MergeNet_image_prob_mergedarray_largerborderoverlap.ipynb

{
 "cells": [
  {
   "cell_type": "code",
   "execution_count": 1,
   "metadata": {
    "collapsed": false
   },
   "outputs": [
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "/home/d/nolearn/local/lib/python2.7/site-packages/matplotlib/font_manager.py:273: UserWarning: Matplotlib is building the font cache using fc-list. This may take a moment.\n",
      "  warnings.warn('Matplotlib is building the font cache using fc-list. This may take a moment.')\n",
      "Using gpu device 0: GeForce GTX TITAN Black (CNMeM is disabled, CuDNN 4007)\n",
      "/home/d/nolearn/local/lib/python2.7/site-packages/theano/tensor/signal/downsample.py:6: UserWarning: downsample module has been moved to the theano.tensor.signal.pool module.\n",
      "  \"downsample module has been moved to the theano.tensor.signal.pool module.\")\n"
     ]
    }
   ],
   "source": [
    "%load_ext autoreload\n",
    "%autoreload 2\n",
    "\n",
    "import mlproof as mlp\n",
    "import mlproof.nets as nets\n",
    "\n",
    "import numpy as np\n",
    "\n",
    "from nolearn.lasagne.visualize import plot_loss\n",
    "from nolearn.lasagne.visualize import plot_conv_weights\n",
    "from nolearn.lasagne.visualize import plot_conv_activity\n",
    "from nolearn.lasagne.visualize import plot_occlusion\n",
    "\n",
    "from matplotlib.pyplot import imshow\n",
    "import matplotlib.pyplot as plt\n",
    "%matplotlib inline"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 2,
   "metadata": {
    "collapsed": false
   },
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "CNN configuration: \n",
      "    Our CNN with different \"legs\" for image, prob, merged_array, border_overlap.\n",
      "    \n"
     ]
    }
   ],
   "source": [
    "cnn = nets.MergeNet()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 3,
   "metadata": {
    "collapsed": false
   },
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Loaded /home/d/patches/cylinder_small1/ in 0.00182795524597 seconds.\n"
     ]
    }
   ],
   "source": [
    "X_train, y_train, X_test, y_test = mlp.Patch.load('cylinder_small1')"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 4,
   "metadata": {
    "collapsed": false
   },
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "# Neural Network with 1511874 learnable parameters\n",
      "\n",
      "## Layer information\n",
      "\n",
      "  #  name          size\n",
      "---  ------------  --------\n",
      "  0  image_input   1x75x75\n",
      "  1  image_conv1   16x63x63\n",
      "  2  image_pool1   16x31x31\n",
      "  3  image_conv2   16x19x19\n",
      "  4  image_pool2   16x9x9\n",
      "  5  prob_input    1x75x75\n",
      "  6  prob_conv1    16x63x63\n",
      "  7  prob_pool1    16x31x31\n",
      "  8  prob_conv2    16x19x19\n",
      "  9  prob_pool2    16x9x9\n",
      " 10  binary_input  1x75x75\n",
      " 11  binary_conv1  16x63x63\n",
      " 12  binary_pool1  16x31x31\n",
      " 13  binary_conv2  16x19x19\n",
      " 14  binary_pool2  16x9x9\n",
      " 15  border_input  1x75x75\n",
      " 16  border_conv1  16x63x63\n",
      " 17  border_pool1  16x31x31\n",
      " 18  border_conv2  16x19x19\n",
      " 19  border_pool2  16x9x9\n",
      " 20  merge         64x9x9\n",
      " 21  hidden3       256\n",
      " 22  dropout3      256\n",
      " 23  output        2\n",
      "\n",
      "  epoch    train loss    valid loss    train/val    valid acc  dur\n",
      "-------  ------------  ------------  -----------  -----------  ------\n",
      "      1       \u001b[36m0.58617\u001b[0m       \u001b[32m0.49436\u001b[0m      1.18570      0.76099  29.39s\n",
      "      2       \u001b[36m0.47847\u001b[0m       \u001b[32m0.43490\u001b[0m      1.10019      0.79881  29.95s\n",
      "      3       \u001b[36m0.44625\u001b[0m       \u001b[32m0.41767\u001b[0m      1.06843      0.81170  29.69s\n",
      "      4       \u001b[36m0.42967\u001b[0m       \u001b[32m0.40724\u001b[0m      1.05510      0.81367  28.55s\n",
      "      5       \u001b[36m0.41904\u001b[0m       \u001b[32m0.40326\u001b[0m      1.03914      0.82225  29.53s\n",
      "      6       \u001b[36m0.40983\u001b[0m       \u001b[32m0.39401\u001b[0m      1.04017      0.82334  29.56s\n",
      "      7       \u001b[36m0.40334\u001b[0m       0.39403      1.02364      0.82782  29.21s\n",
      "      8       \u001b[36m0.39369\u001b[0m       \u001b[32m0.38603\u001b[0m      1.01985      0.83384  29.40s\n",
      "      9       \u001b[36m0.39025\u001b[0m       \u001b[32m0.37387\u001b[0m      1.04381      0.83891  29.71s\n",
      "     10       \u001b[36m0.38351\u001b[0m       \u001b[32m0.37058\u001b[0m      1.03490      0.84018  29.37s\n",
      "     11       \u001b[36m0.37444\u001b[0m       \u001b[32m0.36604\u001b[0m      1.02295      0.84225  29.46s\n",
      "     12       \u001b[36m0.37072\u001b[0m       0.36719      1.00962      0.84192  29.64s\n",
      "     13       \u001b[36m0.36933\u001b[0m       \u001b[32m0.35331\u001b[0m      1.04534      0.85155  28.92s\n",
      "     14       \u001b[36m0.36314\u001b[0m       0.35673      1.01797      0.85204  29.01s\n",
      "     15       \u001b[36m0.36104\u001b[0m       \u001b[32m0.34849\u001b[0m      1.03600      0.85570  28.93s\n",
      "     16       \u001b[36m0.35622\u001b[0m       \u001b[32m0.34819\u001b[0m      1.02304      0.85439  28.99s\n",
      "     17       \u001b[36m0.35419\u001b[0m       \u001b[32m0.34299\u001b[0m      1.03264      0.85712  29.03s\n",
      "     18       \u001b[36m0.35234\u001b[0m       0.35576      0.99040      0.85242  29.59s\n",
      "     19       \u001b[36m0.34569\u001b[0m       0.34374      1.00567      0.85533  28.86s\n",
      "     20       \u001b[36m0.34374\u001b[0m       \u001b[32m0.33996\u001b[0m      1.01111      0.86095  28.82s\n",
      "     21       0.34418       \u001b[32m0.33895\u001b[0m      1.01543      0.85789  28.89s\n",
      "     22       \u001b[36m0.33925\u001b[0m       \u001b[32m0.32964\u001b[0m      1.02916      0.86007  28.72s\n",
      "     23       \u001b[36m0.33449\u001b[0m       \u001b[32m0.32903\u001b[0m      1.01660      0.86226  28.66s\n",
      "     24       \u001b[36m0.33287\u001b[0m       \u001b[32m0.32513\u001b[0m      1.02382      0.86248  28.69s\n",
      "     25       \u001b[36m0.33188\u001b[0m       \u001b[32m0.32354\u001b[0m      1.02579      0.86160  28.75s\n",
      "     26       \u001b[36m0.32760\u001b[0m       0.32767      0.99978      0.86084  28.65s\n",
      "     27       \u001b[36m0.32627\u001b[0m       \u001b[32m0.31328\u001b[0m      1.04147      0.86931  28.64s\n",
      "     28       \u001b[36m0.32368\u001b[0m       0.33061      0.97902      0.85866  28.68s\n",
      "     29       \u001b[36m0.32030\u001b[0m       0.31475      1.01761      0.86883  28.71s\n",
      "     30       \u001b[36m0.31544\u001b[0m       \u001b[32m0.31306\u001b[0m      1.00758      0.86637  28.65s\n",
      "     31       \u001b[36m0.31530\u001b[0m       \u001b[32m0.31044\u001b[0m      1.01566      0.86942  28.65s\n",
      "     32       \u001b[36m0.31516\u001b[0m       0.31191      1.01041      0.86767  28.71s\n",
      "     33       \u001b[36m0.30985\u001b[0m       0.31440      0.98553      0.86384  28.63s\n",
      "     34       0.31194       \u001b[32m0.30842\u001b[0m      1.01141      0.86920  28.68s\n",
      "     35       \u001b[36m0.30640\u001b[0m       \u001b[32m0.30717\u001b[0m      0.99749      0.87144  28.62s\n",
      "     36       \u001b[36m0.30510\u001b[0m       \u001b[32m0.30196\u001b[0m      1.01040      0.87128  28.67s\n",
      "     37       \u001b[36m0.30189\u001b[0m       \u001b[32m0.30139\u001b[0m      1.00164      0.86926  28.61s\n",
      "     38       \u001b[36m0.30117\u001b[0m       \u001b[32m0.29661\u001b[0m      1.01539      0.87615  28.64s\n",
      "     39       \u001b[36m0.29826\u001b[0m       \u001b[32m0.29537\u001b[0m      1.00978      0.87756  28.62s\n",
      "     40       0.29827       0.29644      1.00617      0.87461  28.64s\n",
      "     41       0.29971       0.30135      0.99457      0.87095  28.67s\n",
      "     42       \u001b[36m0.29603\u001b[0m       \u001b[32m0.29397\u001b[0m      1.00698      0.87434  28.61s\n",
      "     43       \u001b[36m0.29294\u001b[0m       \u001b[32m0.29126\u001b[0m      1.00576      0.87516  28.67s\n",
      "     44       \u001b[36m0.29145\u001b[0m       \u001b[32m0.28501\u001b[0m      1.02259      0.88292  28.66s\n",
      "     45       \u001b[36m0.28728\u001b[0m       0.28862      0.99535      0.87855  28.62s\n",
      "     46       \u001b[36m0.28479\u001b[0m       0.29735      0.95778      0.87380  28.65s\n",
      "     47       0.28527       0.29134      0.97919      0.87866  28.58s\n",
      "     48       \u001b[36m0.27966\u001b[0m       \u001b[32m0.28301\u001b[0m      0.98815      0.88342  28.62s\n",
      "     49       0.28079       0.28493      0.98544      0.87632  28.60s\n",
      "     50       \u001b[36m0.27654\u001b[0m       0.28583      0.96749      0.87975  28.57s\n",
      "     51       \u001b[36m0.27494\u001b[0m       0.28348      0.96986      0.87887  28.63s\n",
      "     52       0.27590       \u001b[32m0.28111\u001b[0m      0.98148      0.88090  28.58s\n",
      "     53       \u001b[36m0.27262\u001b[0m       \u001b[32m0.27406\u001b[0m      0.99475      0.88391  28.66s\n",
      "     54       0.27367       0.27534      0.99395      0.88370  28.59s\n",
      "     55       \u001b[36m0.27051\u001b[0m       0.27457      0.98523      0.88369  28.61s\n",
      "     56       \u001b[36m0.26852\u001b[0m       \u001b[32m0.27083\u001b[0m      0.99147      0.88669  28.75s\n",
      "     57       \u001b[36m0.26589\u001b[0m       0.27298      0.97404      0.88483  28.75s\n",
      "     58       \u001b[36m0.26331\u001b[0m       \u001b[32m0.26939\u001b[0m      0.97742      0.88330  28.59s\n",
      "     59       \u001b[36m0.26050\u001b[0m       \u001b[32m0.26748\u001b[0m      0.97388      0.88773  28.70s\n",
      "     60       \u001b[36m0.25905\u001b[0m       \u001b[32m0.25826\u001b[0m      1.00307      0.89391  28.84s\n",
      "     61       \u001b[36m0.25692\u001b[0m       0.26449      0.97138      0.88746  28.92s\n",
      "     62       \u001b[36m0.25509\u001b[0m       0.26550      0.96078      0.88987  28.74s\n",
      "     63       \u001b[36m0.25366\u001b[0m       0.26001      0.97558      0.88839  28.84s\n",
      "     64       0.25374       \u001b[32m0.25490\u001b[0m      0.99545      0.89446  28.84s\n",
      "     65       \u001b[36m0.24973\u001b[0m       0.26225      0.95227      0.88888  28.92s\n",
      "     66       \u001b[36m0.24892\u001b[0m       0.26355      0.94448      0.89118  28.87s\n",
      "     67       \u001b[36m0.24690\u001b[0m       0.25674      0.96167      0.89227  28.67s\n",
      "     68       \u001b[36m0.24295\u001b[0m       \u001b[32m0.24410\u001b[0m      0.99529      0.90118  28.95s\n",
      "     69       \u001b[36m0.24219\u001b[0m       0.25501      0.94972      0.89385  28.99s\n",
      "     70       \u001b[36m0.24062\u001b[0m       0.25580      0.94067      0.89342  28.81s\n",
      "     71       \u001b[36m0.23607\u001b[0m       0.24712      0.95528      0.89620  29.01s\n",
      "     72       \u001b[36m0.23205\u001b[0m       \u001b[32m0.24212\u001b[0m      0.95840      0.89920  28.67s\n",
      "     73       \u001b[36m0.23108\u001b[0m       \u001b[32m0.24112\u001b[0m      0.95836      0.90019  28.69s\n",
      "     74       \u001b[36m0.23065\u001b[0m       0.24392      0.94557      0.89713  28.71s\n",
      "     75       \u001b[36m0.22715\u001b[0m       \u001b[32m0.23702\u001b[0m      0.95834      0.89921  28.68s\n",
      "     76       0.22981       \u001b[32m0.23675\u001b[0m      0.97070      0.90139  28.71s\n",
      "     77       \u001b[36m0.22634\u001b[0m       \u001b[32m0.23420\u001b[0m      0.96644      0.90129  28.72s\n",
      "     78       \u001b[36m0.22612\u001b[0m       \u001b[32m0.22854\u001b[0m      0.98942      0.90708  28.76s\n",
      "     79       \u001b[36m0.21822\u001b[0m       \u001b[32m0.22382\u001b[0m      0.97495      0.91036  28.79s\n",
      "     80       0.21944       0.23340      0.94020      0.90419  28.88s\n",
      "     81       \u001b[36m0.21281\u001b[0m       0.22618      0.94090      0.90588  28.81s\n",
      "     82       0.21345       0.23436      0.91077      0.90358  28.85s\n",
      "     83       0.21445       0.23315      0.91980      0.90397  28.76s\n",
      "     84       \u001b[36m0.21032\u001b[0m       \u001b[32m0.22250\u001b[0m      0.94524      0.90959  28.74s\n",
      "     85       0.21069       0.22757      0.92585      0.90730  28.80s\n",
      "     86       \u001b[36m0.20927\u001b[0m       \u001b[32m0.21446\u001b[0m      0.97580      0.91173  28.72s\n",
      "     87       \u001b[36m0.20740\u001b[0m       0.21815      0.95073      0.91025  28.75s\n",
      "     88       \u001b[36m0.20475\u001b[0m       0.21447      0.95471      0.91708  28.73s\n",
      "     89       0.20626       \u001b[32m0.20186\u001b[0m      1.02177      0.91784  28.72s\n",
      "     90       \u001b[36m0.19923\u001b[0m       0.21295      0.93558      0.91397  28.71s\n",
      "     91       0.19983       0.21036      0.94994      0.91353  28.68s\n",
      "     92       \u001b[36m0.19663\u001b[0m       \u001b[32m0.20106\u001b[0m      0.97794      0.92183  28.71s\n",
      "     93       \u001b[36m0.19411\u001b[0m       0.20417      0.95071      0.91665  28.69s\n",
      "     94       \u001b[36m0.19267\u001b[0m       0.20912      0.92134      0.91495  28.73s\n",
      "     95       \u001b[36m0.19039\u001b[0m       0.20296      0.93809      0.92052  28.70s\n",
      "     96       \u001b[36m0.18919\u001b[0m       \u001b[32m0.20088\u001b[0m      0.94183      0.92124  28.69s\n",
      "     97       \u001b[36m0.18487\u001b[0m       \u001b[32m0.19459\u001b[0m      0.95004      0.92462  28.72s\n",
      "     98       0.18540       \u001b[32m0.19227\u001b[0m      0.96424      0.92861  28.71s\n",
      "     99       \u001b[36m0.18073\u001b[0m       \u001b[32m0.19218\u001b[0m      0.94038      0.92271  28.70s\n",
      "    100       \u001b[36m0.17896\u001b[0m       0.19363      0.92426      0.92206  28.69s\n",
      "    101       0.18254       \u001b[32m0.19030\u001b[0m      0.95922      0.92730  28.86s\n",
      "    102       \u001b[36m0.17724\u001b[0m       \u001b[32m0.18904\u001b[0m      0.93756      0.92960  28.91s\n",
      "    103       \u001b[36m0.17613\u001b[0m       0.19163      0.91911      0.92818  28.74s\n",
      "    104       \u001b[36m0.17347\u001b[0m       \u001b[32m0.18264\u001b[0m      0.94981      0.93293  28.70s\n",
      "    105       \u001b[36m0.16931\u001b[0m       \u001b[32m0.17742\u001b[0m      0.95429      0.93310  28.68s\n",
      "    106       \u001b[36m0.16887\u001b[0m       0.18125      0.93168      0.93205  28.67s\n",
      "    107       \u001b[36m0.16846\u001b[0m       \u001b[32m0.17325\u001b[0m      0.97236      0.93715  28.69s\n",
      "    108       \u001b[36m0.16708\u001b[0m       \u001b[32m0.17214\u001b[0m      0.97065      0.93660  28.65s\n",
      "    109       \u001b[36m0.16392\u001b[0m       0.17880      0.91679      0.93456  28.68s\n",
      "    110       \u001b[36m0.16095\u001b[0m       0.17793      0.90458      0.93479  28.63s\n",
      "    111       \u001b[36m0.15775\u001b[0m       0.17302      0.91171      0.93632  28.66s\n",
      "    112       \u001b[36m0.15716\u001b[0m       \u001b[32m0.17187\u001b[0m      0.91440      0.93676  28.67s\n",
      "    113       \u001b[36m0.15645\u001b[0m       \u001b[32m0.16608\u001b[0m      0.94202      0.93583  28.67s\n",
      "    114       \u001b[36m0.15293\u001b[0m       0.16682      0.91673      0.93687  28.67s\n",
      "    115       \u001b[36m0.14692\u001b[0m       0.16661      0.88181      0.93779  28.62s\n",
      "    116       0.15175       0.16787      0.90397      0.93742  28.69s\n",
      "    117       \u001b[36m0.14578\u001b[0m       \u001b[32m0.16130\u001b[0m      0.90377      0.94107  28.68s\n",
      "    118       \u001b[36m0.14471\u001b[0m       0.16743      0.86431      0.93840  28.68s\n",
      "    119       0.14636       \u001b[32m0.15498\u001b[0m      0.94434      0.94670  28.64s\n",
      "    120       \u001b[36m0.14144\u001b[0m       \u001b[32m0.15342\u001b[0m      0.92191      0.94599  28.67s\n",
      "    121       0.14440       0.15845      0.91133      0.94216  28.69s\n",
      "    122       \u001b[36m0.13805\u001b[0m       \u001b[32m0.14831\u001b[0m      0.93081      0.94725  28.63s\n",
      "    123       \u001b[36m0.13570\u001b[0m       0.14973      0.90630      0.94687  28.67s\n",
      "    124       0.13791       0.15063      0.91559      0.94769  28.65s\n",
      "    125       \u001b[36m0.13041\u001b[0m       0.15525      0.84004      0.94233  28.66s\n",
      "    126       0.13145       \u001b[32m0.14480\u001b[0m      0.90779      0.94660  28.65s\n",
      "    127       \u001b[36m0.12958\u001b[0m       \u001b[32m0.13795\u001b[0m      0.93935      0.95107  28.68s\n",
      "    128       \u001b[36m0.12670\u001b[0m       0.13853      0.91457      0.94982  28.65s\n",
      "    129       \u001b[36m0.12406\u001b[0m       \u001b[32m0.13761\u001b[0m      0.90158      0.95070  28.64s\n",
      "    130       0.12427       \u001b[32m0.13737\u001b[0m      0.90466      0.95080  28.69s\n",
      "    131       0.12432       0.14248      0.87252      0.94938  28.67s\n",
      "    132       0.12463       \u001b[32m0.13342\u001b[0m      0.93406      0.95425  28.65s\n",
      "    133       \u001b[36m0.11870\u001b[0m       0.13543      0.87642      0.95441  28.67s\n",
      "    134       0.12248       0.13553      0.90373      0.95184  28.65s\n",
      "    135       \u001b[36m0.11645\u001b[0m       0.13519      0.86136      0.95311  28.66s\n",
      "    136       0.11733       0.13385      0.87656      0.95086  28.70s\n",
      "    137       \u001b[36m0.11151\u001b[0m       0.13578      0.82121      0.95464  28.63s\n",
      "    138       0.11201       \u001b[32m0.13223\u001b[0m      0.84708      0.95561  28.65s\n",
      "    139       0.11236       \u001b[32m0.12589\u001b[0m      0.89259      0.95534  28.64s\n",
      "    140       0.11545       0.12785      0.90294      0.95873  28.63s\n",
      "    141       0.11193       \u001b[32m0.11965\u001b[0m      0.93547      0.96020  28.65s\n",
      "    142       \u001b[36m0.10980\u001b[0m       0.12152      0.90355      0.95840  28.71s\n",
      "    143       \u001b[36m0.10567\u001b[0m       0.12071      0.87541      0.95955  28.79s\n",
      "    144       \u001b[36m0.10435\u001b[0m       \u001b[32m0.11716\u001b[0m      0.89069      0.96048  28.63s\n",
      "    145       0.10481       \u001b[32m0.11578\u001b[0m      0.90526      0.96206  28.74s\n",
      "    146       \u001b[36m0.10145\u001b[0m       0.11622      0.87291      0.96152  28.67s\n",
      "    147       0.10305       \u001b[32m0.11545\u001b[0m      0.89261      0.96305  28.63s\n",
      "    148       \u001b[36m0.09778\u001b[0m       \u001b[32m0.11414\u001b[0m      0.85670      0.96108  28.68s\n",
      "    149       0.09965       \u001b[32m0.11010\u001b[0m      0.90505      0.96196  28.64s\n",
      "    150       \u001b[36m0.09392\u001b[0m       0.11486      0.81770      0.96261  28.69s\n",
      "    151       0.09801       0.11023      0.88911      0.96731  28.65s\n",
      "    152       \u001b[36m0.09249\u001b[0m       0.11319      0.81713      0.96326  28.65s\n",
      "    153       \u001b[36m0.09232\u001b[0m       \u001b[32m0.10743\u001b[0m      0.85930      0.96676  28.64s\n",
      "    154       0.09401       0.10909      0.86176      0.96605  28.69s\n",
      "    155       0.09523       0.11610      0.82027      0.96518  28.61s\n",
      "    156       \u001b[36m0.09098\u001b[0m       \u001b[32m0.10646\u001b[0m      0.85461      0.96594  28.64s\n",
      "    157       0.09163       \u001b[32m0.10245\u001b[0m      0.89439      0.96852  28.68s\n",
      "    158       \u001b[36m0.08849\u001b[0m       0.10603      0.83452      0.96874  28.64s\n",
      "    159       0.09365       \u001b[32m0.10245\u001b[0m      0.91413      0.96829  28.64s\n",
      "    160       \u001b[36m0.08397\u001b[0m       0.10718      0.78343      0.96485  28.62s\n",
      "    161       0.08451       \u001b[32m0.09964\u001b[0m      0.84813      0.96879  28.61s\n",
      "    162       0.08768       0.10037      0.87349      0.96977  28.67s\n",
      "    163       0.08458       \u001b[32m0.09487\u001b[0m      0.89156      0.97108  28.60s\n",
      "    164       \u001b[36m0.08247\u001b[0m       0.09801      0.84145      0.96917  28.66s\n",
      "    165       0.08518       0.09936      0.85724      0.97125  28.61s\n",
      "    166       \u001b[36m0.08109\u001b[0m       0.10071      0.80521      0.96938  28.65s\n",
      "    167       \u001b[36m0.07999\u001b[0m       0.09838      0.81303      0.97015  28.60s\n",
      "    168       \u001b[36m0.07782\u001b[0m       \u001b[32m0.09210\u001b[0m      0.84492      0.97283  28.61s\n",
      "    169       0.07903       0.10247      0.77118      0.96829  28.68s\n",
      "    170       \u001b[36m0.07744\u001b[0m       0.09660      0.80169      0.97229  28.60s\n",
      "    171       \u001b[36m0.07629\u001b[0m       0.09321      0.81844      0.97218  28.64s\n",
      "    172       \u001b[36m0.07320\u001b[0m       \u001b[32m0.08773\u001b[0m      0.83435      0.97239  28.63s\n",
      "    173       0.07630       0.09391      0.81250      0.97092  28.69s\n",
      "    174       \u001b[36m0.07271\u001b[0m       0.09479      0.76699      0.97174  28.68s\n",
      "    175       \u001b[36m0.07175\u001b[0m       \u001b[32m0.08498\u001b[0m      0.84436      0.97388  28.71s\n",
      "    176       0.07435       0.09619      0.77299      0.97103  28.68s\n",
      "    177       0.07358       0.08939      0.82314      0.97349  28.70s\n",
      "    178       0.07306       0.09259      0.78902      0.97196  28.69s\n",
      "    179       \u001b[36m0.06702\u001b[0m       0.08530      0.78568      0.97556  28.72s\n",
      "    180       \u001b[36m0.06619\u001b[0m       0.08666      0.76376      0.97409  28.66s\n",
      "    181       0.06855       0.08841      0.77535      0.97606  28.69s\n",
      "    182       \u001b[36m0.06475\u001b[0m       0.09011      0.71852      0.97431  28.67s\n",
      "    183       \u001b[36m0.06443\u001b[0m       0.08612      0.74816      0.97540  28.68s\n",
      "    184       0.06822       0.09111      0.74883      0.97573  28.62s\n",
      "    185       \u001b[36m0.06255\u001b[0m       0.08716      0.71764      0.97731  28.66s\n",
      "    186       0.06693       \u001b[32m0.07989\u001b[0m      0.83780      0.97715  28.65s\n",
      "    187       0.06630       \u001b[32m0.07805\u001b[0m      0.84945      0.97879  28.69s\n",
      "    188       0.06356       0.08581      0.74065      0.97742  28.72s\n",
      "    189       0.06366       0.08403      0.75754      0.97600  28.63s\n",
      "    190       \u001b[36m0.05778\u001b[0m       0.08546      0.67614      0.97819  28.64s\n",
      "    191       0.05905       0.08203      0.71987      0.97759  28.63s\n",
      "    192       0.06100       0.08485      0.71888      0.97606  28.71s\n",
      "    193       0.05905       0.07873      0.75007      0.97813  28.68s\n",
      "    194       \u001b[36m0.05657\u001b[0m       \u001b[32m0.07659\u001b[0m      0.73859      0.98114  28.60s\n",
      "    195       0.05734       0.08381      0.68412      0.97704  28.69s\n",
      "    196       \u001b[36m0.05599\u001b[0m       0.08446      0.66293      0.97753  28.64s\n",
      "    197       0.05703       0.08398      0.67912      0.97829  28.61s\n",
      "    198       0.05932       0.07737      0.76672      0.97961  28.66s\n",
      "    199       \u001b[36m0.05336\u001b[0m       \u001b[32m0.07173\u001b[0m      0.74387      0.98147  28.62s\n",
      "    200       0.05419       0.07929      0.68347      0.97983  28.68s\n",
      "    201       \u001b[36m0.05310\u001b[0m       0.08193      0.64809      0.97747  28.64s\n",
      "    202       0.05514       0.07963      0.69239      0.98005  28.62s\n",
      "    203       0.05576       0.07659      0.72810      0.98114  28.65s\n",
      "    204       0.05348       \u001b[32m0.07021\u001b[0m      0.76165      0.98333  28.62s\n",
      "    205       0.05382       0.08008      0.67205      0.98152  28.68s\n",
      "    206       \u001b[36m0.05113\u001b[0m       0.07805      0.65517      0.98016  28.67s\n",
      "    207       \u001b[36m0.04946\u001b[0m       0.07441      0.66474      0.98108  28.63s\n",
      "    208       0.05220       0.07477      0.69804      0.98288  28.66s\n",
      "    209       \u001b[36m0.04842\u001b[0m       0.08079      0.59933      0.97944  28.62s\n",
      "    210       0.05296       \u001b[32m0.06899\u001b[0m      0.76765      0.98382  28.67s\n",
      "    211       0.05009       0.07623      0.65710      0.98016  28.70s\n",
      "    212       0.04954       0.07706      0.64288      0.98169  28.60s\n",
      "    213       0.04876       0.07513      0.64896      0.98097  28.67s\n",
      "    214       0.04866       0.07596      0.64059      0.98212  28.63s\n",
      "    215       \u001b[36m0.04809\u001b[0m       0.07619      0.63120      0.98305  28.63s\n",
      "    216       \u001b[36m0.04802\u001b[0m       \u001b[32m0.06570\u001b[0m      0.73092      0.98524  28.63s\n",
      "    217       0.04994       0.07447      0.67062      0.98201  28.64s\n",
      "    218       \u001b[36m0.04451\u001b[0m       0.07457      0.59686      0.98245  28.66s\n",
      "    219       0.04656       0.08840      0.52675      0.97665  28.61s\n",
      "    220       0.04650       0.07513      0.61892      0.98196  28.63s\n",
      "    221       0.04461       0.07970      0.55977      0.98081  28.62s\n",
      "    222       0.04555       0.07249      0.62843      0.98305  28.62s\n",
      "    223       \u001b[36m0.04245\u001b[0m       0.07705      0.55090      0.98163  28.64s\n",
      "    224       0.04559       0.06633      0.68731      0.98470  28.72s\n",
      "    225       \u001b[36m0.04195\u001b[0m       0.06837      0.61354      0.98245  28.62s\n",
      "    226       0.04362       0.07498      0.58173      0.98256  28.63s\n",
      "    227       \u001b[36m0.03942\u001b[0m       0.06659      0.59200      0.98387  28.62s\n",
      "    228       0.04186       0.07441      0.56250      0.98305  28.61s\n",
      "    229       0.04039       0.07420      0.54442      0.98442  28.63s\n",
      "    230       \u001b[36m0.03850\u001b[0m       0.06741      0.57110      0.98573  28.60s\n",
      "    231       0.04491       0.06705      0.66981      0.98606  28.59s\n",
      "    232       0.04155       0.06637      0.62599      0.98606  28.65s\n",
      "    233       0.04106       0.07750      0.52982      0.98311  28.65s\n",
      "    234       \u001b[36m0.03829\u001b[0m       0.07088      0.54021      0.98491  28.64s\n",
      "    235       0.03840       0.07275      0.52792      0.98464  28.59s\n",
      "    236       0.04058       0.07397      0.54864      0.98464  28.62s\n",
      "    237       0.04028       0.07516      0.53597      0.98196  28.58s\n",
      "    238       0.03886       0.07409      0.52445      0.98327  28.64s\n",
      "    239       \u001b[36m0.03486\u001b[0m       0.07109      0.49030      0.98447  28.56s\n",
      "    240       0.03718       0.06655      0.55865      0.98398  28.64s\n",
      "    241       0.03689       0.07373      0.50028      0.98589  28.63s\n",
      "    242       0.03738       \u001b[32m0.06519\u001b[0m      0.57338      0.98655  28.59s\n",
      "    243       0.03661       0.07007      0.52248      0.98507  28.65s\n",
      "    244       0.03970       0.07239      0.54843      0.98497  28.59s\n",
      "    245       0.03760       0.06599      0.56975      0.98425  28.60s\n",
      "    246       \u001b[36m0.03448\u001b[0m       0.07075      0.48728      0.98431  28.61s\n",
      "    247       0.03733       0.07055      0.52906      0.98573  28.60s\n",
      "    248       0.03637       0.07696      0.47258      0.98081  28.90s\n",
      "    249       0.03663       0.07337      0.49926      0.98497  28.91s\n",
      "    250       0.03458       0.06628      0.52177      0.98622  28.69s\n",
      "    251       0.03514       0.06841      0.51371      0.98458  28.82s\n",
      "    252       \u001b[36m0.03375\u001b[0m       0.07139      0.47272      0.98425  28.81s\n",
      "    253       0.03499       0.07994      0.43771      0.98316  28.77s\n",
      "    254       0.03767       0.06630      0.56816      0.98775  28.89s\n",
      "    255       \u001b[36m0.03348\u001b[0m       0.07149      0.46825      0.98529  28.92s\n",
      "    256       \u001b[36m0.03317\u001b[0m       0.06942      0.47780      0.98546  28.95s\n",
      "    257       0.03460       0.07512      0.46062      0.98611  28.92s\n",
      "    258       0.03406       0.06626      0.51408      0.98737  28.93s\n",
      "    259       0.03392       0.06618      0.51254      0.98513  28.72s\n",
      "    260       \u001b[36m0.03164\u001b[0m       0.06812      0.46454      0.98540  28.61s\n",
      "    261       \u001b[36m0.03016\u001b[0m       0.07522      0.40092      0.98475  28.55s\n",
      "    262       0.03329       0.06897      0.48270      0.98661  28.61s\n",
      "    263       0.03235       0.06942      0.46595      0.98562  28.60s\n",
      "    264       0.03438       0.07100      0.48418      0.98589  28.63s\n",
      "    265       0.03093       0.06722      0.46017      0.98644  28.60s\n",
      "    266       0.03168       0.06894      0.45953      0.98655  28.65s\n",
      "    267       \u001b[36m0.02980\u001b[0m       0.07301      0.40812      0.98573  28.60s\n",
      "    268       0.03225       0.07183      0.44894      0.98475  28.62s\n",
      "    269       0.03120       0.08271      0.37715      0.98475  28.60s\n",
      "    270       \u001b[36m0.02820\u001b[0m       0.07209      0.39121      0.98415  28.69s\n",
      "    271       0.03103       0.07695      0.40320      0.98524  28.66s\n",
      "    272       0.03372       0.07109      0.47433      0.98579  28.70s\n",
      "    273       0.03006       \u001b[32m0.06217\u001b[0m      0.48359      0.98606  28.73s\n",
      "    274       0.03017       0.07216      0.41810      0.98753  28.74s\n",
      "    275       \u001b[36m0.02707\u001b[0m       0.07606      0.35593      0.98497  28.83s\n",
      "    276       0.03192       0.07108      0.44912      0.98655  28.88s\n",
      "    277       0.02904       0.06451      0.45016      0.98808  28.81s\n",
      "    278       0.03023       0.06846      0.44160      0.98497  28.93s\n",
      "    279       0.02844       0.07098      0.40064      0.98552  28.77s\n",
      "    280       0.02982       0.07197      0.41431      0.98748  28.71s\n",
      "    281       0.02941       0.06550      0.44900      0.98726  28.69s\n",
      "    282       0.02814       0.07431      0.37868      0.98693  28.77s\n",
      "    283       \u001b[36m0.02472\u001b[0m       0.07023      0.35202      0.98556  28.64s\n",
      "    284       0.03171       0.06304      0.50298      0.98611  28.67s\n",
      "    285       0.02979       0.07468      0.39888      0.98573  28.66s\n",
      "    286       0.02871       0.06727      0.42680      0.98704  28.66s\n",
      "    287       0.02802       0.06681      0.41931      0.98765  28.78s\n",
      "    288       0.02504       0.06447      0.38844      0.98644  28.66s\n",
      "    289       0.02676       0.07276      0.36779      0.98606  28.62s\n",
      "    290       0.02724       0.07092      0.38416      0.98556  28.70s\n",
      "    291       0.02534       0.06584      0.38491      0.98638  28.60s\n",
      "    292       0.02710       0.06983      0.38801      0.98464  28.71s\n",
      "    293       0.02487       0.06923      0.35922      0.98644  28.63s\n",
      "    294       \u001b[36m0.02391\u001b[0m       0.06538      0.36579      0.98798  28.76s\n",
      "    295       0.02562       0.07060      0.36288      0.98622  28.76s\n",
      "    296       0.02666       0.06668      0.39987      0.98693  28.74s\n",
      "    297       0.02482       0.06513      0.38115      0.98743  28.75s\n",
      "    298       0.02672       0.06807      0.39252      0.98688  28.59s\n",
      "    299       0.02620       0.06384      0.41036      0.98868  28.63s\n",
      "    300       0.02572       0.06723      0.38259      0.98825  28.68s\n",
      "    301       0.02512       0.07194      0.34918      0.98661  28.61s\n",
      "    302       0.02405       0.07382      0.32586      0.98579  28.62s\n",
      "    303       0.02636       0.07486      0.35217      0.98589  28.64s\n",
      "    304       0.02674       0.06907      0.38714      0.98737  28.58s\n",
      "    305       \u001b[36m0.02259\u001b[0m       0.06601      0.34226      0.98792  28.62s\n",
      "    306       0.02374       0.07650      0.31029      0.98507  28.55s\n",
      "    307       0.02555       0.06976      0.36629      0.98852  28.62s\n",
      "    308       \u001b[36m0.02203\u001b[0m       0.06522      0.33772      0.98988  28.63s\n",
      "    309       0.02454       0.07355      0.33371      0.98655  28.55s\n",
      "    310       0.02409       0.07275      0.33109      0.98638  28.66s\n",
      "    311       0.02492       0.07831      0.31823      0.98556  28.62s\n",
      "    312       0.02263       0.07351      0.30783      0.98710  28.59s\n",
      "    313       0.02437       0.06321      0.38548      0.98765  28.63s\n",
      "    314       \u001b[36m0.02042\u001b[0m       0.07057      0.28940      0.98693  28.63s\n",
      "    315       0.02049       0.06850      0.29919      0.98770  28.73s\n",
      "    316       \u001b[36m0.01936\u001b[0m       0.07230      0.26777      0.98770  30.31s\n",
      "    317       0.02160       0.07606      0.28396      0.98655  30.05s\n",
      "    318       0.02028       0.07107      0.28539      0.98704  29.32s\n",
      "    319       0.02202       0.06509      0.33833      0.98704  28.73s\n",
      "    320       0.02091       0.07973      0.26223      0.98638  28.66s\n",
      "    321       0.02023       0.06947      0.29121      0.98841  29.17s\n",
      "    322       0.02143       0.06904      0.31046      0.98732  28.54s\n",
      "    323       0.02296       0.07154      0.32089      0.98825  29.18s\n",
      "Early stopping.\n",
      "Best valid loss was 0.062167 at epoch 273.\n",
      "Loaded parameters to layer 'image_conv1' (shape 16x1x13x13).\n",
      "Loaded parameters to layer 'image_conv1' (shape 16).\n",
      "Loaded parameters to layer 'image_conv2' (shape 16x16x13x13).\n",
      "Loaded parameters to layer 'image_conv2' (shape 16).\n",
      "Loaded parameters to layer 'prob_conv1' (shape 16x1x13x13).\n",
      "Loaded parameters to layer 'prob_conv1' (shape 16).\n",
      "Loaded parameters to layer 'prob_conv2' (shape 16x16x13x13).\n",
      "Loaded parameters to layer 'prob_conv2' (shape 16).\n",
      "Loaded parameters to layer 'binary_conv1' (shape 16x1x13x13).\n",
      "Loaded parameters to layer 'binary_conv1' (shape 16).\n",
      "Loaded parameters to layer 'binary_conv2' (shape 16x16x13x13).\n",
      "Loaded parameters to layer 'binary_conv2' (shape 16).\n",
      "Loaded parameters to layer 'border_conv1' (shape 16x1x13x13).\n",
      "Loaded parameters to layer 'border_conv1' (shape 16).\n",
      "Loaded parameters to layer 'border_conv2' (shape 16x16x13x13).\n",
      "Loaded parameters to layer 'border_conv2' (shape 16).\n",
      "Loaded parameters to layer 'hidden3' (shape 5184x256).\n",
      "Loaded parameters to layer 'hidden3' (shape 256).\n",
      "Loaded parameters to layer 'output' (shape 256x2).\n",
      "Loaded parameters to layer 'output' (shape 2).\n"
     ]
    }
   ],
   "source": [
    "cnn = cnn.fit({'image_input': X_train['image'],\n",
    "               'prob_input': X_train['prob'],\n",
    "               'binary_input': X_train['merged_array'],\n",
    "               'border_input': X_train['larger_border_overlap']},\n",
    "               y_train)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 5,
   "metadata": {
    "collapsed": false
   },
   "outputs": [
    {
     "data": {
      "text/plain": [
       "0.8646472392638037"
      ]
     },
     "execution_count": 5,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "# this is the score on cylinder_small1 with regularization and patience counter 50 and data augmentation\n",
    "cnn.score({'image_input': X_test['image'],\n",
    "           'prob_input': X_test['prob'],\n",
    "           'binary_input': X_test['merged_array'],\n",
    "           'border_input': X_test['larger_border_overlap']},\n",
    "           y_test)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 6,
   "metadata": {
    "collapsed": false
   },
   "outputs": [
    {
     "data": {
      "text/plain": [
       "<module 'matplotlib.pyplot' from '/home/d/nolearn/local/lib/python2.7/site-packages/matplotlib/pyplot.pyc'>"
      ]
     },
     "execution_count": 6,
     "metadata": {},
     "output_type": "execute_result"
    },
    {
     "data": {
      "image/png": "iVBORw0KGgoAAAANSUhEUgAAAYkAAAEPCAYAAAC3NDh4AAAABHNCSVQICAgIfAhkiAAAAAlwSFlz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\nUEoplSxNEkoppZL1/wktwVcnNXTvAAAAAElFTkSuQmCC\n",
      "text/plain": [
       "<matplotlib.figure.Figure at 0x7f48a84ccc90>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    }
   ],
   "source": [
    "# this is the plot on cylinder_small1 with regularization and patience counter 50 and data augmentation\n",
    "plot_loss(cnn)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 7,
   "metadata": {
    "collapsed": false
   },
   "outputs": [
    {
     "data": {
      "text/plain": [
       "0.8186349693251533"
      ]
     },
     "execution_count": 7,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "# this is the score on cylinder_small1 with regularization and patience counter 50\n",
    "cnn.score({'image_input': X_test['image'],\n",
    "           'prob_input': X_test['prob'],\n",
    "           'binary_input': X_test['merged_array'],\n",
    "           'border_input': X_test['larger_border_overlap']},\n",
    "           y_test)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 9,
   "metadata": {
    "collapsed": false
   },
   "outputs": [
    {
     "data": {
      "text/plain": [
       "<module 'matplotlib.pyplot' from '/home/d/nolearn/local/lib/python2.7/site-packages/matplotlib/pyplot.pyc'>"
      ]
     },
     "execution_count": 9,
     "metadata": {},
     "output_type": "execute_result"
    },
    {
     "data": {
      "image/png": "iVBORw0KGgoAAAANSUhEUgAAAYkAAAEPCAYAAAC3NDh4AAAABHNCSVQICAgIfAhkiAAAAAlwSFlz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\n",
      "text/plain": [
       "<matplotlib.figure.Figure at 0x7febcc158dd0>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    }
   ],
   "source": [
    "# this is the plot on cylinder_small1 with regularization and patience counter 50\n",
    "plot_loss(cnn)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 6,
   "metadata": {
    "collapsed": false
   },
   "outputs": [
    {
     "data": {
      "text/plain": [
       "0.8197852760736196"
      ]
     },
     "execution_count": 6,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "# this is the score on cylinder_small1 with regularization\n",
    "cnn.score({'image_input': X_test['image'],\n",
    "           'prob_input': X_test['prob'],\n",
    "           'binary_input': X_test['merged_array'],\n",
    "           'border_input': X_test['larger_border_overlap']},\n",
    "           y_test)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 7,
   "metadata": {
    "collapsed": false
   },
   "outputs": [
    {
     "data": {
      "text/plain": [
       "<module 'matplotlib.pyplot' from '/home/d/nolearn/local/lib/python2.7/site-packages/matplotlib/pyplot.pyc'>"
      ]
     },
     "execution_count": 7,
     "metadata": {},
     "output_type": "execute_result"
    },
    {
     "data": {
      "image/png": "iVBORw0KGgoAAAANSUhEUgAAAYkAAAEPCAYAAAC3NDh4AAAABHNCSVQICAgIfAhkiAAAAAlwSFlz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\nAElFTkSuQmCC\n",
      "text/plain": [
       "<matplotlib.figure.Figure at 0x7fef8c7d5510>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    }
   ],
   "source": [
    "# this is the plot on cylinder_small1 with regularization\n",
    "plot_loss(cnn)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 5,
   "metadata": {
    "collapsed": false,
    "scrolled": true
   },
   "outputs": [
    {
     "data": {
      "text/plain": [
       "0.8186349693251533"
      ]
     },
     "execution_count": 5,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "# this is the score on cylinder_small1 without regularization\n",
    "cnn.score({'image_input': X_test['image'],\n",
    "           'prob_input': X_test['prob'],\n",
    "           'binary_input': X_test['merged_array'],\n",
    "           'border_input': X_test['larger_border_overlap']},\n",
    "           y_test)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 6,
   "metadata": {
    "collapsed": false,
    "scrolled": true
   },
   "outputs": [
    {
     "data": {
      "text/plain": [
       "<module 'matplotlib.pyplot' from '/home/d/nolearn/local/lib/python2.7/site-packages/matplotlib/pyplot.pyc'>"
      ]
     },
     "execution_count": 6,
     "metadata": {},
     "output_type": "execute_result"
    },
    {
     "data": {
      "image/png": "iVBORw0KGgoAAAANSUhEUgAAAYkAAAEPCAYAAAC3NDh4AAAABHNCSVQICAgIfAhkiAAAAAlwSFlz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\nJJRSSqXo/wEF8GIsvKAbqgAAAABJRU5ErkJggg==\n",
      "text/plain": [
       "<matplotlib.figure.Figure at 0x7f01a6f43990>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    }
   ],
   "source": [
    "# this is the plot on cylinder_small1 without regularization\n",
    "plot_loss(cnn)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {
    "collapsed": true
   },
   "outputs": [],
   "source": []
  }
 ],
 "metadata": {
  "kernelspec": {
   "display_name": "Python 2",
   "language": "python",
   "name": "python2"
  },
  "language_info": {
   "codemirror_mode": {
    "name": "ipython",
    "version": 2
   },
   "file_extension": ".py",
   "mimetype": "text/x-python",
   "name": "python",
   "nbconvert_exporter": "python",
   "pygments_lexer": "ipython2",
   "version": "2.7.6"
  }
 },
 "nbformat": 4,
 "nbformat_minor": 0
}