Keras-PowerLaws.ipynb

{
 "cells": [
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "### Keras \n",
    "\n",
    "https://keras.io/applications/#available-models\n",
    "\n",
    "\n"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 4,
   "metadata": {
    "ExecuteTime": {
     "end_time": "2018-08-19T02:18:20.181612Z",
     "start_time": "2018-08-19T02:18:20.153071Z"
    }
   },
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "3.6.5 |Anaconda, Inc.| (default, Apr 26 2018, 08:42:37) \n",
      "[GCC 4.2.1 Compatible Clang 4.0.1 (tags/RELEASE_401/final)]\n",
      "numpy version 1.14.3\n",
      "tensforflow version 1.10.1\n",
      "keras version 2.2.2\n"
     ]
    }
   ],
   "source": [
    "import numpy as np\n",
    "import pickle, time, sys\n",
    "from copy import deepcopy\n",
    "from shutil import copy\n",
    "import import_ipynb\n",
    "from tqdm import tqdm_notebook as tqdm\n",
    "import RMT_Util\n",
    "\n",
    "import matplotlib\n",
    "import matplotlib.pyplot as plt\n",
    "%matplotlib inline\n",
    "\n",
    "import powerlaw\n",
    "\n",
    "import tensorflow as tf\n",
    "import keras\n",
    "from keras.callbacks import TensorBoard, EarlyStopping\n",
    "from keras.optimizers import SGD\n",
    "from keras.initializers import Constant\n",
    "import keras.backend as K\n",
    "\n",
    "from keras.applications import vgg16, vgg19, inception_resnet_v2, mobilenet, nasnet \n",
    "\n",
    "import sklearn\n",
    "from sklearn.decomposition import TruncatedSVD\n",
    "from sklearn.random_projection import sparse_random_matrix\n",
    "\n",
    "\n",
    "print(sys.version)\n",
    "print(\"numpy version {}\".format(np.__version__))\n",
    "print(\"tensforflow version {}\".format(tf.__version__))\n",
    "print(\"keras version {}\".format(keras.__version__))"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "### Fit Marchenko Pastur to ESD\n"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 2,
   "metadata": {
    "ExecuteTime": {
     "end_time": "2018-08-19T02:18:22.147205Z",
     "start_time": "2018-08-19T02:18:21.103515Z"
    }
   },
   "outputs": [],
   "source": [
    "sess = tf.InteractiveSession()\n",
    "sess.run(tf.global_variables_initializer())"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 3,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Downloading data from https://github.com/fchollet/deep-learning-models/releases/download/v0.1/vgg16_weights_tf_dim_ordering_tf_kernels.h5\n",
      "553467904/553467096 [==============================] - 96s 0us/step\n",
      "0 <keras.engine.input_layer.InputLayer object at 0x10db56be0>\n",
      "1 <keras.layers.convolutional.Conv2D object at 0x10db78240>\n",
      "2 <keras.layers.convolutional.Conv2D object at 0x10db934a8>\n",
      "3 <keras.layers.pooling.MaxPooling2D object at 0x10d8cf860>\n",
      "4 <keras.layers.convolutional.Conv2D object at 0x10d8cfd30>\n",
      "5 <keras.layers.convolutional.Conv2D object at 0x10dbc6320>\n",
      "6 <keras.layers.pooling.MaxPooling2D object at 0x10dbdd0f0>\n",
      "7 <keras.layers.convolutional.Conv2D object at 0x10dbdd748>\n",
      "8 <keras.layers.convolutional.Conv2D object at 0x10dbf9630>\n",
      "9 <keras.layers.convolutional.Conv2D object at 0x10dc13048>\n",
      "10 <keras.layers.pooling.MaxPooling2D object at 0x10dc2bf98>\n",
      "11 <keras.layers.convolutional.Conv2D object at 0x10dc45588>\n",
      "12 <keras.layers.convolutional.Conv2D object at 0x10dc62438>\n",
      "13 <keras.layers.convolutional.Conv2D object at 0x10dc78518>\n",
      "14 <keras.layers.pooling.MaxPooling2D object at 0x10dc963c8>\n",
      "15 <keras.layers.convolutional.Conv2D object at 0x10dc96d68>\n",
      "16 <keras.layers.convolutional.Conv2D object at 0x10dccd240>\n",
      "17 <keras.layers.convolutional.Conv2D object at 0x10dcea320>\n",
      "18 <keras.layers.pooling.MaxPooling2D object at 0x10dd031d0>\n",
      "19 <keras.layers.core.Flatten object at 0x10dd03b00>\n",
      "20 <keras.layers.core.Dense object at 0x10dd1a588>\n",
      "21 <keras.layers.core.Dense object at 0x10dd35e80>\n",
      "22 <keras.layers.core.Dense object at 0x10dd52518>\n"
     ]
    }
   ],
   "source": [
    "alphas = []\n",
    "for kmodel in [vgg16.VGG16]:#, vgg19.VGG19, ]:\n",
    "#for kmodel in [inception_resnet_v2]:\n",
    "    kmodel(weights=None)\n",
    "    model = kmodel(weights='imagenet')\n",
    "  \n",
    "    for il, l in enumerate(model.layers):\n",
    "        print(il,l)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 8,
   "metadata": {
    "ExecuteTime": {
     "end_time": "2018-08-19T02:27:09.898958Z",
     "start_time": "2018-08-19T02:18:22.148949Z"
    }
   },
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "0 <keras.engine.input_layer.InputLayer object at 0x10ed28898>\n",
      "1 <keras.layers.convolutional.Conv2D object at 0x10ecf4710>\n",
      "2 <keras.layers.convolutional.Conv2D object at 0x10ecf4b38>\n",
      "3 <keras.layers.pooling.MaxPooling2D object at 0x10ecea0b8>\n",
      "4 <keras.layers.convolutional.Conv2D object at 0x10ecea898>\n",
      "5 <keras.layers.convolutional.Conv2D object at 0x10eccc828>\n",
      "6 <keras.layers.pooling.MaxPooling2D object at 0x10ecc2588>\n",
      "7 <keras.layers.convolutional.Conv2D object at 0x10ecc2d68>\n",
      "8 <keras.layers.convolutional.Conv2D object at 0x10ddc7278>\n",
      "9 <keras.layers.convolutional.Conv2D object at 0x1a1f624080>\n",
      "10 <keras.layers.pooling.MaxPooling2D object at 0x1a1f63c588>\n",
      "11 <keras.layers.convolutional.Conv2D object at 0x1a1f6543c8>\n",
      "12 <keras.layers.convolutional.Conv2D object at 0x1a1f6733c8>\n",
      "13 <keras.layers.convolutional.Conv2D object at 0x1a1f68d470>\n",
      "14 <keras.layers.pooling.MaxPooling2D object at 0x1a1f6a62e8>\n",
      "15 <keras.layers.convolutional.Conv2D object at 0x1a1f6a6ba8>\n",
      "16 <keras.layers.convolutional.Conv2D object at 0x1a1f6db128>\n",
      "17 <keras.layers.convolutional.Conv2D object at 0x1a1f6f5588>\n",
      "18 <keras.layers.pooling.MaxPooling2D object at 0x1a1f712400>\n",
      "19 <keras.layers.core.Flatten object at 0x1a1f712908>\n",
      "20 <keras.layers.core.Dense object at 0x1a2081a390>\n",
      "analyzing\n"
     ]
    },
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "Calculating best minimal value for power law fit\n",
      "/Users/charleshmartin/anaconda3/lib/python3.6/site-packages/powerlaw.py:700: RuntimeWarning: divide by zero encountered in true_divide\n",
      "  (Theoretical_CDF * (1 - Theoretical_CDF))\n",
      "/Users/charleshmartin/anaconda3/lib/python3.6/site-packages/powerlaw.py:700: RuntimeWarning: invalid value encountered in true_divide\n",
      "  (Theoretical_CDF * (1 - Theoretical_CDF))\n",
      "Assuming nested distributions\n"
     ]
    },
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<Figure size 432x288 with 1 Axes>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "% spikes outside bulk 0.00\n",
      "% sigma 1.7930\n"
     ]
    },
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "RMT_Util.ipynb:7: RuntimeWarning: divide by zero encountered in true_divide\n",
      "  \"## RMT Util\\n\",\n",
      "RMT_Util.ipynb:7: RuntimeWarning: invalid value encountered in sqrt\n",
      "  \"## RMT Util\\n\",\n"
     ]
    },
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<Figure size 432x288 with 1 Axes>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "2.1147459382101115 0.025720056731000085 TPL\n"
     ]
    },
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<Figure size 432x288 with 1 Axes>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "21 <keras.layers.core.Dense object at 0x1a20832c50>\n",
      "analyzing\n"
     ]
    },
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "Calculating best minimal value for power law fit\n",
      "Assuming nested distributions\n"
     ]
    },
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<Figure size 432x288 with 1 Axes>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "% spikes outside bulk 0.00\n",
      "% sigma 1.3976\n"
     ]
    },
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "RMT_Util.ipynb:7: RuntimeWarning: invalid value encountered in multiply\n",
      "  \"## RMT Util\\n\",\n"
     ]
    },
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<Figure size 432x288 with 1 Axes>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "2.0334381348824206 0.03389597526031429 TPL\n"
     ]
    },
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<Figure size 432x288 with 1 Axes>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "22 <keras.layers.core.Dense object at 0x1a208512e8>\n",
      "analyzing\n"
     ]
    },
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "Calculating best minimal value for power law fit\n",
      "Assuming nested distributions\n"
     ]
    },
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<Figure size 432x288 with 1 Axes>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "% spikes outside bulk 0.00\n",
      "% sigma 1.9904\n"
     ]
    },
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<Figure size 432x288 with 1 Axes>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "3.0306202560474653 0.03604779539528424 PL\n"
     ]
    },
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<Figure size 432x288 with 1 Axes>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    }
   ],
   "source": [
    "alphas = []\n",
    "for kmodel in [vgg16.VGG16]:#, vgg19.VGG19, ]:\n",
    "#for kmodel in [inception_resnet_v2]:\n",
    "    kmodel(weights=None)\n",
    "    model = kmodel(weights='imagenet')\n",
    "  \n",
    "    for il, l in enumerate(model.layers):\n",
    "        print(il,l)\n",
    "        if (isinstance(l, keras.layers.core.Dense)):  \n",
    "            print(\"analyzing\")\n",
    "            if (l.get_weights()):\n",
    "                W = l.get_weights()[0]\n",
    "                M, N = np.min(W.shape), np.max(W.shape)\n",
    "                Q=N/M\n",
    "\n",
    "                svd = TruncatedSVD(n_components=M-1, n_iter=7, random_state=10)\n",
    "                svd.fit(W) \n",
    "                sv = svd.singular_values_\n",
    "                evals_imagenet = sv*sv\n",
    "\n",
    "                evals = evals_imagenet\n",
    "                fit = powerlaw.Fit(evals, xmax=np.max(evals))\n",
    "                alpha, D, best = fit.alpha, fit.D, RMT_Util.best_dist(fit)\n",
    "\n",
    "                fig2 = fit.plot_pdf(color='b', linewidth=2)\n",
    "                fit.power_law.plot_pdf(color='b', linestyle='--', ax=fig2)\n",
    "                fit.plot_ccdf(color='r', linewidth=2, ax=fig2)\n",
    "                fit.power_law.plot_ccdf(color='r', linestyle='--', ax=fig2)\n",
    "                plt.show()\n",
    "\n",
    "                #if (alpha > 4.0):\n",
    "                RMT_Util.plot_ESD_and_fit(model=None, eigenvalues=evals_imagenet, layer=None, \n",
    "                                            Q=Q, num_spikes=0, sigma=None, color='orange')\n",
    "                plt.title(\"L {} ({}x{}) PL {} \".format(il, M,N, alpha))\n",
    "                plt.show()          \n",
    "                print(alpha, D, best)\n",
    "\n",
    "\n",
    "\n",
    "                plt.loglog(evals_imagenet)\n",
    "                plt.show()\n",
    "\n",
    "                alphas.append(alpha)\n",
    "\n"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 9,
   "metadata": {
    "ExecuteTime": {
     "end_time": "2018-08-19T01:56:21.546448Z",
     "start_time": "2018-08-19T01:56:20.999724Z"
    }
   },
   "outputs": [
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<Figure size 432x288 with 1 Axes>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    }
   ],
   "source": [
    "plt.hist(evals_imagenet,bins=200,log=True, density=True)\n",
    "plt.xscale('log')\n",
    "plt.show()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {
    "ExecuteTime": {
     "end_time": "2018-08-19T02:04:27.588048Z",
     "start_time": "2018-08-19T02:04:27.585312Z"
    }
   },
   "outputs": [],
   "source": [
    "import powerlaw\n",
    "evals = evals_imagenet\n"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {
    "ExecuteTime": {
     "end_time": "2018-08-19T02:05:17.272815Z",
     "start_time": "2018-08-19T02:05:17.008226Z"
    }
   },
   "outputs": [],
   "source": []
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {},
   "outputs": [],
   "source": []
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {},
   "outputs": [],
   "source": []
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {
    "ExecuteTime": {
     "end_time": "2018-08-19T01:51:43.232804Z",
     "start_time": "2018-08-19T01:51:43.113440Z"
    }
   },
   "outputs": [],
   "source": [
    "print(len(alphas))\n",
    "plt.hist(alphas);"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {
    "ExecuteTime": {
     "end_time": "2018-08-05T03:46:51.992507Z",
     "start_time": "2018-08-05T03:46:51.927926Z"
    }
   },
   "outputs": [],
   "source": [
    "\n",
    "def analyze_model(model):\n",
    "    alphas = []\n",
    "    for l in model.layers:\n",
    "        print(l)\n",
    "        if l.get_weights():\n",
    "            W = l.get_weights()[0]\n",
    "            W = RMT_Util.reshape_tensor(W)\n",
    "            M, N = np.min(W.shape), np.max(W.shape)\n",
    "            if ((N>43) and (M>43)):\n",
    "                Q=N/M\n",
    "\n",
    "                svd = TruncatedSVD(n_components=M-1, n_iter=7, random_state=10)\n",
    "                svd.fit(W) \n",
    "                sv = svd.singular_values_\n",
    "                evals_imagenet = sv*sv\n",
    "\n",
    "                RMT_Util.plot_ESD_and_fit(model=None, eigenvalues=evals_imagenet, layer=None, \n",
    "                                            Q=Q, num_spikes=0, sigma=None, color='orange')\n",
    "                plt.title(\"Layer ({}x{})\".format(M,N))\n",
    "                plt.show()     \n",
    "\n",
    "                alpha, D, best = RMT_Util.fit_powerlaw(evals_imagenet)\n",
    "                print(alpha, D, best)\n",
    "                alphas.append(alpha)\n",
    "    return alphas"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {
    "ExecuteTime": {
     "end_time": "2018-08-19T01:59:28.306532Z",
     "start_time": "2018-08-19T01:59:28.294276Z"
    }
   },
   "outputs": [],
   "source": []
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {
    "ExecuteTime": {
     "end_time": "2018-08-04T14:37:24.878056Z",
     "start_time": "2018-08-04T13:58:06.018963Z"
    }
   },
   "outputs": [],
   "source": [
    "for kmodel in [inception_resnet_v2.InceptionResNetV2]:\n",
    "    kmodel(weights=None)\n",
    "    model = kmodel(weights='imagenet')\n",
    "    alphas = analyze_model(model)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {
    "ExecuteTime": {
     "end_time": "2018-08-05T00:17:18.445830Z",
     "start_time": "2018-08-05T00:17:18.110673Z"
    }
   },
   "outputs": [],
   "source": [
    "print(len(alphas))\n",
    "plt.hist(alphas,bins=50);\n",
    "plt.title(\"power law fits, InceptionResNetV2\")"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {
    "ExecuteTime": {
     "end_time": "2018-08-05T00:23:54.489894Z",
     "start_time": "2018-08-05T00:17:46.116584Z"
    }
   },
   "outputs": [],
   "source": [
    "for kmodel in [mobilenet.MobileNet]:\n",
    "    kmodel(weights=None)\n",
    "    model = kmodel(weights='imagenet')\n",
    "    alphas = analyze_model(model)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {
    "ExecuteTime": {
     "end_time": "2018-08-05T00:30:13.658040Z",
     "start_time": "2018-08-05T00:30:13.478758Z"
    }
   },
   "outputs": [],
   "source": [
    "print(len(alphas))\n",
    "plt.hist(alphas,bins=30);\n",
    "plt.title(\"power law fits, MobileNet\")"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {
    "ExecuteTime": {
     "end_time": "2018-08-05T02:29:17.436198Z",
     "start_time": "2018-08-05T00:43:11.697296Z"
    }
   },
   "outputs": [],
   "source": [
    "for kmodel in [nasnet.NASNetLarge]:\n",
    "    kmodel(weights=None)\n",
    "    model = kmodel(weights='imagenet')\n",
    "    alphas = analyze_model(model)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {
    "ExecuteTime": {
     "end_time": "2018-08-05T02:29:17.698767Z",
     "start_time": "2018-08-05T02:29:17.487071Z"
    }
   },
   "outputs": [],
   "source": [
    "print(len(alphas))\n",
    "plt.hist(alphas,bins=30);\n",
    "plt.title(\"power law fits, NasNetLarge\")"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {
    "ExecuteTime": {
     "end_time": "2018-08-05T15:21:46.052233Z",
     "start_time": "2018-08-05T03:46:57.587030Z"
    }
   },
   "outputs": [],
   "source": [
    "for kmodel in [nasnet.NASNetMobile]:\n",
    "    kmodel(weights=None)\n",
    "    model = kmodel(weights='imagenet')\n",
    "    alphas = analyze_model(model)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {
    "ExecuteTime": {
     "end_time": "2018-08-05T03:46:32.069232Z",
     "start_time": "2018-08-05T03:14:43.675Z"
    }
   },
   "outputs": [],
   "source": [
    "print(len(alphas))\n",
    "plt.hist(alphas,bins=30);\n",
    "plt.title(\"power law fits, NasNetMobile\")"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {
    "ExecuteTime": {
     "end_time": "2018-08-19T01:37:08.444092Z",
     "start_time": "2018-08-19T01:37:04.034096Z"
    }
   },
   "outputs": [],
   "source": [
    "!pip install oct2py"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {},
   "outputs": [],
   "source": []
  }
 ],
 "metadata": {
  "kernelspec": {
   "display_name": "Python 3",
   "language": "python",
   "name": "python3"
  },
  "language_info": {
   "codemirror_mode": {
    "name": "ipython",
    "version": 3
   },
   "file_extension": ".py",
   "mimetype": "text/x-python",
   "name": "python",
   "nbconvert_exporter": "python",
   "pygments_lexer": "ipython3",
   "version": "3.6.5"
  }
 },
 "nbformat": 4,
 "nbformat_minor": 2
}