DeepBeam.ipynb

{
 "cells": [
  {
   "cell_type": "code",
   "execution_count": 1,
   "metadata": {},
   "outputs": [],
   "source": [
    "import numpy as np\n",
    "import matplotlib.pyplot as plt\n",
    "from scipy.interpolate import interpn\n",
    "import os\n",
    "import glob\n",
    "import random\n",
    "\n",
    "import config\n",
    "import utils\n"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 2,
   "metadata": {},
   "outputs": [
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "Using TensorFlow backend.\n"
     ]
    }
   ],
   "source": [
    "from keras.models import Model\n",
    "from keras.layers import BatchNormalization, Conv2D, Conv2DTranspose, MaxPooling2D, Dropout, UpSampling2D, Input, concatenate, Activation,LeakyReLU\n",
    "from keras import backend as K\n",
    "import tensorflow as tf\n",
    "from keras.callbacks import ModelCheckpoint\n",
    "from keras.optimizers import Adam, SGD\n"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 8,
   "metadata": {},
   "outputs": [],
   "source": [
    "# Set model_filename to either name of a file to which a new trained model has to be saved or to a name of a file from which\n",
    "# trained model weights have to be loaded\n",
    "\n",
    "model_filename = 'regresja1.h5'\n"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 4,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "(3, 300, 6, 495) (300, 4)\n"
     ]
    }
   ],
   "source": [
    "# Read all simulated profiles for a regular grid of primary beam parameters, for fields 3x3, 10x10, and 30x30\n",
    "\n",
    "dataPoints = [(str(e),str(se),str(s),str(an)) for e in config.simulatedEnergies for se in config.simulatedEnergyDispersions for s in config.simulatedSourceSizes \n",
    "              for an in config.simulatedAngularDivergences]\n",
    "\n",
    "random.seed(config.SEED)\n",
    "random.shuffle(dataPoints)\n",
    "\n",
    "profiles = utils.readProfiles(config.profileDIR,dataPoints)\n",
    "\n",
    "profiles = np.asarray(profiles)\n",
    "y_train = np.asarray(dataPoints,dtype=np.float)\n",
    "\n",
    "print(profiles.shape,y_train.shape)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 5,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "(300, 348, 1)\n",
      "(300, 141, 1)\n",
      "(300, 348, 1)\n",
      "(300, 231, 1)\n",
      "(300, 491, 1)\n"
     ]
    }
   ],
   "source": [
    "\n",
    "X_train = []\n",
    "\n",
    "for nfield,(field,Ranges) in enumerate(zip(config.analyzedProfiles,config.analyzedRanges)):\n",
    "    if field != None:\n",
    "        for profile,Range in zip(field,Ranges):\n",
    "            inp = profiles[nfield,:,profile,Range[0]:Range[1]]\n",
    "            inp = np.reshape(inp,inp.shape + (1,))\n",
    "            print(inp.shape)\n",
    "            X_train.append(inp)\n",
    "            "
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 6,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "(348, 1)\n",
      "(141, 1)\n",
      "(348, 1)\n",
      "(231, 1)\n",
      "(491, 1)\n"
     ]
    }
   ],
   "source": [
    "from keras.layers import Conv1D,MaxPooling1D,Concatenate,Flatten,Dense\n",
    "\n",
    "def conv_block(inp,filters=16,kernel_size=3,strides = 1,kernel_initializer= 'glorot_uniform',padding='same',activation='relu'):\n",
    "    c = Conv1D(filters,kernel_size,kernel_initializer=kernel_initializer,padding=padding,activation=activation,strides = strides)(inp)\n",
    "    c = Conv1D(filters,kernel_size,kernel_initializer=kernel_initializer,padding=padding,activation=activation,strides = strides)(c)\n",
    "    c = MaxPooling1D(2)(c)\n",
    "    return c\n",
    "\n",
    "inputs = []\n",
    "outputs = []\n",
    "for nmod in range(len(config.allRanges)):\n",
    "    shape = (config.allRanges[nmod][1]-config.allRanges[nmod][0],1)\n",
    "    print(shape)\n",
    "    W = int(np.log(shape[0]/3)/np.log(2))\n",
    "    inp = Input(shape)\n",
    "    F = 16\n",
    "    x = inp\n",
    "    for w in range(W):\n",
    "        x = conv_block(x,filters = F)\n",
    "        F *= 2\n",
    "    out = Conv1D(1,1) (x)\n",
    "    inputs.append(inp)\n",
    "    outputs.append(out)\n",
    "\n",
    "c = Concatenate(axis=1)(outputs)\n",
    "c = Flatten()(c)\n",
    "c = Dense(100,activation='relu')(c)\n",
    "c = Dense(100,activation='relu')(c)\n",
    "out = Dense(4,activation=None)(c)\n",
    "\n",
    "model = Model(inputs = inputs,outputs=[out])\n",
    "#model.summary()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 7,
   "metadata": {},
   "outputs": [],
   "source": [
    "\n",
    "callback_checkpoint = ModelCheckpoint(\n",
    "    model_filename, \n",
    "    verbose=1, \n",
    "    monitor='val_loss', \n",
    "    save_best_only=True\n",
    ")\n",
    "\n",
    "model.compile(\n",
    "    optimizer=Adam(lr=0.0001), \n",
    "    loss = 'mse',\n",
    "    metrics=[tf.keras.metrics.MeanAbsoluteError()]\n",
    ")\n"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {},
   "outputs": [],
   "source": [
    "# Run only if the trained model does not yet exist exists\n",
    "\n",
    "history = model.fit(\n",
    "    X_train,y_train,\n",
    "    steps_per_epoch=100,\n",
    "    epochs=300,\n",
    "    validation_split=0.2,\n",
    "    validation_steps = 10,\n",
    "    callbacks=[callback_checkpoint]\n",
    ")\n"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 9,
   "metadata": {},
   "outputs": [],
   "source": [
    "# Run if the trained model already exists\n",
    "\n",
    "model.load_weights(model_filename)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 10,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "(3, 25, 6, 495) (25, 4)\n"
     ]
    }
   ],
   "source": [
    "groundTruthFilename = config.testProfilesDIR + config.groundTruthFileName\n",
    "\n",
    "testGoals = open(groundTruthFilename)\n",
    "lines = testGoals.readlines()\n",
    "y_test = [l[:-1].split()[1:5] for l in lines[:-1]]\n",
    "fileNames = [l[:-1].split()[0] for l in lines[:-1]]\n",
    "testGoals.close()\n",
    "\n",
    "fields = ['fields','fields10','fields30']\n",
    "testProfiles3 = []\n",
    "testProfiles10 = []\n",
    "testProfiles30 = []\n",
    "\n",
    "for fileName in fileNames:\n",
    "    name = config.testProfilesDIR + fileName + '_' + fields[0] + '.npz'\n",
    "    file = np.load(name)\n",
    "    testProfiles3.append(file[file.files[1]])\n",
    "\n",
    "    name = config.testProfilesDIR + fileName + '_' + fields[1] + '.npz'\n",
    "    file = np.load(name)\n",
    "    testProfiles10.append(file[file.files[1]])\n",
    "\n",
    "    name = config.testProfilesDIR + fileName + '_' + fields[2] + '.npz'\n",
    "    file = np.load(name)\n",
    "    testProfiles30.append(file[file.files[1]])\n",
    "        \n",
    "\n",
    "testProfiles = []\n",
    "testProfiles.append(testProfiles3)\n",
    "testProfiles.append(testProfiles10)\n",
    "testProfiles.append(testProfiles30)\n",
    "\n",
    "testProfiles = np.asarray(testProfiles,dtype=np.float)\n",
    "y_test = np.asarray(y_test,dtype=np.float)\n",
    "\n",
    "print(testProfiles.shape,y_test.shape)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 11,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "(25, 348, 1)\n",
      "(25, 141, 1)\n",
      "(25, 348, 1)\n",
      "(25, 231, 1)\n",
      "(25, 491, 1)\n"
     ]
    }
   ],
   "source": [
    "\n",
    "X_test = []\n",
    "\n",
    "for nfield,(field,Ranges) in enumerate(zip(config.analyzedProfiles,config.analyzedRanges)):\n",
    "    if field != None:\n",
    "        for profile,Range in zip(field,Ranges):\n",
    "            inp = testProfiles[nfield,:,profile,Range[0]:Range[1]]\n",
    "            inp = np.reshape(inp,inp.shape + (1,))\n",
    "            print(inp.shape)\n",
    "            X_test.append(inp)\n"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 12,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "(25, 4)\n"
     ]
    }
   ],
   "source": [
    "y_predicted = model.predict(X_test)\n",
    "print(y_predicted.shape)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 13,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<Figure size 720x720 with 1 Axes>"
      ]
     },
     "metadata": {
      "needs_background": "light"
     },
     "output_type": "display_data"
    },
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<Figure size 720x720 with 1 Axes>"
      ]
     },
     "metadata": {
      "needs_background": "light"
     },
     "output_type": "display_data"
    },
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<Figure size 720x720 with 1 Axes>"
      ]
     },
     "metadata": {
      "needs_background": "light"
     },
     "output_type": "display_data"
    },
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<Figure size 720x720 with 1 Axes>"
      ]
     },
     "metadata": {
      "needs_background": "light"
     },
     "output_type": "display_data"
    }
   ],
   "source": [
    "from scipy.stats import pearsonr\n",
    "\n",
    "for par in range(4):\n",
    "    plt.figure(figsize=(10,10))\n",
    "    corr, _ = pearsonr(y_test[:,par], y_predicted[:,par])\n",
    "    plt.plot(y_test[:,par],y_predicted[:,par],'og',label='R='+str(np.round(corr,3)))\n",
    "    plt.legend(loc='upper left')\n",
    "    plt.show()"
   ]
  },
  {
   "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": "3.6.10"
  }
 },
 "nbformat": 4,
 "nbformat_minor": 4
}