cl_train_linear.ipynb

{
 "metadata": {
  "name": "ChromaLuma - training"
 },
 "nbformat": 3,
 "nbformat_minor": 0,
 "worksheets": [
  {
   "cells": [
    {
     "cell_type": "code",
     "collapsed": false,
     "input": [
      "import sklearn\n",
      "import sklearn.cross_validation\n",
      "import sklearn.linear_model\n",
      "import sklearn.metrics\n",
      "import cPickle as pickle\n",
      "import numpy as np\n",
      "import glob\n",
      "import os"
     ],
     "language": "python",
     "metadata": {},
     "outputs": [
      {
       "output_type": "stream",
       "stream": "stderr",
       "text": [
        "/usr/local/lib/python2.7/dist-packages/scikits/__init__.py:1: UserWarning: Module dap was already imported from None, but /usr/lib/python2.7/dist-packages is being added to sys.path\n",
        "  __import__('pkg_resources').declare_namespace(__name__)\n"
       ]
      }
     ],
     "prompt_number": 3
    },
    {
     "cell_type": "code",
     "collapsed": false,
     "input": [
      "# Set the RNG for reproducibility\n",
      "SEED = hash('yodawg')"
     ],
     "language": "python",
     "metadata": {},
     "outputs": [],
     "prompt_number": 4
    },
    {
     "cell_type": "heading",
     "level": 1,
     "metadata": {},
     "source": [
      "Training - beat-level independent classifier"
     ]
    },
    {
     "cell_type": "code",
     "collapsed": false,
     "input": [
      "def load_songs(path, features='raw-compressed', labels='labels-minmaj'):\n",
      "    '''Load songs and labels:\n",
      "\n",
      "    Arguments\n",
      "    ---------\n",
      "    path: (str) glob fragment giving the path to npy files.\n",
      "            eg. '/q/porkpie/porkpie-p0/drspeech/data/chromaluma/beatles/*'\n",
      "            or  '/q/porkpie/porkpie-p0/drspeech/data/chromaluma/uspop2002/'\n",
      "\n",
      "    features: (str) extension to specify which features to load\n",
      "            eg. 'raw-compressed' or 'cdl-compressed'\n",
      "\n",
      "    labels: (str) extension to specify which labels to load\n",
      "            eg. 'labels-minmaj' or 'beat'\n",
      "\n",
      "    Returns\n",
      "    -------\n",
      "    (song_id, names, X, Y)\n",
      "        \n",
      "        song_id is a list of the constituent songs\n",
      "        names is a list of the song filenames\n",
      "        X is a list of np.arrays containing features\n",
      "        Y is a list of np.arrays containing labels\n",
      "            \n",
      "    '''\n",
      "    \n",
      "    feature_files = glob.glob('%s/*-%s.npy' % (path, features))\n",
      "    feature_files.sort()\n",
      "    \n",
      "    label_files   = glob.glob('%s/*-%s.npy' % (path, labels))\n",
      "    label_files.sort()\n",
      "    \n",
      "    assert(len(feature_files) == len(label_files))\n",
      "        \n",
      "    X       = []\n",
      "    Y       = []\n",
      "    song_id = []\n",
      "    names   = []\n",
      "    \n",
      "\n",
      "    for (i, (X_f, Y_f)) in enumerate(zip(feature_files, label_files)):\n",
      "        X_new, Y_new = np.load(X_f), np.load(Y_f)\n",
      "        \n",
      "        X.append(X_new)\n",
      "        Y.append(Y_new)\n",
      "            \n",
      "        song_id.append(i)\n",
      "        names.append(X_f)\n",
      "          \n",
      "    return song_id, names, X, Y"
     ],
     "language": "python",
     "metadata": {},
     "outputs": [],
     "prompt_number": 5
    },
    {
     "cell_type": "code",
     "collapsed": false,
     "input": [
      "def train(X, Y, n_iterations=5):\n",
      "    '''Train a multi-class logistic regression model\n",
      "    \n",
      "    X and Y are squashed into one giant data set\n",
      "\n",
      "    A classifier is trained by k-fold cross-validation, and then\n",
      "    retrained on the full data set with the best validation parameters.\n",
      "\n",
      "    Returns:\n",
      "    h: sklearn.linear_model.SGDClassifier\n",
      "    '''\n",
      "    \n",
      "    # Merge the data\n",
      "    # Stack all beats into one matrix\n",
      "    X = np.vstack(X)\n",
      "    \n",
      "    # Stack all labels into one vector\n",
      "    Y = np.concatenate(Y).astype(int)\n",
      "    \n",
      "    parameters_sgd = {'alpha': [1e-7, 1e-6, 1e-5, 1e-4, 1e-3, 1e-2, 1e-1],\n",
      "                      'class_weight': ['auto'], 'n_iter': [5], 'shuffle': [True]}\n",
      "    \n",
      "    splits = sklearn.cross_validation.KFold(len(Y), k=n_iterations, shuffle=True, random_state=SEED)\n",
      "    \n",
      "    h = sklearn.grid_search.GridSearchCV(\n",
      "                sklearn.linear_model.SGDClassifier(loss='log'),\n",
      "                parameters_sgd, n_jobs=4, cv=splits)\n",
      "    \n",
      "    h.fit(X, Y)\n",
      "    \n",
      "    return h.best_estimator_"
     ],
     "language": "python",
     "metadata": {},
     "outputs": [],
     "prompt_number": 6
    },
    {
     "cell_type": "code",
     "collapsed": false,
     "input": [
      "def analyze_labels(Y):\n",
      "    Y = np.concatenate(Y).astype(int)\n",
      "    print Y.min(), Y.max()\n",
      "    counts = np.zeros(Y.max()+1, dtype=int)\n",
      "    \n",
      "    for lab in Y:\n",
      "        counts[lab] = counts[lab] + 1\n",
      "        \n",
      "    figure()\n",
      "    bar(range(len(counts)), counts)\n",
      "    \n",
      "    print counts.min(), counts.max()"
     ],
     "language": "python",
     "metadata": {},
     "outputs": [],
     "prompt_number": 7
    },
    {
     "cell_type": "code",
     "collapsed": false,
     "input": [
      "with open('/home/bmcfee/git/death2chroma/Training_Scripts/minmaj_dict.pickle', 'r') as f:\n",
      "    chord_classes, chord_indices = pickle.load(f)"
     ],
     "language": "python",
     "metadata": {},
     "outputs": [],
     "prompt_number": 8
    },
    {
     "cell_type": "code",
     "collapsed": false,
     "input": [
      "#song_id, names, X, Y = load_songs('/home/bmcfee/git/death2chroma/data/beatles/*', features='wrapCL', labels='labels-minmaj')\n",
      "#song_id, names, X, Y = load_songs('/home/bmcfee/git/death2chroma/data/beatles/*', features='CL', labels='labels-minmaj')\n",
      "song_id, names, X, Y = load_songs('/home/bmcfee/git/death2chroma/data/beatles/*', features='raw-compressed', labels='labels-minmaj')\n",
      "#song_id, names, X, Y = load_songs('/home/bmcfee/git/death2chroma/data/beatles/*', features='encoded-compressed', labels='labels-minmaj')"
     ],
     "language": "python",
     "metadata": {},
     "outputs": [],
     "prompt_number": 19
    },
    {
     "cell_type": "code",
     "collapsed": false,
     "input": [
      "frame_classifier = train(X, Y)"
     ],
     "language": "python",
     "metadata": {},
     "outputs": [],
     "prompt_number": 289
    },
    {
     "cell_type": "code",
     "collapsed": false,
     "input": [
      "#with open('/home/bmcfee/git/death2chroma/data/frame_classifier_encoded_beatles.pickle', 'w') as f:\n",
      "with open('/home/bmcfee/git/death2chroma/data/frame_classifier_beatles.pickle', 'w') as f:\n",
      "#with open('/home/bmcfee/git/death2chroma/data/frame_classifier_beatles_CL.pickle', 'w') as f:\n",
      "#with open('/home/bmcfee/git/death2chroma/data/frame_classifier_beatles_wrapCL.pickle', 'w') as f:\n",
      "    pickle.dump(frame_classifier, f, protocol=-1)"
     ],
     "language": "python",
     "metadata": {},
     "outputs": [],
     "prompt_number": 290
    },
    {
     "cell_type": "code",
     "collapsed": false,
     "input": [
      "NOTES     = ['C', 'D', 'E', 'F', 'G', 'A', 'B']\n",
      "positions = np.array([0,     8,   16,   20,   28,  36,  44])\n",
      "#for i in range(4):\n",
      "#    if i == 0:\n",
      "#        P = positions\n",
      "#    else:\n",
      "#        P = np.hstack((P, positions + i * 48))\n",
      "\n",
      "imshow(np.abs(frame_classifier.coef_), aspect='auto', origin='lower', interpolation='nearest'), colorbar()\n",
      "#xticks(P, NOTES * 4)\n",
      "xticks(positions, NOTES)\n",
      "yticks(range(25), [chord_classes[chord_indices[i]][0] for i in range(25)])\n",
      "pass"
     ],
     "language": "python",
     "metadata": {},
     "outputs": [
      {
       "output_type": "display_data",
       "png": "iVBORw0KGgoAAAANSUhEUgAAAdsAAAFrCAYAAACZsCNBAAAABHNCSVQICAgIfAhkiAAAAAlwSFlz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\nAAAAAElFTkSuQmCC\n"
      }
     ],
     "prompt_number": 26
    },
    {
     "cell_type": "heading",
     "level": 1,
     "metadata": {},
     "source": [
      "Prediction"
     ]
    },
    {
     "cell_type": "code",
     "collapsed": false,
     "input": [
      "def output_predicted_labels(path, names, beats, Ypred, ext='prediction'):\n",
      "    \n",
      "    for (song_i, beats_i, Ypred_i) in zip(names, beats, Ypred):\n",
      "        song_i = song_i[:song_i.index('.npy')]\n",
      "        outfile = '%s/%s-%s.npy' % (path, os.path.basename(song_i), ext)\n",
      "        np.save(outfile, zip(beats_i, Ypred_i))\n",
      "        pass\n",
      "    pass"
     ],
     "language": "python",
     "metadata": {},
     "outputs": [],
     "prompt_number": 312
    },
    {
     "cell_type": "code",
     "collapsed": false,
     "input": [
      "def frame_eval(Y_true, Y_pred):\n",
      "    Y_true = np.concatenate(Y_true)\n",
      "    Y_pred = np.concatenate(Y_pred)\n",
      "    print sklearn.metrics.classification_report(Y_true, Y_pred)"
     ],
     "language": "python",
     "metadata": {},
     "outputs": [],
     "prompt_number": 17
    },
    {
     "cell_type": "code",
     "collapsed": false,
     "input": [
      "# Reload the classifier\n",
      "#with open('/home/bmcfee/git/death2chroma/data/frame_classifier_encoded_beatles.pickle', 'r') as f:\n",
      "with open('/home/bmcfee/git/death2chroma/data/frame_classifier_beatles.pickle', 'r') as f:\n",
      "#with open('/home/bmcfee/git/death2chroma/data/frame_classifier_beatles_CL.pickle', 'r') as f:\n",
      "#with open('/home/bmcfee/git/death2chroma/data/frame_classifier_beatles_wrapCL.pickle', 'r') as f:\n",
      "    frame_classifier = pickle.load(f)"
     ],
     "language": "python",
     "metadata": {},
     "outputs": [],
     "prompt_number": 14
    },
    {
     "cell_type": "code",
     "collapsed": false,
     "input": [
      "# Load the test data\n",
      "#_, names, X, beats = load_songs('/home/bmcfee/git/death2chroma/data/beatles/*', features='wrapCL', labels='beats')\n",
      "#_, names, X, beats = load_songs('/home/bmcfee/git/death2chroma/data/beatles/*', features='CL', labels='beats')\n",
      "_, names, X, beats = load_songs('/home/bmcfee/git/death2chroma/data/beatles/*', features='raw-compressed', labels='beats')\n",
      "#_, names, X, beats = load_songs('/home/bmcfee/git/death2chroma/data/beatles/*', features='encoded-compressed', labels='beats')\n",
      "\n",
      "#_, names, X, beats = load_songs('/home/bmcfee/git/death2chroma/data/uspop2002-npy', features='wrapCL', labels='beats')\n",
      "#_, names, X, beats = load_songs('/home/bmcfee/git/death2chroma/data/uspop2002-npy', features='CL', labels='beats')\n",
      "#_, names, X, beats = load_songs('/home/bmcfee/git/death2chroma/data/uspop2002-npy/', features='raw-compressed', labels='beats')\n",
      "#_, names, X, beats = load_songs('/home/bmcfee/git/death2chroma/data/uspop2002-npy/', features='encoded-compressed', labels='beats')\n"
     ],
     "language": "python",
     "metadata": {},
     "outputs": [],
     "prompt_number": 12
    },
    {
     "cell_type": "code",
     "collapsed": false,
     "input": [
      "SONG_NUMBER = 81\n",
      "#SONG_NUMBER = 0\n",
      "confusion = sklearn.metrics.confusion_matrix(Y[SONG_NUMBER], Ypred[SONG_NUMBER], labels=range(25)).astype(float)\n",
      "z_c = confusion.sum(axis=1, keepdims=True)\n",
      "z_c[z_c==0] = 1\n",
      "confusion = confusion / z_c\n",
      "imshow(confusion, aspect='auto', interpolation='nearest', origin='lower')\n",
      "title(os.path.basename(names[SONG_NUMBER]))\n",
      "ylabel('Y_true'), xlabel('Y_pred')\n",
      "xticks(range(25), [chord_classes[chord_indices[i]][0] for i in range(25)], rotation='90')\n",
      "yticks(range(25), [chord_classes[chord_indices[i]][0] for i in range(25)])\n",
      "colorbar()\n",
      "\n",
      "print sklearn.metrics.classification_report(Y[SONG_NUMBER], Ypred[SONG_NUMBER])"
     ],
     "language": "python",
     "metadata": {},
     "outputs": [
      {
       "output_type": "stream",
       "stream": "stdout",
       "text": [
        "             precision    recall  f1-score   support\n",
        "\n",
        "          3       0.00      0.00      0.00         0\n",
        "          4       0.94      0.72      0.82        40\n",
        "          6       0.89      0.72      0.80       127\n",
        "          7       0.00      0.00      0.00         0\n",
        "         11       0.00      0.00      0.00         7\n",
        "         12       0.00      0.00      0.00         0\n",
        "         13       0.74      0.78      0.76        51\n",
        "         14       0.00      0.00      0.00         0\n",
        "         18       0.00      0.00      0.00         0\n",
        "         22       0.50      1.00      0.67         2\n",
        "         23       0.00      0.00      0.00         0\n",
        "\n",
        "avg / total       0.84      0.72      0.77       227\n",
        "\n"
       ]
      },
      {
       "output_type": "display_data",
       "png": "iVBORw0KGgoAAAANSUhEUgAAAeUAAAGkCAYAAAAPND2rAAAABHNCSVQICAgIfAhkiAAAAAlwSFlz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\nCcqUCSGEECvx/wHX3cI9gMZW4AAAAABJRU5ErkJggg==\n"
      }
     ],
     "prompt_number": 23
    },
    {
     "cell_type": "code",
     "collapsed": false,
     "input": [
      "figure(figsize=(18,4))\n",
      "imshow(np.vstack((Y[SONG_NUMBER], Ypred[SONG_NUMBER])), aspect='auto', interpolation='nearest', origin='lower'), colorbar()\n",
      "pass"
     ],
     "language": "python",
     "metadata": {},
     "outputs": [
      {
       "output_type": "display_data",
       "png": "iVBORw0KGgoAAAANSUhEUgAAA6IAAAD9CAYAAACmyMo4AAAABHNCSVQICAgIfAhkiAAAAAlwSFlz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\nAAAAgKn+PwsR/iOAW/B0AAAAAElFTkSuQmCC\n"
      }
     ],
     "prompt_number": 25
    },
    {
     "cell_type": "code",
     "collapsed": false,
     "input": [
      "Ypred = map(lambda z: frame_classifier.predict(z), X)"
     ],
     "language": "python",
     "metadata": {},
     "outputs": [],
     "prompt_number": 15
    },
    {
     "cell_type": "code",
     "collapsed": false,
     "input": [
      "output_predicted_labels('/home/bmcfee/git/death2chroma/data/predictions_uspop', names, beats, Ypred, ext='prediction-linear')\n",
      "#output_predicted_labels('/home/bmcfee/git/death2chroma/data/predictions', names, beats, Ypred, ext='prediction-linear')"
     ],
     "language": "python",
     "metadata": {},
     "outputs": [],
     "prompt_number": 305
    },
    {
     "cell_type": "code",
     "collapsed": false,
     "input": [
      "frame_eval(Y, Ypred)"
     ],
     "language": "python",
     "metadata": {},
     "outputs": [
      {
       "output_type": "stream",
       "stream": "stdout",
       "text": [
        "             precision    recall  f1-score   support\n",
        "\n",
        "          0       0.53      0.63      0.57      1549\n",
        "          1       0.67      0.56      0.61      6749\n",
        "          2       0.26      0.50      0.34       273\n",
        "          3       0.22      0.60      0.32       314\n",
        "          4       0.54      0.54      0.54      2591\n",
        "          5       0.51      0.59      0.54       842\n",
        "          6       0.74      0.67      0.70      7315\n",
        "          7       0.47      0.56      0.51      1578\n",
        "          8       0.63      0.71      0.67      1272\n",
        "          9       0.46      0.54      0.50      1483\n",
        "         10       0.52      0.64      0.57      1113\n",
        "         11       0.51      0.61      0.56      2308\n",
        "         12       0.40      0.46      0.43      1571\n",
        "         13       0.72      0.62      0.66      8269\n",
        "         14       0.43      0.49      0.46      1419\n",
        "         15       0.31      0.49      0.38       608\n",
        "         16       0.38      0.61      0.47       235\n",
        "         17       0.68      0.61      0.64      7047\n",
        "         18       0.40      0.75      0.53       570\n",
        "         19       0.47      0.60      0.53       576\n",
        "         20       0.61      0.61      0.61      2568\n",
        "         21       0.50      0.62      0.55      1004\n",
        "         22       0.40      0.32      0.36      1858\n",
        "         23       0.70      0.61      0.65      5895\n",
        "         24       0.46      0.59      0.52       789\n",
        "\n",
        "avg / total       0.61      0.59      0.60     59796\n",
        "\n"
       ]
      }
     ],
     "prompt_number": 22
    }
   ],
   "metadata": {}
  }
 ]
}