\n",
        "\n",
        "        for count in range(10):\n",
        "            split_y = y[sr * 3 * count: (sr * 3 * (count + 1)) - 1]\n",
        "            S = librosa.feature.melspectrogram(y=split_y, sr=sr)\n",
        "            S_DB = librosa.amplitude_to_db(S, ref=np.max)\n",
        "            librosa.display.specshow(S_DB, sr=sr, hop_length=512, )\n",
        "\n",
        "\n",
        "            plt.savefig(split_MFCC_path + genre + '/' + audio_clip.split('.')[0]\n",
        "                        + audio_clip.split('.')[1] + '.' + str(count) + '.png')"
      ]
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "dkscFFbEHlsr"
      },
      "source": [
        "30 Second Clip Feature Extraction"
      ]
    },
    {
      "cell_type": "code",
      "execution_count": null,
      "metadata": {
        "id": "UAxGfjYiHlXq"
      },
      "outputs": [],
      "source": [
        "audio_path =  path + \"/genres_original/\"\n",
        "\n",
        "header = 'filename length chroma_stft_mean chroma_stft_var rms_mean rms_var spectral_centroid_mean \\\n",
        "    spectral_centroid_var spectral_bandwidth_mean spectral_bandwidth_var \\\n",
        "    rolloff_mean rolloff_var zero_crossing_rate_mean zero_crossing_rate_var \\\n",
        "    harmony_mean harmony_var perceptr_mean perceptr_var tempo'\n",
        "\n",
        "for i in range(1, 21):\n",
        "    header += f' mfcc{i}_mean'\n",
        "    header += f' mfcc{i}_var'\n",
        "\n",
        "header += ' label'\n",
        "header = header.split()\n",
        "\n",
        "\n",
        "file = open(path + 'features_30_sec.csv', 'w', newline='')\n",
        "with file:\n",
        "    writer = csv.writer(file)\n",
        "    writer.writerow(header)\n",
        "\n",
        "for genre in os.listdir(audio_path):\n",
        "  for audio_clip in os.listdir(audio_path + genre):\n",
        "\n",
        "        curr_audio_path = audio_path + genre + '/' + audio_clip\n",
        "        y, sr = librosa.load(curr_audio_path, mono=True, duration=30)\n",
        "        length = librosa.get_duration(y=y, sr=sr) * sr\n",
        "        chroma_stft = librosa.feature.chroma_stft(y=y, sr=sr)\n",
        "        rms = librosa.feature.rms(y=y)\n",
        "        spec_cent = librosa.feature.spectral_centroid(y=y, sr=sr)\n",
        "        spec_bw = librosa.feature.spectral_bandwidth(y=y, sr=sr)\n",
        "        rolloff = librosa.feature.spectral_rolloff(y=y, sr=sr)\n",
        "        zcr = librosa.feature.zero_crossing_rate(y)\n",
        "        mfcc = librosa.feature.mfcc(y=y, sr=sr)\n",
        "        harm, perc = librosa.effects.hpss(y)\n",
        "        tempo = librosa.beat.tempo(y=y, sr=sr)\n",
        "\n",
        "        to_append = f'{audio_clip} {length} {np.mean(chroma_stft)} {np.var(chroma_stft)} \\\n",
        "        {np.mean(rms)} {np.var(rms)} {np.mean(spec_cent)} {np.var(spec_cent)} \\\n",
        "         {np.mean(spec_bw)} {np.var(spec_bw)} {np.mean(rolloff)} {np.var(rolloff)} \\\n",
        "         {np.mean(zcr)} {np.var(zcr)} {np.mean(harm)} {np.var(harm)} \\\n",
        "         {np.mean(perc)} {np.var(perc)} {np.mean(tempo)}'\n",
        "\n",
        "        for e in mfcc:\n",
        "            to_append += f' {np.mean(e)}'\n",
        "            to_append += f' {np.var(e)}'\n",
        "\n",
        "        to_append += f' {genre}'\n",
        "\n",
        "        file = open(path + 'features_30_sec.csv', 'a', newline='')\n",
        "        with file:\n",
        "            writer = csv.writer(file)\n",
        "            writer.writerow(to_append.split())\n"
      ]
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "ZemLo_abnBWj"
      },
      "source": [
        "3 Second Clip Splitting and Feature Extraction \\\n",
        "Ran locally to save drive space"
      ]
    },
    {
      "cell_type": "code",
      "execution_count": null,
      "metadata": {
        "id": "l7_iI1VhnIpd"
      },
      "outputs": [],
      "source": [
        "audio_path = path + \"/genres_original/\"\n",
        "new_audio_path = path + \"/genres_split/\"\n",
        "\n",
        "\n",
        "header = 'filename length chroma_stft_mean chroma_stft_var rms_mean rms_var spectral_centroid_mean \\\n",
        "    spectral_centroid_var spectral_bandwidth_mean spectral_bandwidth_var \\\n",
        "    rolloff_mean rolloff_var zero_crossing_rate_mean zero_crossing_rate_var \\\n",
        "    harmony_mean harmony_var perceptr_mean perceptr_var tempo'\n",
        "\n",
        "for i in range(1, 21):\n",
        "    header += f' mfcc{i}_mean'\n",
        "    header += f' mfcc{i}_var'\n",
        "\n",
        "header += ' label'\n",
        "header = header.split()\n",
        "\n",
        "# Updated name to be 'features_10_sec.csv'\n",
        "file = open('features_10_sec.csv', 'w', newline='')\n",
        "with file:\n",
        "    writer = csv.writer(file)\n",
        "    writer.writerow(header)\n",
        "\n",
        "\n",
        "for genre in os.listdir(audio_path):\n",
        "    for audio_clip in os.listdir(audio_path + genre):\n",
        "        curr_audio_path = audio_path + genre + '/' + audio_clip\n",
        "        x, sr = librosa.load(curr_audio_path, mono=True, duration=30)\n",
        "\n",
        "        # Split the audio file into 10s intervals and rename it and make features for the split files.\n",
        "        file_num = 0\n",
        "        for i in range(0, len(x), 3 * sr):\n",
        "            y = x[i: (3 * sr) + i]\n",
        "            update_filename = audio_clip[:-3] + str(file_num) + \".wav\"\n",
        "            update_audio_path = new_audio_path + genre + '/' + update_filename\n",
        "\n",
        "            # Write the new audio to update_audio_path directory.\n",
        "            # Will need to manually make the folder path for each genres.\n",
        "            sf.write(update_audio_path, y, sr)\n",
        "            file_num += 1\n",
        "\n",
        "            # Moved all the feature extraction codes inside the audio split for-loop.\n",
        "            length = librosa.get_duration(y=y, sr=sr) * sr\n",
        "            chroma_stft = librosa.feature.chroma_stft(y=y, sr=sr)\n",
        "            rms = librosa.feature.rms(y=y)\n",
        "            spec_cent = librosa.feature.spectral_centroid(y=y, sr=sr)\n",
        "            spec_bw = librosa.feature.spectral_bandwidth(y=y, sr=sr)\n",
        "            rolloff = librosa.feature.spectral_rolloff(y=y, sr=sr)\n",
        "            zcr = librosa.feature.zero_crossing_rate(y)\n",
        "            mfcc = librosa.feature.mfcc(y=y, sr=sr)\n",
        "            harm, perc = librosa.effects.hpss(y)\n",
        "            tempo = librosa.beat.tempo(y=y, sr=sr)\n",
        "\n",
        "            # Update: changed the file name to be \"update_curr_audio_path\"\n",
        "            to_append = f'{update_filename} {length} {np.mean(chroma_stft)} {np.var(chroma_stft)} \\\n",
        "            {np.mean(rms)} {np.var(rms)} {np.mean(spec_cent)} {np.var(spec_cent)} \\\n",
        "             {np.mean(spec_bw)} {np.var(spec_bw)} {np.mean(rolloff)} {np.var(rolloff)} \\\n",
        "             {np.mean(zcr)} {np.var(zcr)} {np.mean(harm)} {np.var(harm)} \\\n",
        "             {np.mean(perc)} {np.var(perc)} {np.mean(tempo)}'\n",
        "\n",
        "            for e in mfcc:\n",
        "                to_append += f' {np.mean(e)}'\n",
        "                to_append += f' {np.var(e)}'\n",
        "\n",
        "            to_append += f' {genre}'\n",
        "\n",
        "            file = open('features_10_sec.csv', 'a', newline='')\n",
        "            with file:\n",
        "                writer = csv.writer(file)\n",
        "                writer.writerow(to_append.split())"
      ]
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "mgJRdsDZgngI"
      },
      "source": [
        "# Baseline Model Section"
      ]
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "nZUAZiUGP8yj"
      },
      "source": [
        "## Subsection 1. 3 second model\n"
      ]
    },
    {
      "cell_type": "code",
      "execution_count": null,
      "metadata": {
        "id": "-qrQcwtUgslp"
      },
      "outputs": [],
      "source": [
        "## Code for Baseline Model will go here"
      ]
    },
    {
      "cell_type": "code",
      "execution_count": null,
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 299
        },
        "id": "XJh33lpWV0MY",
        "outputId": "99a58526-c2e6-4b5a-86fb-3bf7eb380f6a"
      },
      "outputs": [
        {
          "data": {
            "text/html": [
              "\n",
              "  <div id=\"df-37aa7936-b9d6-482d-bb20-9631e85615e4\">\n",
              "    <div class=\"colab-df-container\">\n",
              "      <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>filename</th>\n",
              "      <th>length</th>\n",
              "      <th>chroma_stft_mean</th>\n",
              "      <th>chroma_stft_var</th>\n",
              "      <th>rms_mean</th>\n",
              "      <th>rms_var</th>\n",
              "      <th>spectral_centroid_mean</th>\n",
              "      <th>spectral_centroid_var</th>\n",
              "      <th>spectral_bandwidth_mean</th>\n",
              "      <th>spectral_bandwidth_var</th>\n",
              "      <th>...</th>\n",
              "      <th>mfcc16_var</th>\n",
              "      <th>mfcc17_mean</th>\n",
              "      <th>mfcc17_var</th>\n",
              "      <th>mfcc18_mean</th>\n",
              "      <th>mfcc18_var</th>\n",
              "      <th>mfcc19_mean</th>\n",
              "      <th>mfcc19_var</th>\n",
              "      <th>mfcc20_mean</th>\n",
              "      <th>mfcc20_var</th>\n",
              "      <th>label</th>\n",
              "    </tr>\n",
              "  </thead>\n",
              "  <tbody>\n",
              "    <tr>\n",
              "      <th>0</th>\n",
              "      <td>blues.00000.0.wav</td>\n",
              "      <td>66150</td>\n",
              "      <td>0.335555</td>\n",
              "      <td>0.090997</td>\n",
              "      <td>0.130189</td>\n",
              "      <td>0.003559</td>\n",
              "      <td>1773.358004</td>\n",
              "      <td>169450.82950</td>\n",
              "      <td>1972.334258</td>\n",
              "      <td>117272.64020</td>\n",
              "      <td>...</td>\n",
              "      <td>39.547073</td>\n",
              "      <td>-3.230046</td>\n",
              "      <td>36.606857</td>\n",
              "      <td>0.696385</td>\n",
              "      <td>37.766129</td>\n",
              "      <td>-5.035945</td>\n",
              "      <td>33.668549</td>\n",
              "      <td>-0.239585</td>\n",
              "      <td>43.818882</td>\n",
              "      <td>blues</td>\n",
              "    </tr>\n",
              "    <tr>\n",
              "      <th>1</th>\n",
              "      <td>blues.00000.1.wav</td>\n",
              "      <td>66150</td>\n",
              "      <td>0.343523</td>\n",
              "      <td>0.086782</td>\n",
              "      <td>0.112119</td>\n",
              "      <td>0.001491</td>\n",
              "      <td>1817.244034</td>\n",
              "      <td>90766.29725</td>\n",
              "      <td>2010.751494</td>\n",
              "      <td>65940.66624</td>\n",
              "      <td>...</td>\n",
              "      <td>64.819778</td>\n",
              "      <td>-6.025472</td>\n",
              "      <td>40.548805</td>\n",
              "      <td>0.127131</td>\n",
              "      <td>51.048939</td>\n",
              "      <td>-2.808956</td>\n",
              "      <td>97.221489</td>\n",
              "      <td>5.771882</td>\n",
              "      <td>60.360352</td>\n",
              "      <td>blues</td>\n",
              "    </tr>\n",
              "    <tr>\n",
              "      <th>2</th>\n",
              "      <td>blues.00000.2.wav</td>\n",
              "      <td>66150</td>\n",
              "      <td>0.347746</td>\n",
              "      <td>0.092495</td>\n",
              "      <td>0.130895</td>\n",
              "      <td>0.004552</td>\n",
              "      <td>1790.722357</td>\n",
              "      <td>110071.20700</td>\n",
              "      <td>2088.184750</td>\n",
              "      <td>73391.49800</td>\n",
              "      <td>...</td>\n",
              "      <td>68.306793</td>\n",
              "      <td>-1.714475</td>\n",
              "      <td>28.136948</td>\n",
              "      <td>2.329553</td>\n",
              "      <td>47.211426</td>\n",
              "      <td>-1.925621</td>\n",
              "      <td>52.922432</td>\n",
              "      <td>2.466996</td>\n",
              "      <td>33.164001</td>\n",
              "      <td>blues</td>\n",
              "    </tr>\n",
              "    <tr>\n",
              "      <th>3</th>\n",
              "      <td>blues.00000.3.wav</td>\n",
              "      <td>66150</td>\n",
              "      <td>0.363863</td>\n",
              "      <td>0.087207</td>\n",
              "      <td>0.131349</td>\n",
              "      <td>0.002338</td>\n",
              "      <td>1660.545231</td>\n",
              "      <td>109496.93630</td>\n",
              "      <td>1967.920582</td>\n",
              "      <td>79805.90135</td>\n",
              "      <td>...</td>\n",
              "      <td>48.543201</td>\n",
              "      <td>-3.786986</td>\n",
              "      <td>28.419544</td>\n",
              "      <td>1.153315</td>\n",
              "      <td>35.682701</td>\n",
              "      <td>-3.501979</td>\n",
              "      <td>50.610340</td>\n",
              "      <td>3.580636</td>\n",
              "      <td>32.325874</td>\n",
              "      <td>blues</td>\n",
              "    </tr>\n",
              "    <tr>\n",
              "      <th>4</th>\n",
              "      <td>blues.00000.4.wav</td>\n",
              "      <td>66150</td>\n",
              "      <td>0.335481</td>\n",
              "      <td>0.088482</td>\n",
              "      <td>0.142370</td>\n",
              "      <td>0.001734</td>\n",
              "      <td>1634.465077</td>\n",
              "      <td>77425.41923</td>\n",
              "      <td>1954.633566</td>\n",
              "      <td>57359.69560</td>\n",
              "      <td>...</td>\n",
              "      <td>30.829542</td>\n",
              "      <td>0.635797</td>\n",
              "      <td>44.645561</td>\n",
              "      <td>1.591108</td>\n",
              "      <td>51.415859</td>\n",
              "      <td>-3.364909</td>\n",
              "      <td>26.421085</td>\n",
              "      <td>0.501505</td>\n",
              "      <td>29.109531</td>\n",
              "      <td>blues</td>\n",
              "    </tr>\n",
              "  </tbody>\n",
              "</table>\n",
              "<p>5 rows × 60 columns</p>\n",
              "</div>\n",
              "      <button class=\"colab-df-convert\" onclick=\"convertToInteractive('df-37aa7936-b9d6-482d-bb20-9631e85615e4')\"\n",
              "              title=\"Convert this dataframe to an interactive table.\"\n",
              "              style=\"display:none;\">\n",
              "        \n",
              "  <svg xmlns=\"http://www.w3.org/2000/svg\" height=\"24px\"viewBox=\"0 0 24 24\"\n",
              "       width=\"24px\">\n",
              "    <path d=\"M0 0h24v24H0V0z\" fill=\"none\"/>\n",
              "    <path d=\"M18.56 5.44l.94 2.06.94-2.06 2.06-.94-2.06-.94-.94-2.06-.94 2.06-2.06.94zm-11 1L8.5 8.5l.94-2.06 2.06-.94-2.06-.94L8.5 2.5l-.94 2.06-2.06.94zm10 10l.94 2.06.94-2.06 2.06-.94-2.06-.94-.94-2.06-.94 2.06-2.06.94z\"/><path d=\"M17.41 7.96l-1.37-1.37c-.4-.4-.92-.59-1.43-.59-.52 0-1.04.2-1.43.59L10.3 9.45l-7.72 7.72c-.78.78-.78 2.05 0 2.83L4 21.41c.39.39.9.59 1.41.59.51 0 1.02-.2 1.41-.59l7.78-7.78 2.81-2.81c.8-.78.8-2.07 0-2.86zM5.41 20L4 18.59l7.72-7.72 1.47 1.35L5.41 20z\"/>\n",
              "  </svg>\n",
              "      </button>\n",
              "      \n",
              "  <style>\n",
              "    .colab-df-container {\n",
              "      display:flex;\n",
              "      flex-wrap:wrap;\n",
              "      gap: 12px;\n",
              "    }\n",
              "\n",
              "    .colab-df-convert {\n",
              "      background-color: #E8F0FE;\n",
              "      border: none;\n",
              "      border-radius: 50%;\n",
              "      cursor: pointer;\n",
              "      display: none;\n",
              "      fill: #1967D2;\n",
              "      height: 32px;\n",
              "      padding: 0 0 0 0;\n",
              "      width: 32px;\n",
              "    }\n",
              "\n",
              "    .colab-df-convert:hover {\n",
              "      background-color: #E2EBFA;\n",
              "      box-shadow: 0px 1px 2px rgba(60, 64, 67, 0.3), 0px 1px 3px 1px rgba(60, 64, 67, 0.15);\n",
              "      fill: #174EA6;\n",
              "    }\n",
              "\n",
              "    [theme=dark] .colab-df-convert {\n",
              "      background-color: #3B4455;\n",
              "      fill: #D2E3FC;\n",
              "    }\n",
              "\n",
              "    [theme=dark] .colab-df-convert:hover {\n",
              "      background-color: #434B5C;\n",
              "      box-shadow: 0px 1px 3px 1px rgba(0, 0, 0, 0.15);\n",
              "      filter: drop-shadow(0px 1px 2px rgba(0, 0, 0, 0.3));\n",
              "      fill: #FFFFFF;\n",
              "    }\n",
              "  </style>\n",
              "\n",
              "      <script>\n",
              "        const buttonEl =\n",
              "          document.querySelector('#df-37aa7936-b9d6-482d-bb20-9631e85615e4 button.colab-df-convert');\n",
              "        buttonEl.style.display =\n",
              "          google.colab.kernel.accessAllowed ? 'block' : 'none';\n",
              "\n",
              "        async function convertToInteractive(key) {\n",
              "          const element = document.querySelector('#df-37aa7936-b9d6-482d-bb20-9631e85615e4');\n",
              "          const dataTable =\n",
              "            await google.colab.kernel.invokeFunction('convertToInteractive',\n",
              "                                                     [key], {});\n",
              "          if (!dataTable) return;\n",
              "\n",
              "          const docLinkHtml = 'Like what you see? Visit the ' +\n",
              "            '<a target=\"_blank\" href=https://colab.research.google.com/notebooks/data_table.ipynb>data table notebook</a>'\n",
              "            + ' to learn more about interactive tables.';\n",
              "          element.innerHTML = '';\n",
              "          dataTable['output_type'] = 'display_data';\n",
              "          await google.colab.output.renderOutput(dataTable, element);\n",
              "          const docLink = document.createElement('div');\n",
              "          docLink.innerHTML = docLinkHtml;\n",
              "          element.appendChild(docLink);\n",
              "        }\n",
              "      </script>\n",
              "    </div>\n",
              "  </div>\n",
              "  "
            ],
            "text/plain": [
              "            filename  length  chroma_stft_mean  chroma_stft_var  rms_mean  \\\n",
              "0  blues.00000.0.wav   66150          0.335555         0.090997  0.130189   \n",
              "1  blues.00000.1.wav   66150          0.343523         0.086782  0.112119   \n",
              "2  blues.00000.2.wav   66150          0.347746         0.092495  0.130895   \n",
              "3  blues.00000.3.wav   66150          0.363863         0.087207  0.131349   \n",
              "4  blues.00000.4.wav   66150          0.335481         0.088482  0.142370   \n",
              "\n",
              "    rms_var  spectral_centroid_mean  spectral_centroid_var  \\\n",
              "0  0.003559             1773.358004           169450.82950   \n",
              "1  0.001491             1817.244034            90766.29725   \n",
              "2  0.004552             1790.722357           110071.20700   \n",
              "3  0.002338             1660.545231           109496.93630   \n",
              "4  0.001734             1634.465077            77425.41923   \n",
              "\n",
              "   spectral_bandwidth_mean  spectral_bandwidth_var  ...  mfcc16_var  \\\n",
              "0              1972.334258            117272.64020  ...   39.547073   \n",
              "1              2010.751494             65940.66624  ...   64.819778   \n",
              "2              2088.184750             73391.49800  ...   68.306793   \n",
              "3              1967.920582             79805.90135  ...   48.543201   \n",
              "4              1954.633566             57359.69560  ...   30.829542   \n",
              "\n",
              "   mfcc17_mean  mfcc17_var  mfcc18_mean  mfcc18_var  mfcc19_mean  mfcc19_var  \\\n",
              "0    -3.230046   36.606857     0.696385   37.766129    -5.035945   33.668549   \n",
              "1    -6.025472   40.548805     0.127131   51.048939    -2.808956   97.221489   \n",
              "2    -1.714475   28.136948     2.329553   47.211426    -1.925621   52.922432   \n",
              "3    -3.786986   28.419544     1.153315   35.682701    -3.501979   50.610340   \n",
              "4     0.635797   44.645561     1.591108   51.415859    -3.364909   26.421085   \n",
              "\n",
              "   mfcc20_mean  mfcc20_var  label  \n",
              "0    -0.239585   43.818882  blues  \n",
              "1     5.771882   60.360352  blues  \n",
              "2     2.466996   33.164001  blues  \n",
              "3     3.580636   32.325874  blues  \n",
              "4     0.501505   29.109531  blues  \n",
              "\n",
              "[5 rows x 60 columns]"
            ]
          },
          "execution_count": 25,
          "metadata": {},
          "output_type": "execute_result"
        }
      ],
      "source": [
        "knn_path = path + \"/features_3_sec.csv\"\n",
        "knn_data = pd.read_csv(knn_path)\n",
        "knn_data.head()"
      ]
    },
    {
      "cell_type": "code",
      "execution_count": null,
      "metadata": {
        "id": "Vkt5dUcqV2Ct"
      },
      "outputs": [],
      "source": [
        "knn_data = knn_data.iloc[0:, 1:]\n",
        "knn_y = knn_data[\"label\"]\n",
        "knn_x = knn_data.loc[:, knn_data.columns != 'label']\n",
        "\n",
        "# Normalize X\n",
        "xcols = knn_x.columns\n",
        "min_max_scaler = preprocessing.MinMaxScaler()\n",
        "np_scaled = min_max_scaler.fit_transform(knn_x)\n",
        "\n",
        "knn_x = pd.DataFrame(np_scaled, columns = xcols)"
      ]
    },
    {
      "cell_type": "code",
      "execution_count": null,
      "metadata": {
        "id": "cC0IIGkBV3av"
      },
      "outputs": [],
      "source": [
        "knn_X_train, knn_X_test, knn_y_train, knn_y_test = train_test_split(knn_x, knn_y, test_size=0.2, random_state=32, shuffle=True)"
      ]
    },
    {
      "cell_type": "code",
      "execution_count": null,
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/"
        },
        "id": "MAjr1CyUZ4lM",
        "outputId": "9754e3e7-db40-4ea8-b3f1-e287f3d23b6a"
      },
      "outputs": [
        {
          "name": "stdout",
          "output_type": "stream",
          "text": [
            "Accuracy of KNN model for 3 seconds interval dataset, with 10 neighbours:  0.868\n"
          ]
        }
      ],
      "source": [
        "knn_10 = KNeighborsClassifier(n_neighbors=10)\n",
        "knn_10.fit(knn_X_train, knn_y_train)\n",
        "y_pred_10 = knn_10.predict(knn_X_test)\n",
        "print('Accuracy of KNN model for 3 seconds interval dataset, with 10 neighbours: ', round(accuracy_score(knn_y_test, y_pred_10), 5))"
      ]
    },
    {
      "cell_type": "code",
      "execution_count": null,
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/"
        },
        "id": "OtAVLDVEV4on",
        "outputId": "2d04f036-8635-43d0-990b-420daed3eaff"
      },
      "outputs": [
        {
          "name": "stdout",
          "output_type": "stream",
          "text": [
            "Accuracy of KNN model for 3 seconds interval dataset, with 20 neighbours:  0.8105\n"
          ]
        }
      ],
      "source": [
        "knn = KNeighborsClassifier(n_neighbors=20)\n",
        "knn.fit(knn_X_train, knn_y_train)\n",
        "y_pred = knn.predict(knn_X_test)\n",
        "print('Accuracy of KNN model for 3 seconds interval dataset, with 20 neighbours: ', round(accuracy_score(knn_y_test, y_pred), 5))"
      ]
    },
    {
      "cell_type": "code",
      "execution_count": null,
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/"
        },
        "id": "Y9eZ5WwLaOC-",
        "outputId": "7e210095-3770-476b-bfa9-6989386596fc"
      },
      "outputs": [
        {
          "name": "stdout",
          "output_type": "stream",
          "text": [
            "Accuracy of KNN model for 3 seconds interval dataset, with all neighbours:  0.692\n"
          ]
        }
      ],
      "source": [
        "knn_all = KNeighborsClassifier(n_neighbors=100)\n",
        "knn_all.fit(knn_X_train, knn_y_train)\n",
        "y_pred_all = knn_all.predict(knn_X_test)\n",
        "print('Accuracy of KNN model for 3 seconds interval dataset, with all neighbours: ', round(accuracy_score(knn_y_test, y_pred_all), 5))"
      ]
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "F0AgU90vQLiJ"
      },
      "source": [
        "## Subsection 2. 30 Second Model"
      ]
    },
    {
      "cell_type": "code",
      "execution_count": null,
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 299
        },
        "id": "AfFeyzQPQO7m",
        "outputId": "8aa70851-ed7b-47b4-9e2c-234618ed69d5"
      },
      "outputs": [
        {
          "data": {
            "text/html": [
              "\n",
              "  <div id=\"df-0ee9952c-a644-4aac-990a-86df61f93404\">\n",
              "    <div class=\"colab-df-container\">\n",
              "      <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>filename</th>\n",
              "      <th>length</th>\n",
              "      <th>chroma_stft_mean</th>\n",
              "      <th>chroma_stft_var</th>\n",
              "      <th>rms_mean</th>\n",
              "      <th>rms_var</th>\n",
              "      <th>spectral_centroid_mean</th>\n",
              "      <th>spectral_centroid_var</th>\n",
              "      <th>spectral_bandwidth_mean</th>\n",
              "      <th>spectral_bandwidth_var</th>\n",
              "      <th>...</th>\n",
              "      <th>mfcc16_var</th>\n",
              "      <th>mfcc17_mean</th>\n",
              "      <th>mfcc17_var</th>\n",
              "      <th>mfcc18_mean</th>\n",
              "      <th>mfcc18_var</th>\n",
              "      <th>mfcc19_mean</th>\n",
              "      <th>mfcc19_var</th>\n",
              "      <th>mfcc20_mean</th>\n",
              "      <th>mfcc20_var</th>\n",
              "      <th>label</th>\n",
              "    </tr>\n",
              "  </thead>\n",
              "  <tbody>\n",
              "    <tr>\n",
              "      <th>0</th>\n",
              "      <td>jazz.00000.wav</td>\n",
              "      <td>661500.0</td>\n",
              "      <td>0.344568</td>\n",
              "      <td>0.085005</td>\n",
              "      <td>0.046768</td>\n",
              "      <td>0.001543</td>\n",
              "      <td>1504.404759</td>\n",
              "      <td>554635.808130</td>\n",
              "      <td>1754.409592</td>\n",
              "      <td>283727.376842</td>\n",
              "      <td>...</td>\n",
              "      <td>50.317745</td>\n",
              "      <td>-1.502332</td>\n",
              "      <td>41.169983</td>\n",
              "      <td>0.221480</td>\n",
              "      <td>55.749470</td>\n",
              "      <td>-1.991628</td>\n",
              "      <td>50.034637</td>\n",
              "      <td>-3.348313</td>\n",
              "      <td>49.907917</td>\n",
              "      <td>jazz</td>\n",
              "    </tr>\n",
              "    <tr>\n",
              "      <th>1</th>\n",
              "      <td>jazz.00003.wav</td>\n",
              "      <td>661500.0</td>\n",
              "      <td>0.319313</td>\n",
              "      <td>0.081005</td>\n",
              "      <td>0.059496</td>\n",
              "      <td>0.001298</td>\n",
              "      <td>813.293729</td>\n",
              "      <td>92950.399709</td>\n",
              "      <td>1058.898921</td>\n",
              "      <td>220025.916341</td>\n",
              "      <td>...</td>\n",
              "      <td>47.350842</td>\n",
              "      <td>-2.272845</td>\n",
              "      <td>54.629692</td>\n",
              "      <td>3.883317</td>\n",
              "      <td>38.872822</td>\n",
              "      <td>0.344252</td>\n",
              "      <td>50.070782</td>\n",
              "      <td>-2.214156</td>\n",
              "      <td>56.304241</td>\n",
              "      <td>jazz</td>\n",
              "    </tr>\n",
              "    <tr>\n",
              "      <th>2</th>\n",
              "      <td>jazz.00002.wav</td>\n",
              "      <td>661500.0</td>\n",
              "      <td>0.267640</td>\n",
              "      <td>0.083949</td>\n",
              "      <td>0.080840</td>\n",
              "      <td>0.000992</td>\n",
              "      <td>1064.810564</td>\n",
              "      <td>161183.061715</td>\n",
              "      <td>1050.419172</td>\n",
              "      <td>62157.266400</td>\n",
              "      <td>...</td>\n",
              "      <td>39.972137</td>\n",
              "      <td>-12.295748</td>\n",
              "      <td>54.781208</td>\n",
              "      <td>-6.769270</td>\n",
              "      <td>45.765415</td>\n",
              "      <td>-3.495534</td>\n",
              "      <td>41.461681</td>\n",
              "      <td>-8.644752</td>\n",
              "      <td>73.453781</td>\n",
              "      <td>jazz</td>\n",
              "    </tr>\n",
              "    <tr>\n",
              "      <th>3</th>\n",
              "      <td>jazz.00005.wav</td>\n",
              "      <td>661500.0</td>\n",
              "      <td>0.248236</td>\n",
              "      <td>0.096418</td>\n",
              "      <td>0.107555</td>\n",
              "      <td>0.002236</td>\n",
              "      <td>1268.061497</td>\n",
              "      <td>286543.046719</td>\n",
              "      <td>1546.300122</td>\n",
              "      <td>236694.493580</td>\n",
              "      <td>...</td>\n",
              "      <td>229.317825</td>\n",
              "      <td>0.863595</td>\n",
              "      <td>405.584015</td>\n",
              "      <td>0.953356</td>\n",
              "      <td>332.244568</td>\n",
              "      <td>1.736951</td>\n",
              "      <td>193.131317</td>\n",
              "      <td>-1.345491</td>\n",
              "      <td>212.923080</td>\n",
              "      <td>jazz</td>\n",
              "    </tr>\n",
              "    <tr>\n",
              "      <th>4</th>\n",
              "      <td>jazz.00004.wav</td>\n",
              "      <td>661500.0</td>\n",
              "      <td>0.171782</td>\n",
              "      <td>0.082596</td>\n",
              "      <td>0.108786</td>\n",
              "      <td>0.002178</td>\n",
              "      <td>1039.623420</td>\n",
              "      <td>122521.889902</td>\n",
              "      <td>1422.303098</td>\n",
              "      <td>77324.475593</td>\n",
              "      <td>...</td>\n",
              "      <td>162.859573</td>\n",
              "      <td>1.991310</td>\n",
              "      <td>181.093201</td>\n",
              "      <td>4.985074</td>\n",
              "      <td>162.339539</td>\n",
              "      <td>6.582728</td>\n",
              "      <td>393.453186</td>\n",
              "      <td>-2.426803</td>\n",
              "      <td>473.187347</td>\n",
              "      <td>jazz</td>\n",
              "    </tr>\n",
              "  </tbody>\n",
              "</table>\n",
              "<p>5 rows × 60 columns</p>\n",
              "</div>\n",
              "      <button class=\"colab-df-convert\" onclick=\"convertToInteractive('df-0ee9952c-a644-4aac-990a-86df61f93404')\"\n",
              "              title=\"Convert this dataframe to an interactive table.\"\n",
              "              style=\"display:none;\">\n",
              "        \n",
              "  <svg xmlns=\"http://www.w3.org/2000/svg\" height=\"24px\"viewBox=\"0 0 24 24\"\n",
              "       width=\"24px\">\n",
              "    <path d=\"M0 0h24v24H0V0z\" fill=\"none\"/>\n",
              "    <path d=\"M18.56 5.44l.94 2.06.94-2.06 2.06-.94-2.06-.94-.94-2.06-.94 2.06-2.06.94zm-11 1L8.5 8.5l.94-2.06 2.06-.94-2.06-.94L8.5 2.5l-.94 2.06-2.06.94zm10 10l.94 2.06.94-2.06 2.06-.94-2.06-.94-.94-2.06-.94 2.06-2.06.94z\"/><path d=\"M17.41 7.96l-1.37-1.37c-.4-.4-.92-.59-1.43-.59-.52 0-1.04.2-1.43.59L10.3 9.45l-7.72 7.72c-.78.78-.78 2.05 0 2.83L4 21.41c.39.39.9.59 1.41.59.51 0 1.02-.2 1.41-.59l7.78-7.78 2.81-2.81c.8-.78.8-2.07 0-2.86zM5.41 20L4 18.59l7.72-7.72 1.47 1.35L5.41 20z\"/>\n",
              "  </svg>\n",
              "      </button>\n",
              "      \n",
              "  <style>\n",
              "    .colab-df-container {\n",
              "      display:flex;\n",
              "      flex-wrap:wrap;\n",
              "      gap: 12px;\n",
              "    }\n",
              "\n",
              "    .colab-df-convert {\n",
              "      background-color: #E8F0FE;\n",
              "      border: none;\n",
              "      border-radius: 50%;\n",
              "      cursor: pointer;\n",
              "      display: none;\n",
              "      fill: #1967D2;\n",
              "      height: 32px;\n",
              "      padding: 0 0 0 0;\n",
              "      width: 32px;\n",
              "    }\n",
              "\n",
              "    .colab-df-convert:hover {\n",
              "      background-color: #E2EBFA;\n",
              "      box-shadow: 0px 1px 2px rgba(60, 64, 67, 0.3), 0px 1px 3px 1px rgba(60, 64, 67, 0.15);\n",
              "      fill: #174EA6;\n",
              "    }\n",
              "\n",
              "    [theme=dark] .colab-df-convert {\n",
              "      background-color: #3B4455;\n",
              "      fill: #D2E3FC;\n",
              "    }\n",
              "\n",
              "    [theme=dark] .colab-df-convert:hover {\n",
              "      background-color: #434B5C;\n",
              "      box-shadow: 0px 1px 3px 1px rgba(0, 0, 0, 0.15);\n",
              "      filter: drop-shadow(0px 1px 2px rgba(0, 0, 0, 0.3));\n",
              "      fill: #FFFFFF;\n",
              "    }\n",
              "  </style>\n",
              "\n",
              "      <script>\n",
              "        const buttonEl =\n",
              "          document.querySelector('#df-0ee9952c-a644-4aac-990a-86df61f93404 button.colab-df-convert');\n",
              "        buttonEl.style.display =\n",
              "          google.colab.kernel.accessAllowed ? 'block' : 'none';\n",
              "\n",
              "        async function convertToInteractive(key) {\n",
              "          const element = document.querySelector('#df-0ee9952c-a644-4aac-990a-86df61f93404');\n",
              "          const dataTable =\n",
              "            await google.colab.kernel.invokeFunction('convertToInteractive',\n",
              "                                                     [key], {});\n",
              "          if (!dataTable) return;\n",
              "\n",
              "          const docLinkHtml = 'Like what you see? Visit the ' +\n",
              "            '<a target=\"_blank\" href=https://colab.research.google.com/notebooks/data_table.ipynb>data table notebook</a>'\n",
              "            + ' to learn more about interactive tables.';\n",
              "          element.innerHTML = '';\n",
              "          dataTable['output_type'] = 'display_data';\n",
              "          await google.colab.output.renderOutput(dataTable, element);\n",
              "          const docLink = document.createElement('div');\n",
              "          docLink.innerHTML = docLinkHtml;\n",
              "          element.appendChild(docLink);\n",
              "        }\n",
              "      </script>\n",
              "    </div>\n",
              "  </div>\n",
              "  "
            ],
            "text/plain": [
              "         filename    length  chroma_stft_mean  chroma_stft_var  rms_mean  \\\n",
              "0  jazz.00000.wav  661500.0          0.344568         0.085005  0.046768   \n",
              "1  jazz.00003.wav  661500.0          0.319313         0.081005  0.059496   \n",
              "2  jazz.00002.wav  661500.0          0.267640         0.083949  0.080840   \n",
              "3  jazz.00005.wav  661500.0          0.248236         0.096418  0.107555   \n",
              "4  jazz.00004.wav  661500.0          0.171782         0.082596  0.108786   \n",
              "\n",
              "    rms_var  spectral_centroid_mean  spectral_centroid_var  \\\n",
              "0  0.001543             1504.404759          554635.808130   \n",
              "1  0.001298              813.293729           92950.399709   \n",
              "2  0.000992             1064.810564          161183.061715   \n",
              "3  0.002236             1268.061497          286543.046719   \n",
              "4  0.002178             1039.623420          122521.889902   \n",
              "\n",
              "   spectral_bandwidth_mean  spectral_bandwidth_var  ...  mfcc16_var  \\\n",
              "0              1754.409592           283727.376842  ...   50.317745   \n",
              "1              1058.898921           220025.916341  ...   47.350842   \n",
              "2              1050.419172            62157.266400  ...   39.972137   \n",
              "3              1546.300122           236694.493580  ...  229.317825   \n",
              "4              1422.303098            77324.475593  ...  162.859573   \n",
              "\n",
              "   mfcc17_mean  mfcc17_var  mfcc18_mean  mfcc18_var  mfcc19_mean  mfcc19_var  \\\n",
              "0    -1.502332   41.169983     0.221480   55.749470    -1.991628   50.034637   \n",
              "1    -2.272845   54.629692     3.883317   38.872822     0.344252   50.070782   \n",
              "2   -12.295748   54.781208    -6.769270   45.765415    -3.495534   41.461681   \n",
              "3     0.863595  405.584015     0.953356  332.244568     1.736951  193.131317   \n",
              "4     1.991310  181.093201     4.985074  162.339539     6.582728  393.453186   \n",
              "\n",
              "   mfcc20_mean  mfcc20_var  label  \n",
              "0    -3.348313   49.907917   jazz  \n",
              "1    -2.214156   56.304241   jazz  \n",
              "2    -8.644752   73.453781   jazz  \n",
              "3    -1.345491  212.923080   jazz  \n",
              "4    -2.426803  473.187347   jazz  \n",
              "\n",
              "[5 rows x 60 columns]"
            ]
          },
          "execution_count": 28,
          "metadata": {},
          "output_type": "execute_result"
        }
      ],
      "source": [
        "knn_30_path = path + \"/features_30_sec.csv\"\n",
        "knn_30_data = pd.read_csv(knn_30_path)\n",
        "knn_30_data.head()"
      ]
    },
    {
      "cell_type": "code",
      "execution_count": null,
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/"
        },
        "id": "ucXQKwfAU3eX",
        "outputId": "ac8d5bd5-d6ed-4051-f8cf-872270f70b3b"
      },
      "outputs": [
        {
          "data": {
            "text/plain": [
              "1000"
            ]
          },
          "execution_count": 35,
          "metadata": {},
          "output_type": "execute_result"
        }
      ],
      "source": [
        "len(knn_30_data)"
      ]
    },
    {
      "cell_type": "code",
      "execution_count": null,
      "metadata": {
        "id": "q4QFVwg5QVvY"
      },
      "outputs": [],
      "source": [
        "knn_30_data = knn_30_data.iloc[0:, 1:]\n",
        "knn_30_y = knn_30_data[\"label\"]\n",
        "knn_30_x = knn_30_data.loc[:, knn_30_data.columns != 'label']\n",
        "\n",
        "# Normalize X\n",
        "x30cols = knn_30_x.columns\n",
        "min_max_scaler = preprocessing.MinMaxScaler()\n",
        "np_scaled = min_max_scaler.fit_transform(knn_30_x)\n",
        "\n",
        "knn_30_x = pd.DataFrame(np_scaled, columns = x30cols)"
      ]
    },
    {
      "cell_type": "code",
      "execution_count": null,
      "metadata": {
        "id": "qotBQTsRQnQU"
      },
      "outputs": [],
      "source": [
        "knn30_X_train, knn30_X_test, knn30_y_train, knn30_y_test = train_test_split(knn_30_x, knn_30_y, test_size=0.3, random_state=42)"
      ]
    },
    {
      "cell_type": "code",
      "execution_count": null,
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/"
        },
        "id": "5HJHVckHc6Ly",
        "outputId": "f705b1dc-1fe9-40e7-9f3e-33fedde01e4f"
      },
      "outputs": [
        {
          "name": "stdout",
          "output_type": "stream",
          "text": [
            "Accuracy of KNN model with 30 seconds interval dataset, 5 nearest neighbours:  0.62667\n"
          ]
        }
      ],
      "source": [
        "knn30_5 = KNeighborsClassifier(n_neighbors=10)\n",
        "knn30_5.fit(knn30_X_train, knn30_y_train)\n",
        "y30_5_pred = knn30_5.predict(knn30_X_test)\n",
        "print('Accuracy of KNN model with 30 seconds interval dataset, 5 nearest neighbours: ', round(accuracy_score(knn30_y_test, y30_5_pred), 5))"
      ]
    },
    {
      "cell_type": "code",
      "execution_count": null,
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/"
        },
        "id": "hPbIncw8QvNH",
        "outputId": "049be619-3b44-4cb9-c5fb-544365047d2c"
      },
      "outputs": [
        {
          "name": "stdout",
          "output_type": "stream",
          "text": [
            "Accuracy of KNN model:  0.57333\n"
          ]
        }
      ],
      "source": [
        "knn30 = KNeighborsClassifier(n_neighbors=20)\n",
        "knn30.fit(knn30_X_train, knn30_y_train)\n",
        "y30_pred = knn30.predict(knn30_X_test)\n",
        "print('Accuracy of KNN model: ', round(accuracy_score(knn30_y_test, y30_pred), 5))"
      ]
    },
    {
      "cell_type": "code",
      "execution_count": null,
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/"
        },
        "id": "_0PidIpEUuwH",
        "outputId": "332f7477-54dc-422d-9325-4e3338cb76fc"
      },
      "outputs": [
        {
          "name": "stdout",
          "output_type": "stream",
          "text": [
            "Accuracy of KNN model:  0.44667\n"
          ]
        }
      ],
      "source": [
        "knn30_ALL = KNeighborsClassifier(n_neighbors=100)\n",
        "knn30_ALL.fit(knn30_X_train, knn30_y_train)\n",
        "y30_ALL_pred = knn30_ALL.predict(knn30_X_test)\n",
        "print('Accuracy of KNN model: ', round(accuracy_score(knn30_y_test, y30_ALL_pred), 5))"
      ]
    },
    {
      "cell_type": "code",
      "execution_count": null,
      "metadata": {
        "id": "wWGG0H7cvKW8"
      },
      "outputs": [],
      "source": [
        "from sklearn.decomposition import PCA\n",
        "\n",
        "pca = PCA(n_components=2)\n",
        "principalComponents = pca.fit_transform(knn_x)\n",
        "principalDf = pd.DataFrame(data = principalComponents, columns = ['principal component 1', 'principal component 2'])\n",
        "\n",
        "pcadf = pd.concat([principalDf, knn_y], axis = 1)\n",
        "\n",
        "pca_y = pcadf[\"label\"]\n",
        "pca_x = pcadf.loc[:, pcadf.columns != 'label']"
      ]
    },
    {
      "cell_type": "code",
      "execution_count": null,
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/"
        },
        "id": "EA8HGLaRv4nB",
        "outputId": "9347047d-88fe-4286-ac89-ea76d6a088a3"
      },
      "outputs": [
        {
          "name": "stderr",
          "output_type": "stream",
          "text": [
            "/usr/local/lib/python3.7/dist-packages/sklearn/base.py:451: UserWarning: X does not have valid feature names, but KNeighborsClassifier was fitted with feature names\n",
            "  \"X does not have valid feature names, but\"\n"
          ]
        }
      ],
      "source": [
        "from matplotlib.colors import ListedColormap\n",
        "# split data into training and testing set\n",
        "X_train, X_test, y_train, y_test = train_test_split(pca_x, pca_y, test_size = 0.33, random_state = 41)\n",
        "\n",
        "# init vars\n",
        "n_neighbors = 5\n",
        "h = .02  # step size in the mesh\n",
        "\n",
        "# Create color maps\n",
        "cmap_light = ListedColormap(['#FFAAAA', '#AAAAFF'])\n",
        "cmap_bold  = ListedColormap(['#FF0000', '#0000FF'])\n",
        "\n",
        "knn_pca = KNeighborsClassifier(n_neighbors=10)\n",
        "knn_pca.fit(X_train, y_train)\n",
        "\n",
        "x_min, x_max = pca_x.to_numpy()[:, 0].min() - 1, pca_x.to_numpy()[:, 0].max() + 1\n",
        "y_min, y_max = pca_x.to_numpy()[:, 1].min() - 1, pca_x.to_numpy()[:, 1].max() + 1\n",
        "xx, yy = np.meshgrid(np.arange(x_min, x_max, h),\n",
        "                      np.arange(y_min, y_max, h))\n",
        "Z = knn_pca.predict(np.c_[xx.ravel(), yy.ravel()])"
      ]
    },
    {
      "cell_type": "code",
      "execution_count": null,
      "metadata": {
        "id": "RybLH_a2wogW"
      },
      "outputs": [],
      "source": [
        "color_list = ListedColormap([\"Accent\", \"Blues\", \"BuGn\", \"PiYG\", \"PuBuGn_r\", \"Purples\", \"RdYlBu_r\", \"autumn\",\"bone\", \"gist_heat\"])\n",
        "cmap_light = ListedColormap(['#FFAAAA', '#AAAAFF'])\n",
        "\n",
        "Z = Z.reshape(xx.shape)\n",
        "fig = plt.figure()\n",
        "\n",
        "plt.pcolormesh(xx, yy, Z, cmap = cmap_light)\n",
        "\n",
        "# Plot also the training points, x-axis = 'Glucose', y-axis = \"BMI\"\n",
        "# plt.scatter(pca_x.to_numpy()[:, 0], pca_x.to_numpy()[:, 1], c = color_list)\n",
        "# plt.xlim(xx.min(), xx.max())\n",
        "# plt.ylim(yy.min(), yy.max())\n",
        "# plt.title(\"0/1 outcome classification (k = %i, weights = '%s')\" % (5, \"Default\"))\n",
        "# plt.show()"
      ]
    },
    {
      "cell_type": "code",
      "execution_count": null,
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/"
        },
        "id": "UYGkEq0r2ArL",
        "outputId": "c674e5dc-d0ad-4a52-e492-4c48dba8575d"
      },
      "outputs": [
        {
          "data": {
            "text/plain": [
              "(192, 215)"
            ]
          },
          "execution_count": 160,
          "metadata": {},
          "output_type": "execute_result"
        }
      ],
      "source": [
        "Z.shape"
      ]
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "sY8WWwJtwaLc"
      },
      "source": [
        "# CNN Model"
      ]
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "YZ_rgHx2w6Gh"
      },
      "source": [
        "Data Loading & Splitting"
      ]
    },
    {
      "cell_type": "code",
      "execution_count": null,
      "metadata": {
        "id": "KOeWIW16UPFk"
      },
      "outputs": [],
      "source": [
        "def get_data(batch_size = 32):\n",
        "\n",
        "  mel_path = path + \"/MFCC_imgs\"\n",
        "  transform = transforms.Compose([transforms.ToTensor(),\n",
        "      transforms.Normalize((0.5, 0.5, 0.5, 0.5), (0.5, 0.5, 0.5, 0.5))])  #convert data into a tensor, and then normalize it\n",
        "\n",
        "  torch.manual_seed(620)\n",
        "\n",
        "  train_set = torchvision.datasets.ImageFolder(mel_path + \"/training\", loader = plt.imread, transform = transform)\n",
        "  valid_set = torchvision.datasets.ImageFolder(mel_path + \"/valid\", loader = plt.imread, transform = transform)\n",
        "\n",
        "  train_loader = torch.utils.data.DataLoader(train_set, batch_size=batch_size,\n",
        "                                               num_workers=1, shuffle=True)\n",
        "  val_loader = torch.utils.data.DataLoader(valid_set, batch_size=batch_size,\n",
        "                                               num_workers=1, shuffle=True)\n",
        "\n",
        "  return train_loader, val_loader"
      ]
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "XhDoHt3wUEYt"
      },
      "source": [
        "Accuracy Function"
      ]
    },
    {
      "cell_type": "code",
      "execution_count": null,
      "metadata": {
        "id": "4DXNK3BdUEBl"
      },
      "outputs": [],
      "source": [
        "def get_accuracy(net, loader, use_cuda = True):\n",
        "  total_accuracy = 0\n",
        "  total_epoch = 0\n",
        "\n",
        "  for imgs, labels in loader:\n",
        "    output = net(imgs)\n",
        "    pred = output.max(1, keepdim = True)[1]\n",
        "    total_accuracy += pred.eq(labels.view_as(pred)).sum().item()\n",
        "    total_epoch += labels.shape[0]\n",
        "  accuracy = float(total_accuracy) / total_epoch\n",
        "\n",
        "  return accuracy"
      ]
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "IrjHyiijwieb"
      },
      "source": [
        "Initialize CNN\n"
      ]
    },
    {
      "cell_type": "code",
      "execution_count": null,
      "metadata": {
        "id": "sL11GmedwgCw"
      },
      "outputs": [],
      "source": [
        "%matplotlib inline\n",
        "\n",
        "genre_count = 10 #num_genres\n",
        "\n",
        "#include drop_out\n",
        "\n",
        "\n",
        "class CNN(nn.Module):\n",
        "  def __init__(self, c1 = {\"in_ch\":4,\"out_ch\":8,\"k\":3, \"str\":1,\"pad\":0}, c2 = {\"in_ch\":8,\"out_ch\":16,\"k\":3,\"str\":1,\"pad\":0},\n",
        "               c3 =  {\"in_ch\":16,\"out_ch\":32,\"k\":3,\"str\":1,\"pad\":0}, c4 =  {\"in_ch\":32,\"out_ch\":64, \"k\":3,\"str\":1,\"pad\":0}, hyparam_fc1 = {\"size\":64}, gc = genre_count, dropout = 0.25):\n",
        "    torch.manual_seed(1000)\n",
        "    super(CNN, self).__init__()\n",
        "    self.conv1 = nn.Conv2d(c1[\"in_ch\"],c1[\"out_ch\"],c1[\"k\"],c1[\"str\"],padding = c1[\"pad\"])\n",
        "    self.pool1 = nn.MaxPool2d(2,2)\n",
        "    self.conv2 = nn.Conv2d(c2[\"in_ch\"],c2[\"out_ch\"],c2[\"k\"],c2[\"str\"],padding = c2[\"pad\"])\n",
        "    self.pool2 = nn.MaxPool2d(2,2)\n",
        "    self.conv3 = nn.Conv2d(c3[\"in_ch\"],c3[\"out_ch\"],c3[\"k\"],c3[\"str\"],padding = c3[\"pad\"])\n",
        "    self.pool3 = nn.MaxPool2d(2,2)\n",
        "    self.conv4 = nn.Conv2d(c4[\"in_ch\"],c4[\"out_ch\"],c4[\"k\"],c4[\"str\"],padding = c4[\"pad\"])\n",
        "\n",
        "    #computing input layer size assumming 4x288x432 sized images\n",
        "    self.paramsx1 = floor((288 - c1[\"k\"] + c1[\"pad\"])/c1[\"str\"]) + 1  #Convolution layer 1 x output\n",
        "    self.paramsx2 = floor((self.paramsx1 - 2)/2) + 1  #Pooling layer 1 x output\n",
        "    self.paramsx3 = floor((self.paramsx2 - c2[\"k\"] + c2[\"pad\"])/c2[\"str\"]) + 1   #Convolution layer 2 x output\n",
        "    self.paramsx4 = floor((self.paramsx3 - 2)/2) + 1  #Pooling layer 2 x output\n",
        "    self.paramsx5 = floor((self.paramsx4 - c3[\"k\"] + c3[\"pad\"])/c3[\"str\"]) + 1   #Convolution layer 2 x output\n",
        "    self.paramsx6 = floor((self.paramsx5 - 2)/2) + 1  #Pooling layer 2 x output\n",
        "    self.paramsx7 = floor((self.paramsx6 - c4[\"k\"] + c4[\"pad\"])/c4[\"str\"]) + 1   #Convolution layer 2 x output\n",
        "    self.paramsx8 = floor((self.paramsx7 - 2)/2) + 1  #Pooling layer 2 x output\n",
        "    # print(\"self.paramsx8 = \", self.paramsx8)\n",
        "\n",
        "    self.paramsy1 = floor((432 - c1[\"k\"] + c1[\"pad\"])/c1[\"str\"]) + 1\n",
        "    self.paramsy2 = floor((self.paramsy1 - 2)/2) + 1  #Pooling layer 1 y output\n",
        "    self.paramsy3 = floor((self.paramsy2 - c2[\"k\"] + c2[\"pad\"])/c2[\"str\"]) + 1   #Convolution layer 2 y output\n",
        "    self.paramsy4 = floor((self.paramsy3 - 2)/2) + 1  #Pooling layer 2 y output\n",
        "    self.paramsy5 = floor((self.paramsy4 - c3[\"k\"] + c3[\"pad\"])/c3[\"str\"]) + 1   #Convolution layer 2 x output\n",
        "    self.paramsy6 = floor((self.paramsy5 - 2)/2) + 1  #Pooling layer 2 x output\n",
        "    self.paramsy7 = floor((self.paramsy6 - c4[\"k\"] + c4[\"pad\"])/c4[\"str\"]) + 1   #Convolution layer 2 x output\n",
        "    self.paramsy8 = floor((self.paramsy7 - 2)/2) + 1  #Pooling layer 2 x output\n",
        "    # print(\"self.paramsy8 = \", self.paramsy8)\n",
        "\n",
        "\n",
        "    self.fc1 = nn.Linear(c4[\"out_ch\"]*self.paramsx8*self.paramsy8,hyparam_fc1[\"size\"])\n",
        "    self.fc2 = nn.Linear(hyparam_fc1[\"size\"],gc)\n",
        "    self.dropout = nn.Dropout(dropout)\n",
        "\n",
        "  def forward(self,x):\n",
        "    x = self.pool1(F.relu(self.conv1(x)))\n",
        "    x = self.pool1(F.relu(self.conv2(x)))\n",
        "    x = self.pool2(F.relu(self.conv3(x)))\n",
        "    x = self.pool3(F.relu(self.conv4(x)))\n",
        "    # print(\"x.shape = \", x.shape)\n",
        "    # print(\"x.shape[1] = \", x.shape[1])\n",
        "    x = x.view(-1,x.shape[1]*self.paramsx8*self.paramsy8)\n",
        "    x = F.relu(self.fc1(x))\n",
        "    x = self.fc2(x)  #use nn.crossentropy loss to automatically apply signmoid activation\n",
        "    return x"
      ]
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "O_mpMmq8UJfO"
      },
      "source": [
        "Training Function\n"
      ]
    },
    {
      "cell_type": "code",
      "execution_count": null,
      "metadata": {
        "id": "DM4pCWEbUQc8"
      },
      "outputs": [],
      "source": [
        "def train(model, lr = 0.001, epochs = 18, criterion = \"CE\", optimizer = \"Adam\", mom=0.5, batch_size = 32, weight_decay = 0):\n",
        "  '''inputs: nn model, training and validation data (torch.utils.data.DataLoader class), learning rate lr, no. of epochs, loss function criterion\n",
        "  outputs: trained model, accuracy plots'''\n",
        "  train_data, val_data = get_data(batch_size = batch_size)\n",
        "  #print(\"train_data = \", train_data)\n",
        "  loss_dict = {\"CE\": nn.CrossEntropyLoss()}\n",
        "  loss_f = loss_dict[criterion]\n",
        "\n",
        "  optim_dict = {\"SGD\": torch.optim.SGD(model.parameters(), lr, momentum=mom, weight_decay = weight_decay),\"Adam\":torch.optim.Adam(model.parameters(), lr = lr, weight_decay = weight_decay)}\n",
        "  opt = optim_dict[optimizer]\n",
        "  #optimizer = optim.Adam(model.parameters(), lr=lr, weight_decay = weight_decay)\n",
        "  eps = []\n",
        "  losses = []\n",
        "  train_accuracy = []\n",
        "  val_accuracy = []\n",
        "  patience = 0\n",
        "\n",
        "  #train model\n",
        "  start_time = time.time()\n",
        "  for epoch in range(epochs):\n",
        "    print(\"epoch = \", epoch)\n",
        "    iteration = 0\n",
        "\n",
        "    #early stop with patience\n",
        "    if epoch > 1:\n",
        "      if val_accuracy[epoch - 2] > val_accuracy[epoch - 1]:\n",
        "        patience+=1\n",
        "        print(\"patience = \", patience)\n",
        "      if patience > 8:\n",
        "        print(\"Error: Validation Accuracy Run-Off\")\n",
        "        break\n",
        "      if epoch > 13 and val_accuracy[epoch - 1]<0.25:\n",
        "        print(\"Error: Low Validation Accuracy\")\n",
        "        break\n",
        "\n",
        "\n",
        "    for (images, labels) in train_data:\n",
        "      #print(\"batch_no. = \", iteration)\n",
        "      iteration+=1\n",
        "      #print(\"shape =\", images.shape)\n",
        "\n",
        "    #make passes\n",
        "      outputs = model(images)    #make prediction; softmax applied by criterion\n",
        "      loss = loss_f(outputs, labels)\n",
        "      loss.backward()    #calculate gradient\n",
        "      opt.step()   #update model parameters\n",
        "      opt.zero_grad()   #clean up\n",
        "      #optimizer.step()\n",
        "      #optimizer.zero_grad()\n",
        "\n",
        "    #record information for epoch\n",
        "    eps.append(epoch)\n",
        "    losses.append(float(loss)/batch_size)\n",
        "    train_accuracy.append(get_accuracy(model,train_data))\n",
        "    val_acc = get_accuracy(model,val_data)\n",
        "    print(\"val acc: \", val_acc)\n",
        "    val_accuracy.append(val_acc)\n",
        "  end_time = time.time()\n",
        "  elapsed_time = end_time - start_time\n",
        "  print(\"Total time elapsed: {:.2f} seconds\".format(elapsed_time))\n",
        "  print(\"Final Training Loss: \", losses[-1])\n",
        "  print(\"Final Training Accuracy: \", train_accuracy[-1])\n",
        "  print(\"Final Val Accuracy: \", val_accuracy[-1])\n",
        "\n",
        "  #plotting\n",
        "  plt.title(\"Training Loss vs. Epochs\")\n",
        "  plt.plot(eps, losses, label=\"Training\")\n",
        "  plt.xlabel(\"Epochs\")\n",
        "  plt.ylabel(\"Loss\")\n",
        "  plt.show()\n",
        "\n",
        "  plt.title(\"Training Accuracy vs. Epochs\")\n",
        "  plt.plot(eps, train_accuracy, label=\"Train\")\n",
        "  plt.plot(eps, val_accuracy, label=\"Validation\")\n",
        "  plt.xlabel(\"Epochs\")\n",
        "  plt.ylabel(\"Accuracy\")\n",
        "  plt.legend(loc='best')\n",
        "  plt.show()\n",
        "\n",
        "  return model\n",
        "\n",
        "\n"
      ]
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "IvsHeTpW1dNR"
      },
      "source": [
        "Training & Hyperparameter Optimization"
      ]
    },
    {
      "cell_type": "code",
      "execution_count": null,
      "metadata": {
        "id": "44wla5SIv_CJ"
      },
      "outputs": [],
      "source": [
        "model = CNN()\n",
        "\n",
        "try4 = train(model, epochs = 1, batch_size = 64, lr = 0.001, optimizer =\"Adam\")"
      ]
    },
    {
      "cell_type": "code",
      "execution_count": null,
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 1000
        },
        "id": "PH_Z8xN4SIlc",
        "outputId": "2655566b-a97a-47bd-ab2f-06910f96dafc"
      },
      "outputs": [
        {
          "name": "stdout",
          "output_type": "stream",
          "text": [
            "train_data =  <torch.utils.data.dataloader.DataLoader object at 0x7efdd9af6050>\n",
            "epoch =  0\n",
            "val acc:  0.22\n",
            "epoch =  1\n",
            "val acc:  0.22\n",
            "epoch =  2\n",
            "val acc:  0.26666666666666666\n",
            "epoch =  3\n",
            "val acc:  0.32\n",
            "epoch =  4\n",
            "val acc:  0.37333333333333335\n",
            "epoch =  5\n",
            "val acc:  0.3466666666666667\n",
            "epoch =  6\n",
            "patience =  1\n",
            "val acc:  0.38\n",
            "epoch =  7\n",
            "val acc:  0.38666666666666666\n",
            "epoch =  8\n",
            "val acc:  0.4066666666666667\n",
            "epoch =  9\n",
            "val acc:  0.46\n",
            "epoch =  10\n",
            "val acc:  0.5466666666666666\n",
            "epoch =  11\n",
            "val acc:  0.48\n",
            "epoch =  12\n",
            "patience =  2\n",
            "val acc:  0.49333333333333335\n",
            "epoch =  13\n",
            "val acc:  0.52\n",
            "epoch =  14\n",
            "val acc:  0.5466666666666666\n",
            "epoch =  15\n",
            "val acc:  0.5533333333333333\n",
            "epoch =  16\n",
            "val acc:  0.56\n",
            "epoch =  17\n",
            "val acc:  0.5466666666666666\n",
            "epoch =  18\n",
            "patience =  3\n",
            "val acc:  0.5466666666666666\n",
            "epoch =  19\n",
            "val acc:  0.5466666666666666\n",
            "Total time elapsed: 1475.18 seconds\n",
            "Final Training Loss:  0.010166766121983528\n",
            "Final Training Accuracy:  0.7364705882352941\n",
            "Final Val Accuracy:  0.5466666666666666\n"
          ]
        },
        {
          "data": {
            "image/png": "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\n",
            "text/plain": [
              "<Figure size 432x288 with 1 Axes>"
            ]
          },
          "metadata": {},
          "output_type": "display_data"
        },
        {
          "data": {
            "image/png": "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\n",
            "text/plain": [
              "<Figure size 432x288 with 1 Axes>"
            ]
          },
          "metadata": {},
          "output_type": "display_data"
        }
      ],
      "source": [
        "model5 = CNN()\n",
        "\n",
        "try5 = train(model5, epochs = 20, batch_size = 64, lr = 0.001, optimizer =\"Adam\")"
      ]
    },
    {
      "cell_type": "code",
      "execution_count": null,
      "metadata": {
        "colab": {
          "background_save": true,
          "base_uri": "https://localhost:8080/"
        },
        "id": "L4bLN6ZxQJ5I",
        "outputId": "981ff22e-1e92-4716-b6f5-341083a933e9"
      },
      "outputs": [
        {
          "name": "stdout",
          "output_type": "stream",
          "text": [
            "train_data =  <torch.utils.data.dataloader.DataLoader object at 0x7efdd878c950>\n",
            "epoch =  0\n",
            "val acc:  0.23333333333333334\n",
            "epoch =  1\n",
            "val acc:  0.18\n",
            "epoch =  2\n",
            "patience =  1\n",
            "val acc:  0.31333333333333335\n",
            "epoch =  3\n",
            "val acc:  0.2733333333333333\n",
            "epoch =  4\n",
            "patience =  2\n",
            "val acc:  0.36\n",
            "epoch =  5\n",
            "val acc:  0.3466666666666667\n",
            "epoch =  6\n",
            "patience =  3\n",
            "val acc:  0.36\n",
            "epoch =  7\n",
            "val acc:  0.3933333333333333\n",
            "epoch =  8\n",
            "val acc:  0.38666666666666666\n",
            "epoch =  9\n",
            "patience =  4\n",
            "val acc:  0.38\n",
            "epoch =  10\n",
            "patience =  5\n",
            "val acc:  0.37333333333333335\n",
            "epoch =  11\n",
            "patience =  6\n",
            "val acc:  0.46\n",
            "epoch =  12\n",
            "val acc:  0.44666666666666666\n",
            "epoch =  13\n",
            "patience =  7\n",
            "val acc:  0.4666666666666667\n",
            "epoch =  14\n",
            "val acc:  0.52\n",
            "epoch =  15\n",
            "val acc:  0.5666666666666667\n",
            "epoch =  16\n",
            "val acc:  0.5466666666666666\n",
            "epoch =  17\n",
            "patience =  8\n",
            "val acc:  0.5333333333333333\n",
            "epoch =  18\n",
            "patience =  9\n",
            "Error: Validation Accuracy Run-Off\n",
            "Total time elapsed: 1322.87 seconds\n",
            "Final Training Loss:  0.016259804368019104\n",
            "Final Training Accuracy:  0.7411764705882353\n",
            "Final Val Accuracy:  0.5333333333333333\n"
          ]
        },
        {
          "data": {
            "image/png": "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\n",
            "text/plain": [
              "<Figure size 432x288 with 1 Axes>"
            ]
          },
          "metadata": {},
          "output_type": "display_data"
        },
        {
          "data": {
            "image/png": "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\n",
            "text/plain": [
              "<Figure size 432x288 with 1 Axes>"
            ]
          },
          "metadata": {},
          "output_type": "display_data"
        }
      ],
      "source": [
        "model6 = CNN()\n",
        "\n",
        "try6 = train(model6, epochs = 20, batch_size = 64, lr = 0.001, optimizer =\"Adam\", weight_decay = 0.001)"
      ]
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "g7sz4ah_SIEI"
      },
      "source": [
        "Qualitative Testing"
      ]
    },
    {
      "cell_type": "code",
      "execution_count": null,
      "metadata": {
        "id": "snTV5IDIfA3Q"
      },
      "outputs": [],
      "source": [
        "def qual_test(model, batch_size = 16):\n",
        "  mel_path = path + \"/Qual_testing\"\n",
        "  transform = transforms.Compose([transforms.ToTensor(),\n",
        "                                transforms.Normalize((0.5, 0.5, 0.5, 0.5), (0.5, 0.5, 0.5, 0.5))])  #convert data into a tensor, and then normalize it\n",
        "  torch.manual_seed(1000)\n",
        "  genres = [\"blues\",\"classical\",\"country\",\"disco\",\"hiphop\",\"jazz\",\"metal\",\"pop\",\"reggae\",\"rock\"]\n",
        "  accuracies = []\n",
        "  for genre in genres:\n",
        "    print(\"genre = \",genre)\n",
        "    v_set = torchvision.datasets.ImageFolder(mel_path + '/' + genre, loader = plt.imread, transform = transform)\n",
        "    v_set_loader = torch.utils.data.DataLoader(v_set, batch_size=batch_size, num_workers=1, shuffle=True)\n",
        "    accuracy = get_accuracy(model,v_set_loader)\n",
        "    accuracies.append(accuracy)\n",
        "    print(\"accuracy = \", accuracy)\n",
        "\n",
        "  plt.title(\"Validation Accuracy By Genre\")\n",
        "  plt.scatter(genres, accuracies)\n",
        "  plt.xlabel(\"Genres\")\n",
        "  plt.ylabel(\"Validation Accuracy\")\n",
        "  plt.show()"
      ]
    },
    {
      "cell_type": "code",
      "execution_count": null,
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 635
        },
        "id": "1nUbxU9516X6",
        "outputId": "d7d3a954-ac90-4aef-f9d5-cedfab1a65d5"
      },
      "outputs": [
        {
          "name": "stdout",
          "output_type": "stream",
          "text": [
            "genre =  blues\n",
            "accuracy =  0.4666666666666667\n",
            "genre =  classical\n",
            "accuracy =  0.0\n",
            "genre =  country\n",
            "accuracy =  0.0\n",
            "genre =  disco\n",
            "accuracy =  0.0\n",
            "genre =  hiphop\n",
            "accuracy =  0.06666666666666667\n",
            "genre =  jazz\n",
            "accuracy =  0.13333333333333333\n",
            "genre =  metal\n",
            "accuracy =  0.0\n",
            "genre =  pop\n",
            "accuracy =  0.0\n",
            "genre =  reggae\n",
            "accuracy =  0.0\n",
            "genre =  rock\n",
            "accuracy =  0.0\n"
          ]
        },
        {
          "data": {
            "image/png": "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\n",
            "text/plain": [
              "<Figure size 432x288 with 1 Axes>"
            ]
          },
          "metadata": {},
          "output_type": "display_data"
        }
      ],
      "source": [
        "batch_size = 16\n",
        "mel_path = path + \"/Qual_testing\"\n",
        "transform = transforms.Compose([transforms.ToTensor(),\n",
        "                                transforms.Normalize((0.5, 0.5, 0.5, 0.5), (0.5, 0.5, 0.5, 0.5))])  #convert data into a tensor, and then normalize it\n",
        "torch.manual_seed(1000)\n",
        "genres = [\"blues\",\"classical\",\"country\",\"disco\",\"hiphop\",\"jazz\",\"metal\",\"pop\",\"reggae\",\"rock\"]\n",
        "sets = []\n",
        "accuracies = []\n",
        "\n",
        "for genre in genres:\n",
        "  print(\"genre = \",genre)\n",
        "  v_set = torchvision.datasets.ImageFolder(mel_path + '/' + genre, loader = plt.imread, transform = transform)\n",
        "  v_set_loader = torch.utils.data.DataLoader(v_set, batch_size=batch_size, num_workers=1, shuffle=True)\n",
        "  accuracy = get_accuracy(model,v_set_loader)\n",
        "  accuracies.append(accuracy)\n",
        "  print(\"accuracy = \", accuracy)\n",
        "\n",
        "\n",
        "plt.title(\"Validation Accuracy By Genre\")\n",
        "plt.scatter(genres, accuracies)\n",
        "plt.xlabel(\"Genres\")\n",
        "plt.ylabel(\"Validation Accuracy\")\n",
        "plt.show()"
      ]
    },
    {
      "cell_type": "code",
      "execution_count": null,
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 635
        },
        "id": "H0OCU2L5fogu",
        "outputId": "a39bad5b-19ad-4b87-8d8a-7f9b76575d04"
      },
      "outputs": [
        {
          "name": "stdout",
          "output_type": "stream",
          "text": [
            "genre =  blues\n",
            "accuracy =  0.4666666666666667\n",
            "genre =  classical\n",
            "accuracy =  0.0\n",
            "genre =  country\n",
            "accuracy =  0.0\n",
            "genre =  disco\n",
            "accuracy =  0.0\n",
            "genre =  hiphop\n",
            "accuracy =  0.06666666666666667\n",
            "genre =  jazz\n",
            "accuracy =  0.13333333333333333\n",
            "genre =  metal\n",
            "accuracy =  0.0\n",
            "genre =  pop\n",
            "accuracy =  0.0\n",
            "genre =  reggae\n",
            "accuracy =  0.0\n",
            "genre =  rock\n",
            "accuracy =  0.0\n"
          ]
        },
        {
          "data": {
            "image/png": "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\n",
            "text/plain": [
              "<Figure size 432x288 with 1 Axes>"
            ]
          },
          "metadata": {},
          "output_type": "display_data"
        }
      ],
      "source": [
        "qual_test(try6)"
      ]
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "QIGhsAxFF2Qz"
      },
      "source": [
        "# Transformer"
      ]
    },
    {
      "cell_type": "code",
      "execution_count": null,
      "metadata": {
        "id": "MrzzKPwOfwRn"
      },
      "outputs": [],
      "source": [
        "def patchify(images, n_x_patches, n_y_patches):\n",
        "  n, c, w, h = images.shape\n",
        "\n",
        "  ps_x = w//n_x_patches   #patch size in x\n",
        "  ps_y = h//n_y_patches   #patch size in y\n",
        "\n",
        "  patches = torch.zeros(n, n_x_patches*n_y_patches, c*ps_x*ps_y)\n",
        "\n",
        "\n",
        "\n",
        "  idx = 0\n",
        "  for image in images:\n",
        "\n",
        "    for i in range(n_x_patches):\n",
        "      for j in range(n_y_patches):\n",
        "        patch = image[:, i*ps_x:(i+1)*ps_x, j*ps_y:(j+1)*ps_y]\n",
        "\n",
        "        patches[idx,i * n_y_patches + j] = patch.flatten()\n",
        "\n",
        "    idx+=1\n",
        "\n",
        "  return patches   #3d Tensor of N images containing n_x x n_y patches next to each other. Each patch has c x patch_size_x x patch_size_y dimensions.\n"
      ]
    },
    {
      "cell_type": "code",
      "execution_count": null,
      "metadata": {
        "id": "i1f7K_c9i8rF"
      },
      "outputs": [],
      "source": [
        "def get_positional_embeddings(seq_len,d):    #always rectangular\n",
        "  embeddings = torch.zeros(seq_len,d)\n",
        "\n",
        "  for i in range(seq_len):\n",
        "    for j in range(d):\n",
        "      if j%2 == 0:\n",
        "        embeddings[i][j] = np.sin(i/(10000**(j/d)))\n",
        "\n",
        "      else:\n",
        "        embeddings[i][j] = np.cos(i/10000**((j-1)/d))\n",
        "\n",
        "  return embeddings"
      ]
    },
    {
      "cell_type": "code",
      "execution_count": null,
      "metadata": {
        "id": "dgr9x1n5jDRE"
      },
      "outputs": [],
      "source": [
        "class Encoder(nn.Module):\n",
        "  torch.manual_seed(620)\n",
        "  def __init__(self, input_size = 100, hidden_size = 100, num_heads = 4):   #Num_heads = num_patches\n",
        "    super(Encoder,self).__init__()\n",
        "    self.linear_q = nn.Linear(input_size, hidden_size)\n",
        "    self.linear_k = nn.Linear(input_size, hidden_size)\n",
        "    self.linear_v = nn.Linear(input_size, hidden_size)\n",
        "    self.linear_x = nn.Linear(input_size, hidden_size)\n",
        "    self.attention = nn.MultiheadAttention(hidden_size, num_heads, batch_first = True)   #Batch_first ensures the same dimensions as the patchified tensor.\n",
        "\n",
        "    self.ff = nn.Sequential(nn.Linear(hidden_size, hidden_size),nn.ReLU(),nn.Linear(hidden_size,hidden_size)) #feed forward neural network in encoder\n",
        "\n",
        "    self.norm = nn.LayerNorm(hidden_size)\n",
        "\n",
        "\n",
        "  def forward(self, x): #x is image vector in this case?\n",
        "    q, k, v = self.linear_q(x),self.linear_k(x),self.linear_v(x)  #get q, k and v layers for x\n",
        "    x = self.norm(self.linear_x(x) + self.attention(q,k,v)[0]) #residual connection. MhA gives tuple: (attn_output, attn_output_weights)\n",
        "    x = self.norm(x + self.ff(x))\n",
        "\n",
        "    return x\n"
      ]
    },
    {
      "cell_type": "code",
      "execution_count": null,
      "metadata": {
        "id": "nZ5D7PKAlH_U"
      },
      "outputs": [],
      "source": [
        "##When you call this class, please change:\n",
        "#1) c,w,h to reflect the right image dimensions\n",
        "#2) n_x_patches and n_y_patches to ensure w//n_x_patches == h//n_y_patches\n",
        "#3) num_classes to reflect the appropriate # of genres for the datatset\n",
        "\n",
        "%matplotlib inline\n",
        "\n",
        "class Transformer(nn.Module):\n",
        "  torch.manual_seed(620)\n",
        "  def __init__(self, num_class = 10, num_blocks = 2, n_x_patches = 24,\n",
        "               n_y_patches = 36, c = 4, w = 288, h = 432, d = 64, dropout = 0.25):   #num_class = genre_count; d = hidden size\n",
        "\n",
        "    super(Transformer, self).__init__()\n",
        "\n",
        "    #define parameters for forward method\n",
        "    self.n_x_patches = n_x_patches\n",
        "    self.n_y_patches = n_y_patches\n",
        "    self.num_blocks = num_blocks\n",
        "\n",
        "    #linear projection of each token to a flattened vectors\n",
        "    self.linear_x = nn.Linear(c*w//n_x_patches*h//n_y_patches, d)\n",
        "\n",
        "    #create classification token\n",
        "    self.class_token = nn.Parameter(torch.rand(1, d))\n",
        "\n",
        "    #get positional embeddings spanning all token positions\n",
        "    self.pos = get_positional_embeddings(n_x_patches*n_y_patches + 1,d)   #+1 for classification token\n",
        "\n",
        "    #encoder\n",
        "    self.encoder = Encoder(d, d)\n",
        "\n",
        "    #dropout layer\n",
        "    self.dropout = nn.Dropout(dropout)\n",
        "\n",
        "    #classficiation layer\n",
        "    self.fc = nn.Linear(d, num_class)\n",
        "\n",
        "\n",
        "\n",
        "  def forward(self, x):  #x is batch of N images\n",
        "    n,c,w,h = x.shape\n",
        "\n",
        "    #construct patches\n",
        "    patches = patchify(x, self.n_x_patches, self.n_y_patches)  #N x (n_x_patches x n_y_patches)\n",
        "\n",
        "    #Obtain linear projections for all tokens of N images\n",
        "    toks = self.linear_x(patches)\n",
        "\n",
        "    #Add classification token\n",
        "    tokens = torch.stack([torch.vstack((self.class_token, toks[i])) for i in range(len(toks))]) #Make classification token come first in all N images\n",
        "\n",
        "    #Expand positional embeddings to match N images\n",
        "    pos = self.pos.repeat(n,1,1)\n",
        "    output = tokens + pos\n",
        "\n",
        "    for i in range(self.num_blocks):\n",
        "      output = self.encoder(output)\n",
        "\n",
        "    #Get classification token only\n",
        "    final_out = output[:,0]\n",
        "\n",
        "    #Dropout layer\n",
        "    final_out = self.dropout(final_out)\n",
        "\n",
        "    #pass to classification FC NN\n",
        "    return self.fc(final_out)"
      ]
    },
    {
      "cell_type": "code",
      "execution_count": null,
      "metadata": {
        "id": "q4G1L3S7jHze"
      },
      "outputs": [],
      "source": [
        "##Please make sure get_data function redirects to the right dataset that you want to use\n",
        "\n",
        "def ttrain(model, lr = 0.001, epochs = 20, criterion = \"CE\", optimizer = \"Adam\", mom=0.5, batch_size = 32, weight_decay = 0.001):\n",
        "\n",
        "  '''inputs: nn model, training and validation data (torch.utils.data.DataLoader class), learning rate lr, no. of epochs, loss function criterion\n",
        "  outputs: trained model, accuracy plots'''\n",
        "\n",
        "  train_data, val_data = get_data(batch_size) #Please make sure get_data function redirects to the right dataset that you want to use\n",
        "  loss_dict = {\"CE\": nn.CrossEntropyLoss()}\n",
        "  loss_f = loss_dict[criterion]\n",
        "\n",
        "  optim_dict = {\"SGD\": torch.optim.SGD(model.parameters(), lr, momentum=mom, weight_decay = weight_decay),\"Adam\":torch.optim.Adam(model.parameters(), lr = lr, weight_decay = weight_decay)}\n",
        "  opt = optim_dict[optimizer]\n",
        "\n",
        "  eps = []\n",
        "  losses = []\n",
        "  train_accuracy = []\n",
        "  val_accuracy = []\n",
        "  patience = 0\n",
        "\n",
        "  #train model\n",
        "  start_time = time.time()\n",
        "  for epoch in range(epochs):\n",
        "    print(\"epoch = \", epoch)\n",
        "    iteration = 0\n",
        "\n",
        "    #early stop with patience\n",
        "    if epoch > 1:\n",
        "      if val_accuracy[epoch - 2] > val_accuracy[epoch - 1]:\n",
        "        patience+=1\n",
        "        print(\"patience = \", patience)\n",
        "      if patience > 12:\n",
        "        print(\"Error: Validation Accuracy Run-Off\")\n",
        "        break\n",
        "      #if epoch > 13 and val_accuracy[epoch - 1]<0.5:\n",
        "        #print(\"Error: Low Validation Accuracy\")\n",
        "        #break\n",
        "\n",
        "      if val_accuracy[epoch - 1] > 0.91:\n",
        "        break\n",
        "\n",
        "\n",
        "    for (images, labels) in train_data:\n",
        "      #print(\"batch_no. = \", iteration)\n",
        "      #iteration+=1\n",
        "      #print(\"shape =\", images.shape)\n",
        "\n",
        "    #make passes\n",
        "      outputs = model(images)    #make prediction; softmax applied by criterion\n",
        "      loss = loss_f(outputs, labels)\n",
        "      loss.backward()    #calculate gradient\n",
        "      opt.step()   #update model parameters\n",
        "      opt.zero_grad()   #clean up\n",
        "\n",
        "    #record information for epoch\n",
        "    eps.append(epoch)\n",
        "    losses.append(float(loss)/batch_size)\n",
        "    train_accuracy.append(get_accuracy(model,train_data))\n",
        "    val_acc = get_accuracy(model,val_data)\n",
        "    print(\"val acc: \", val_acc)\n",
        "    val_accuracy.append(val_acc)\n",
        "\n",
        "  end_time = time.time()\n",
        "  elapsed_time = end_time - start_time\n",
        "  print(\"Total time elapsed: {:.2f} seconds\".format(elapsed_time))\n",
        "  print(\"Final Training Loss: \", losses[-1])\n",
        "  print(\"Final Training Accuracy: \", train_accuracy[-1])\n",
        "  print(\"Final Val Accuracy: \", val_accuracy[-1])\n",
        "\n",
        "  #plotting\n",
        "  plt.title(\"Training Loss vs. Epochs\")\n",
        "  plt.plot(eps, losses, label=\"Training\")\n",
        "  plt.xlabel(\"Epochs\")\n",
        "  plt.ylabel(\"Loss\")\n",
        "  plt.show()\n",
        "\n",
        "  plt.title(\"Accuracy vs. Epochs\")\n",
        "  plt.plot(eps, train_accuracy, label=\"Train\")\n",
        "  plt.plot(eps, val_accuracy, label=\"Validation\")\n",
        "  plt.xlabel(\"Epochs\")\n",
        "  plt.ylabel(\"Accuracy\")\n",
        "  plt.legend(loc='best')\n",
        "  plt.show()\n",
        "\n",
        "  return model"
      ]
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "obU-c-IlbgWH"
      },
      "source": [
        "Transformer Training"
      ]
    },
    {
      "cell_type": "code",
      "execution_count": null,
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/"
        },
        "id": "D_FrpPGPjNYd",
        "outputId": "19e7ff3e-0268-42f4-acd2-c4ab3c2f7ef1"
      },
      "outputs": [
        {
          "name": "stdout",
          "output_type": "stream",
          "text": [
            "epoch =  0\n"
          ]
        }
      ],
      "source": [
        "tmodel1 = Transformer(num_class = 8, num_blocks = 3, n_x_patches = 24,\n",
        "               n_y_patches = 32, c = 4, w = 480, h = 640, d = 64, dropout = 0)\n",
        "ttry_1 = ttrain(tmodel1, weight_decay = 0.001)"
      ]
    },
    {
      "cell_type": "code",
      "execution_count": null,
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/"
        },
        "id": "gaxzmaPMkUkb",
        "outputId": "ae203c74-ec32-4627-d8ff-9701082afa18"
      },
      "outputs": [
        {
          "name": "stdout",
          "output_type": "stream",
          "text": [
            "Using device:  cuda (Tesla T4)\n"
          ]
        }
      ],
      "source": [
        "device = torch.device(\"cuda\" if torch.cuda.is_available() else \"cpu\")\n",
        "print(\"Using device: \", device,f\"({torch.cuda.get_device_name(device)})\" if torch.cuda.is_available() else \"\")"
      ]
    },
    {
      "cell_type": "code",
      "execution_count": null,
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 1000
        },
        "id": "Xc4yTmRvkqXV",
        "outputId": "da65432a-f1d5-4e0c-fc58-fb018fc72201"
      },
      "outputs": [
        {
          "name": "stdout",
          "output_type": "stream",
          "text": [
            "epoch =  0\n",
            "val acc:  0.1\n",
            "epoch =  1\n",
            "val acc:  0.18666666666666668\n",
            "epoch =  2\n",
            "val acc:  0.16666666666666666\n",
            "epoch =  3\n",
            "patience =  1\n",
            "val acc:  0.16\n",
            "epoch =  4\n",
            "patience =  2\n",
            "val acc:  0.18\n",
            "epoch =  5\n",
            "val acc:  0.20666666666666667\n",
            "epoch =  6\n",
            "val acc:  0.26\n",
            "epoch =  7\n",
            "val acc:  0.22666666666666666\n",
            "epoch =  8\n",
            "patience =  3\n",
            "val acc:  0.24\n",
            "epoch =  9\n",
            "val acc:  0.18666666666666668\n",
            "epoch =  10\n",
            "patience =  4\n",
            "val acc:  0.24\n",
            "epoch =  11\n",
            "val acc:  0.18666666666666668\n",
            "epoch =  12\n",
            "patience =  5\n",
            "val acc:  0.25333333333333335\n",
            "epoch =  13\n",
            "val acc:  0.22666666666666666\n",
            "epoch =  14\n",
            "patience =  6\n",
            "Error: Low Validation Accuracy\n",
            "Total time elapsed: 5388.10 seconds\n",
            "Final Training Loss:  0.05907721817493439\n",
            "Final Training Accuracy:  0.2235294117647059\n",
            "Final Val Accuracy:  0.22666666666666666\n"
          ]
        },
        {
          "data": {
            "image/png": "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\n",
            "text/plain": [
              "<Figure size 432x288 with 1 Axes>"
            ]
          },
          "metadata": {},
          "output_type": "display_data"
        },
        {
          "data": {
            "image/png": "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\n",
            "text/plain": [
              "<Figure size 432x288 with 1 Axes>"
            ]
          },
          "metadata": {},
          "output_type": "display_data"
        }
      ],
      "source": [
        "tmodel2 = Transformer(num_class = 10, num_blocks = 6, c = 4, dropout = 0.25)\n",
        "ttry_2 = ttrain(tmodel2, weight_decay = 0.001)"
      ]
    },
    {
      "cell_type": "code",
      "source": [
        "tmodel2 = Transformer(num_class = 10, num_blocks = 3, dropout = 0.2)\n",
        "ttry_2 = ttrain(tmodel2, weight_decay = 0.001)"
      ],
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 1000
        },
        "id": "981xk4b6duqG",
        "outputId": "a3802a59-c50f-4e24-a867-548405fe1672"
      },
      "execution_count": null,
      "outputs": [
        {
          "output_type": "stream",
          "name": "stdout",
          "text": [
            "epoch =  0\n",
            "val acc:  0.18\n",
            "epoch =  1\n",
            "val acc:  0.18666666666666668\n",
            "epoch =  2\n",
            "val acc:  0.17333333333333334\n",
            "epoch =  3\n",
            "patience =  1\n",
            "val acc:  0.26666666666666666\n",
            "epoch =  4\n",
            "val acc:  0.20666666666666667\n",
            "epoch =  5\n",
            "patience =  2\n",
            "val acc:  0.19333333333333333\n",
            "epoch =  6\n",
            "patience =  3\n",
            "val acc:  0.2\n",
            "epoch =  7\n",
            "val acc:  0.23333333333333334\n",
            "epoch =  8\n",
            "val acc:  0.22\n",
            "epoch =  9\n",
            "patience =  4\n",
            "val acc:  0.29333333333333333\n",
            "epoch =  10\n",
            "val acc:  0.18666666666666668\n",
            "epoch =  11\n",
            "patience =  5\n",
            "val acc:  0.2733333333333333\n",
            "epoch =  12\n",
            "val acc:  0.20666666666666667\n",
            "epoch =  13\n",
            "patience =  6\n",
            "val acc:  0.26666666666666666\n",
            "epoch =  14\n",
            "val acc:  0.21333333333333335\n",
            "epoch =  15\n",
            "patience =  7\n",
            "val acc:  0.3333333333333333\n",
            "epoch =  16\n",
            "val acc:  0.2733333333333333\n",
            "epoch =  17\n",
            "patience =  8\n",
            "val acc:  0.24\n",
            "epoch =  18\n",
            "patience =  9\n",
            "val acc:  0.29333333333333333\n",
            "epoch =  19\n",
            "val acc:  0.23333333333333334\n",
            "Total time elapsed: 4250.14 seconds\n",
            "Final Training Loss:  0.057790789753198624\n",
            "Final Training Accuracy:  0.2564705882352941\n",
            "Final Val Accuracy:  0.23333333333333334\n"
          ]
        },
        {
          "output_type": "display_data",
          "data": {
            "text/plain": [
              "<Figure size 432x288 with 1 Axes>"
            ],
            "image/png": "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\n"
          },
          "metadata": {
            "needs_background": "light"
          }
        },
        {
          "output_type": "display_data",
          "data": {
            "text/plain": [
              "<Figure size 432x288 with 1 Axes>"
            ],
            "image/png": "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\n"
          },
          "metadata": {
            "needs_background": "light"
          }
        }
      ]
    },
    {
      "cell_type": "code",
      "execution_count": null,
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 1000
        },
        "id": "MdyCHVouVFXe",
        "outputId": "aa1d3bad-f804-4581-ff92-d7e09c1d2d37"
      },
      "outputs": [
        {
          "output_type": "stream",
          "name": "stdout",
          "text": [
            "epoch =  0\n",
            "val acc:  0.14\n",
            "epoch =  1\n",
            "val acc:  0.19333333333333333\n",
            "epoch =  2\n",
            "val acc:  0.2\n",
            "epoch =  3\n",
            "val acc:  0.21333333333333335\n",
            "epoch =  4\n",
            "val acc:  0.17333333333333334\n",
            "epoch =  5\n",
            "patience =  1\n",
            "val acc:  0.24\n",
            "epoch =  6\n",
            "val acc:  0.19333333333333333\n",
            "epoch =  7\n",
            "patience =  2\n",
            "val acc:  0.28\n",
            "epoch =  8\n",
            "val acc:  0.22666666666666666\n",
            "epoch =  9\n",
            "patience =  3\n",
            "val acc:  0.22\n",
            "epoch =  10\n",
            "patience =  4\n",
            "val acc:  0.24666666666666667\n",
            "epoch =  11\n",
            "val acc:  0.22\n",
            "epoch =  12\n",
            "patience =  5\n",
            "val acc:  0.2733333333333333\n",
            "epoch =  13\n",
            "val acc:  0.20666666666666667\n",
            "epoch =  14\n",
            "patience =  6\n",
            "Error: Low Validation Accuracy\n",
            "Total time elapsed: 3550.39 seconds\n",
            "Final Training Loss:  0.06233758479356766\n",
            "Final Training Accuracy:  0.22941176470588234\n",
            "Final Val Accuracy:  0.20666666666666667\n"
          ]
        },
        {
          "output_type": "display_data",
          "data": {
            "text/plain": [
              "<Figure size 432x288 with 1 Axes>"
            ],
            "image/png": "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\n"
          },
          "metadata": {
            "needs_background": "light"
          }
        },
        {
          "output_type": "display_data",
          "data": {
            "text/plain": [
              "<Figure size 432x288 with 1 Axes>"
            ],
            "image/png": "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\n"
          },
          "metadata": {
            "needs_background": "light"
          }
        }
      ],
      "source": [
        "tmodel3 = Transformer(num_class = 10, num_blocks = 3, c = 4, dropout = 0)\n",
        "ttry_3 = ttrain(tmodel3, lr = 0.0005, weight_decay = 0.001)"
      ]
    },
    {
      "cell_type": "code",
      "execution_count": null,
      "metadata": {
        "id": "b295YBVPWo3D"
      },
      "outputs": [],
      "source": []
    }
  ],
  "metadata": {
    "accelerator": "GPU",
    "colab": {
      "collapsed_sections": [
        "F0AgU90vQLiJ"
      ],
      "provenance": []
    },
    "gpuClass": "standard",
    "kernelspec": {
      "display_name": "Python 3",
      "name": "python3"
    },
    "language_info": {
      "name": "python"
    }
  },
  "nbformat": 4,
  "nbformat_minor": 0
}