sk-autoencoder3-windows.ipynb

{
 "cells": [
  {
   "cell_type": "code",
   "execution_count": 1,
   "metadata": {
    "collapsed": false
   },
   "outputs": [],
   "source": [
    "import json\n",
    "import glob\n",
    "import numpy as np\n",
    "import matplotlib.pyplot as plt\n",
    "\n",
    "\n",
    "%matplotlib inline\n",
    "\n",
    "import keras\n",
    "from keras.models import Sequential\n",
    "from keras.layers.core import Dense, Dropout, Activation\n",
    "from keras import backend as K\n",
    "\n",
    "from IPython import display\n",
    "\n"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 2,
   "metadata": {
    "collapsed": false
   },
   "outputs": [],
   "source": [
    "LAYER_BLOCKS = 0\n",
    "LAYER_COINS = 1\n",
    "LAYER_ENEMIES = 2\n",
    "NLAYERS = 3\n",
    "\n",
    "class GameMap:\n",
    "    def __init__(self):\n",
    "        self.levels = []\n",
    "        self.widths = []\n",
    "        self.heights = []\n",
    "        self.nmaps = 0\n",
    "        \n",
    "        with open(\"Zelda/zelda.json\") as f:\n",
    "            self.tiledefs = json.load(f)\n",
    "            #print(self.tiledefs)\n",
    "        \n",
    "        for fname in sorted(glob.glob(\"Zelda/zelda/zelda_lvl*.txt\")):\n",
    "            #print(fname)\n",
    "            with open(fname) as f:\n",
    "                data = [line.strip() for line in f.readlines()]\n",
    "                self.levels.append(data)\n",
    "                self.heights.append(len(data))\n",
    "                self.widths.append(len(data[0]))\n",
    "                self.nmaps += 1\n",
    "\n",
    "    def getMaps(self, level=0):\n",
    "        layers = np.ndarray((self.heights[level], self.widths[level], NLAYERS), dtype=np.bool)\n",
    "        layers.fill(0)\n",
    "        for r, row in enumerate(self.levels[level]):\n",
    "            for c, tile in enumerate(row):\n",
    "                if tile in ('1','2','3'):\n",
    "                    layers[r,c,LAYER_ENEMIES] = 1\n",
    "                elif tile in ('+'):\n",
    "                    layers[r,c,LAYER_COINS] = 1\n",
    "                elif tile in ('w'):\n",
    "                    pass\n",
    "                else:\n",
    "                    layers[r,c,LAYER_BLOCKS] = 1\n",
    "                \n",
    "        return layers[:,:,LAYER_BLOCKS]\n",
    "    \n",
    "    def getMapsRange(self, from_level=0, to_level=None):\n",
    "        if to_level == None:\n",
    "            to_level = self.nmaps\n",
    "        return np.concatenate([self.getMaps(i) for i in range(from_level, to_level)], axis=1)\n",
    "            \n",
    "    def getMapsRangeWindowed(self, windowsize = 2, from_level=0, to_level=None):\n",
    "        a = self.getMapsRange(from_level, to_level)\n",
    "        windows = np.concatenate([np.roll(a,-shift,axis=1) for shift in range(windowsize)])\n",
    "        return windows\n",
    "            "
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 3,
   "metadata": {
    "collapsed": false
   },
   "outputs": [],
   "source": [
    "gm = GameMap()\n",
    "a = gm.getMapsRangeWindowed(4,0,5)\n",
    "at = a.transpose()\n",
    "\n",
    "\n",
    "#print(json.dumps(gm.tiledefs, indent=2))"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 4,
   "metadata": {
    "collapsed": false
   },
   "outputs": [
    {
     "data": {
      "text/plain": [
       "(0.0, 100.0)"
      ]
     },
     "execution_count": 4,
     "metadata": {},
     "output_type": "execute_result"
    },
    {
     "data": {
      "image/png": "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",
      "text/plain": [
       "<Figure size 1440x432 with 1 Axes>"
      ]
     },
     "metadata": {
      "needs_background": "light"
     },
     "output_type": "display_data"
    }
   ],
   "source": [
    "fig, axes = plt.subplots(1,1,figsize=(20,6))\n",
    "axes.imshow(at.transpose(),cmap=\"gray\", interpolation=\"nearest\")\n",
    "axes.set_xlim(0,100)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 31,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "36\n",
      "[[False False False ...  True  True False]\n",
      " [False  True  True ... False False False]\n",
      " [False False  True ...  True  True False]\n",
      " ...\n",
      " [False  True  True ... False False False]\n",
      " [False  True  True ...  True  True False]\n",
      " [False False False ...  True  True False]]\n"
     ]
    }
   ],
   "source": [
    "print(at.shape[1])\r\n",
    "print(at)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 5,
   "metadata": {
    "collapsed": false
   },
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "36\n"
     ]
    }
   ],
   "source": [
    "in_out_size = at.shape[1]\n",
    "print(in_out_size)\n",
    "hiddensize =2 #int(in_out_size*.25)\n",
    "model = Sequential()\n",
    "model.add(Dense(hiddensize, input_shape=(in_out_size,)))\n",
    "model.add(Activation(\"relu\"))\n",
    "model.add(Dropout(0))\n",
    "model.add(Dense(in_out_size))\n",
    "# leaving off this final relu seems to do a much better job\n",
    "#model.add(Activation(\"relu\"))\n",
    "model.add(Activation(\"sigmoid\"))\n",
    "\n",
    "# \"mse\" seems to be the best loss function for the problem\n",
    "# \"RMSprop\" seems to be the best, with \"adadelta\" also good\n",
    "model.compile(loss='mse', optimizer='RMSprop', metrics=['accuracy'])\n",
    "\n"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 6,
   "metadata": {
    "collapsed": true
   },
   "outputs": [],
   "source": [
    "in_out_size = at.shape[1]\n",
    "model = Sequential()\n",
    "model.add(Dense(hiddensize, input_shape=(in_out_size,)))\n",
    "model.add(Activation(\"relu\"))\n",
    "\n",
    "model.add(Dense(int(in_out_size/2)))\n",
    "model.add(Activation(\"relu\"))\n",
    "\n",
    "model.add(Dense(int(2)))\n",
    "model.add(Activation(\"relu\"))\n",
    "\n",
    "model.add(Dense(int(in_out_size/2)))\n",
    "model.add(Activation(\"relu\"))\n",
    "\n",
    "model.add(Dense(in_out_size))\n",
    "# leaving off this final relu seems to do a much better job\n",
    "#model.add(Activation(\"relu\"))\n",
    "model.add(Activation(\"sigmoid\"))\n",
    "\n",
    "# \"mse\" seems to be the best loss function for the problem\n",
    "# \"RMSprop\" seems to be the best, with \"adadelta\" also good\n",
    "model.compile(loss='mse', optimizer='RMSprop', metrics=['accuracy'])\n",
    "\n"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 7,
   "metadata": {
    "collapsed": false
   },
   "outputs": [],
   "source": [
    "in_out_size = at.shape[1]\n",
    "dropout = 0\n",
    "model = Sequential()\n",
    "model.add(Dense( int(in_out_size*.9), input_shape=(in_out_size,)))\n",
    "model.add(Activation(\"relu\"))\n",
    "model.add(Dropout(dropout))\n",
    "\n",
    "model.add(Dense(int(in_out_size*.8)))\n",
    "model.add(Activation(\"relu\"))\n",
    "model.add(Dropout(dropout))\n",
    "\n",
    "\n",
    "model.add(Dense(int(in_out_size*.5)))\n",
    "model.add(Activation(\"relu\"))\n",
    "model.add(Dropout(dropout))\n",
    "\n",
    "model.add(Dense(int(in_out_size*.8)))\n",
    "model.add(Activation(\"relu\"))\n",
    "model.add(Dropout(dropout))\n",
    "\n",
    "\n",
    "model.add(Dense(int(in_out_size*.9)))\n",
    "model.add(Activation(\"relu\"))\n",
    "model.add(Dropout(dropout))\n",
    "\n",
    "model.add(Dense(in_out_size))\n",
    "\n",
    "# leaving off this final relu seems to do a much better job.  I think you don't want\n",
    "# your final activation to be something with hard cutoffs.\n",
    "#model.add(Activation(\"relu\"))\n",
    "model.add(Activation(\"sigmoid\"))\n",
    "\n",
    "# \"mse\" seems to be the best loss function for the problem\n",
    "# \"RMSprop\" seems to be the best, with \"adadelta\" also good\n",
    "model.compile(loss='mse', optimizer='RMSprop', metrics=['accuracy'])\n"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 8,
   "metadata": {
    "collapsed": true
   },
   "outputs": [],
   "source": [
    "in_out_size = at.shape[1]\n",
    "dropout = 0\n",
    "model = Sequential()\n",
    "model.add(Dense( int(in_out_size*1.5), input_shape=(in_out_size,)))\n",
    "model.add(Activation(\"relu\"))\n",
    "\n",
    "model.add(Dense(int(in_out_size*.8)))\n",
    "model.add(Activation(\"relu\"))\n",
    "\n",
    "model.add(Dense(int(in_out_size*.5)))\n",
    "model.add(Activation(\"relu\"))\n",
    "\n",
    "model.add(Dense(int(in_out_size*.25)))\n",
    "model.add(Activation(\"relu\"))\n",
    "\n",
    "model.add(Dense(int(8)))\n",
    "model.add(Activation(\"relu\"))\n",
    "\n",
    "model.add(Dense(int(4)))\n",
    "model.add(Activation(\"relu\"))\n",
    "\n",
    "model.add(Dense(int(2)))\n",
    "model.add(Activation(\"relu\"))\n",
    "\n",
    "model.add(Dense(int(4)))\n",
    "model.add(Activation(\"relu\"))\n",
    "\n",
    "\n",
    "model.add(Dense(int(8)))\n",
    "model.add(Activation(\"relu\"))\n",
    "\n",
    "\n",
    "model.add(Dense(int(in_out_size*.5)))\n",
    "model.add(Activation(\"relu\"))\n",
    "\n",
    "model.add(Dense(int(in_out_size*.8)))\n",
    "model.add(Activation(\"relu\"))\n",
    "\n",
    "model.add(Dense(int(in_out_size*1.5)))\n",
    "model.add(Activation(\"relu\"))\n",
    "\n",
    "model.add(Dense(in_out_size))\n",
    "\n",
    "# leaving off this final relu seems to do a much better job.  I think you don't want\n",
    "# your final activation to be something with hard cutoffs.\n",
    "#model.add(Activation(\"relu\"))\n",
    "model.add(Activation(\"sigmoid\"))\n",
    "\n",
    "# \"mse\" seems to be the best loss function for the problem\n",
    "# \"RMSprop\" seems to be the best, with \"adadelta\" also good\n",
    "model.compile(loss='mse', optimizer='adadelta', metrics=['accuracy'])\n"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 9,
   "metadata": {
    "collapsed": false
   },
   "outputs": [
    {
     "data": {
      "text/plain": [
       "[<keras.layers.core.Dense at 0x19d35b98670>,\n",
       " <keras.layers.core.Activation at 0x19d4b3b85e0>,\n",
       " <keras.layers.core.Dense at 0x19d356c9400>,\n",
       " <keras.layers.core.Activation at 0x19d356f34f0>,\n",
       " <keras.layers.core.Dense at 0x19d35bd0640>,\n",
       " <keras.layers.core.Activation at 0x19d35b78a60>,\n",
       " <keras.layers.core.Dense at 0x19d35b25c70>,\n",
       " <keras.layers.core.Activation at 0x19d35b6ed00>,\n",
       " <keras.layers.core.Dense at 0x19d35bdc2e0>,\n",
       " <keras.layers.core.Activation at 0x19d35bdc310>,\n",
       " <keras.layers.core.Dense at 0x19d35bb99d0>,\n",
       " <keras.layers.core.Activation at 0x19d35bd83d0>,\n",
       " <keras.layers.core.Dense at 0x19d35be6d30>,\n",
       " <keras.layers.core.Activation at 0x19d35be25e0>,\n",
       " <keras.layers.core.Dense at 0x19d35bdc1c0>,\n",
       " <keras.layers.core.Activation at 0x19d35be2e80>,\n",
       " <keras.layers.core.Dense at 0x19d35bf2f70>,\n",
       " <keras.layers.core.Activation at 0x19d35bf2160>,\n",
       " <keras.layers.core.Dense at 0x19d35beb340>,\n",
       " <keras.layers.core.Activation at 0x19d35bffd60>,\n",
       " <keras.layers.core.Dense at 0x19d35be2a00>,\n",
       " <keras.layers.core.Activation at 0x19d35c0a0a0>,\n",
       " <keras.layers.core.Dense at 0x19d35c0a250>,\n",
       " <keras.layers.core.Activation at 0x19d35c0a790>,\n",
       " <keras.layers.core.Dense at 0x19d35c1b940>,\n",
       " <keras.layers.core.Activation at 0x19d35c1b730>]"
      ]
     },
     "execution_count": 9,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "model.layers"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 10,
   "metadata": {
    "collapsed": false
   },
   "outputs": [],
   "source": [
    "get_hidden_layer_output = K.function([model.layers[0].input],\n",
    "                                  [model.layers[1].output])\n",
    "test_a = gm.getMapsRangeWindowed(4, 0, 1)\n",
    "test_at = test_a.transpose()\n",
    "\n",
    "hidden_layer_output = get_hidden_layer_output([test_at])[0]\n",
    "\n"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 32,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "[[False False False False False False False False False False  True  True\n",
      "   True  True  True  True  True False False  True  True False  True  True\n",
      "   True  True False False  True  True  True False  True  True  True False]\n",
      " [False  True  True  True  True  True  True  True False False  True  True\n",
      "  False  True  True  True  True False False  True  True  True False  True\n",
      "   True  True False False  True False  True False  True  True  True False]\n",
      " [False  True  True False  True  True  True  True False False  True  True\n",
      "   True False  True  True  True False False  True False  True False  True\n",
      "   True  True False False False False False False  True False False False]\n",
      " [False  True  True  True False  True  True  True False False  True False\n",
      "   True False  True  True  True False False False False False False  True\n",
      "  False False False False  True False False False  True  True False False]\n",
      " [False  True False  True False  True  True  True False False False False\n",
      "  False False  True False False False False  True False False False  True\n",
      "   True False False False  True  True False False False  True  True False]\n",
      " [False False False False False  True False False False False  True False\n",
      "  False False  True  True False False False  True  True False False False\n",
      "   True  True False False  True  True  True False  True  True  True False]\n",
      " [False  True False False False  True  True False False False  True  True\n",
      "  False False False  True  True False False  True  True  True False  True\n",
      "   True  True False False  True  True  True False  True False  True False]\n",
      " [False  True  True False False False  True  True False False  True  True\n",
      "   True False  True  True  True False False  True  True  True False  True\n",
      "  False  True False False  True  True  True  True  True  True False False]\n",
      " [False  True  True  True False  True  True  True False False  True  True\n",
      "   True False  True False  True False False  True  True  True  True  True\n",
      "   True False False False  True  True  True  True  True  True  True False]\n",
      " [False  True  True  True False  True False  True False False  True  True\n",
      "   True  True  True  True False False False  True  True  True  True  True\n",
      "   True  True False False False False False False False False False False]\n",
      " [False  True  True  True  True  True  True False False False  True  True\n",
      "   True  True  True  True  True False False False False False False False\n",
      "  False False False False False False False False False False False False]\n",
      " [False  True  True  True  True  True  True  True False False False False\n",
      "  False False False False False False False False False False False False\n",
      "  False False False False  True  True  True  True  True  True  True False]\n",
      " [False False False False False False False False False False False False\n",
      "  False False False False False False False  True  True  True  True  True\n",
      "   True  True False False  True  True False  True  True  True  True False]]\n"
     ]
    }
   ],
   "source": [
    "print(test_at)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 11,
   "metadata": {
    "collapsed": false
   },
   "outputs": [
    {
     "data": {
      "text/plain": [
       "array([[0.00000000e+00, 6.29661560e-01, 8.20247978e-02, 0.00000000e+00,\n",
       "        8.67218018e-01, 3.92732441e-01, 0.00000000e+00, 0.00000000e+00,\n",
       "        0.00000000e+00, 9.33991671e-01, 7.28970647e-01, 4.41427469e-01,\n",
       "        0.00000000e+00, 0.00000000e+00, 5.55253208e-01, 1.12505007e+00,\n",
       "        9.98876810e-01, 0.00000000e+00, 5.26795268e-01, 3.36838305e-01,\n",
       "        0.00000000e+00, 5.31002164e-01, 0.00000000e+00, 0.00000000e+00,\n",
       "        4.07137662e-01, 0.00000000e+00, 0.00000000e+00, 1.21322000e+00,\n",
       "        0.00000000e+00, 1.73623443e-01, 0.00000000e+00, 9.58450437e-01,\n",
       "        4.62044358e-01, 0.00000000e+00, 0.00000000e+00, 0.00000000e+00,\n",
       "        7.64807940e-01, 0.00000000e+00, 0.00000000e+00, 0.00000000e+00,\n",
       "        0.00000000e+00, 0.00000000e+00, 1.39812863e+00, 2.04116750e+00,\n",
       "        0.00000000e+00, 9.18872356e-01, 1.30832508e-01, 5.01152039e-01,\n",
       "        6.99465752e-01, 0.00000000e+00, 0.00000000e+00, 7.04111338e-01,\n",
       "        4.42569852e-01, 0.00000000e+00],\n",
       "       [0.00000000e+00, 1.75657660e-01, 0.00000000e+00, 0.00000000e+00,\n",
       "        3.80614311e-01, 2.80672967e-01, 5.25690615e-01, 0.00000000e+00,\n",
       "        0.00000000e+00, 6.44190252e-01, 6.17616415e-01, 0.00000000e+00,\n",
       "        0.00000000e+00, 3.88854682e-01, 1.40634924e-01, 1.23450863e+00,\n",
       "        1.13346279e+00, 1.81651890e-01, 0.00000000e+00, 0.00000000e+00,\n",
       "        3.06382209e-01, 0.00000000e+00, 0.00000000e+00, 0.00000000e+00,\n",
       "        0.00000000e+00, 3.09519112e-01, 0.00000000e+00, 1.90632248e+00,\n",
       "        0.00000000e+00, 5.14630795e-01, 0.00000000e+00, 1.50877440e+00,\n",
       "        0.00000000e+00, 1.53992623e-02, 0.00000000e+00, 0.00000000e+00,\n",
       "        6.47827387e-02, 0.00000000e+00, 2.85287738e-01, 0.00000000e+00,\n",
       "        0.00000000e+00, 7.79804468e-01, 1.13726020e+00, 1.82974160e+00,\n",
       "        0.00000000e+00, 1.20032287e+00, 0.00000000e+00, 0.00000000e+00,\n",
       "        6.43562675e-01, 3.13750148e-01, 0.00000000e+00, 1.13391924e+00,\n",
       "        0.00000000e+00, 0.00000000e+00],\n",
       "       [0.00000000e+00, 7.91836858e-01, 1.19256243e-01, 0.00000000e+00,\n",
       "        5.46662807e-01, 3.15863550e-01, 6.84346080e-01, 0.00000000e+00,\n",
       "        0.00000000e+00, 8.46215010e-01, 5.64771771e-01, 4.68982637e-01,\n",
       "        0.00000000e+00, 6.40471399e-01, 1.76141113e-01, 1.11795664e+00,\n",
       "        8.67872834e-01, 5.63973784e-02, 5.26785553e-02, 0.00000000e+00,\n",
       "        0.00000000e+00, 6.62976444e-01, 0.00000000e+00, 0.00000000e+00,\n",
       "        1.59573331e-01, 3.69414777e-01, 0.00000000e+00, 1.29085243e+00,\n",
       "        0.00000000e+00, 0.00000000e+00, 0.00000000e+00, 6.40304565e-01,\n",
       "        4.98844892e-01, 6.86550081e-01, 2.19217956e-01, 0.00000000e+00,\n",
       "        4.59195018e-01, 3.34054232e-03, 1.97112784e-01, 0.00000000e+00,\n",
       "        0.00000000e+00, 0.00000000e+00, 1.59675753e+00, 9.86763716e-01,\n",
       "        0.00000000e+00, 1.51744890e+00, 0.00000000e+00, 0.00000000e+00,\n",
       "        7.57444322e-01, 0.00000000e+00, 0.00000000e+00, 7.66305864e-01,\n",
       "        1.25271142e-01, 0.00000000e+00],\n",
       "       [0.00000000e+00, 2.66179889e-01, 5.46941608e-02, 2.11147204e-01,\n",
       "        3.93261671e-01, 2.75158882e-01, 1.84822217e-01, 4.29471880e-02,\n",
       "        0.00000000e+00, 1.17916262e+00, 7.13046193e-01, 0.00000000e+00,\n",
       "        0.00000000e+00, 2.84137011e-01, 4.48640764e-01, 1.53623223e+00,\n",
       "        8.61531496e-01, 0.00000000e+00, 0.00000000e+00, 5.37460566e-01,\n",
       "        6.39293611e-01, 2.06456378e-01, 0.00000000e+00, 0.00000000e+00,\n",
       "        5.93818426e-02, 1.70146793e-01, 0.00000000e+00, 9.83052135e-01,\n",
       "        0.00000000e+00, 7.61152744e-01, 0.00000000e+00, 1.18042600e+00,\n",
       "        2.85672188e-01, 0.00000000e+00, 4.40911829e-01, 0.00000000e+00,\n",
       "        6.58878684e-01, 0.00000000e+00, 7.32393116e-02, 0.00000000e+00,\n",
       "        0.00000000e+00, 6.05115443e-02, 1.44276512e+00, 1.43650782e+00,\n",
       "        0.00000000e+00, 1.20687556e+00, 3.41019928e-01, 3.73238206e-01,\n",
       "        2.81772435e-01, 0.00000000e+00, 0.00000000e+00, 5.93022227e-01,\n",
       "        6.82282448e-02, 0.00000000e+00],\n",
       "       [8.12003613e-02, 5.99866271e-01, 0.00000000e+00, 0.00000000e+00,\n",
       "        9.17364359e-01, 0.00000000e+00, 2.90262192e-01, 0.00000000e+00,\n",
       "        0.00000000e+00, 3.69606227e-01, 8.32231879e-01, 6.93195909e-02,\n",
       "        0.00000000e+00, 0.00000000e+00, 8.29003036e-01, 1.35367894e+00,\n",
       "        8.92405033e-01, 2.79957920e-01, 0.00000000e+00, 1.73076525e-01,\n",
       "        3.50144953e-02, 2.47300401e-01, 0.00000000e+00, 0.00000000e+00,\n",
       "        1.72786236e-01, 3.62147182e-01, 0.00000000e+00, 1.45263970e+00,\n",
       "        0.00000000e+00, 0.00000000e+00, 0.00000000e+00, 1.46190369e+00,\n",
       "        2.81279117e-01, 0.00000000e+00, 0.00000000e+00, 0.00000000e+00,\n",
       "        1.71830833e-01, 0.00000000e+00, 0.00000000e+00, 0.00000000e+00,\n",
       "        5.32919288e-01, 3.07142854e-01, 1.51489842e+00, 2.05876517e+00,\n",
       "        2.68922150e-02, 1.63438225e+00, 9.91982222e-03, 0.00000000e+00,\n",
       "        2.63581067e-01, 0.00000000e+00, 0.00000000e+00, 1.12667549e+00,\n",
       "        4.39229608e-02, 0.00000000e+00],\n",
       "       [0.00000000e+00, 6.15303159e-01, 0.00000000e+00, 0.00000000e+00,\n",
       "        7.60403037e-01, 3.34396422e-01, 0.00000000e+00, 0.00000000e+00,\n",
       "        0.00000000e+00, 8.24910760e-01, 1.10015666e+00, 2.71981955e-03,\n",
       "        0.00000000e+00, 7.51650035e-02, 0.00000000e+00, 1.31057739e+00,\n",
       "        1.14767492e+00, 0.00000000e+00, 0.00000000e+00, 6.30055904e-01,\n",
       "        4.32772994e-01, 2.86623061e-01, 0.00000000e+00, 0.00000000e+00,\n",
       "        0.00000000e+00, 1.50928661e-01, 0.00000000e+00, 2.35085273e+00,\n",
       "        0.00000000e+00, 6.78757489e-01, 1.86175749e-01, 1.09241688e+00,\n",
       "        8.14177841e-02, 4.16449428e-01, 2.48669714e-01, 0.00000000e+00,\n",
       "        3.11234146e-01, 0.00000000e+00, 4.47007596e-01, 0.00000000e+00,\n",
       "        0.00000000e+00, 4.86767143e-02, 1.30822623e+00, 1.92209077e+00,\n",
       "        0.00000000e+00, 1.66244590e+00, 0.00000000e+00, 4.37601388e-01,\n",
       "        1.29465079e+00, 0.00000000e+00, 0.00000000e+00, 9.22617495e-01,\n",
       "        3.98450941e-01, 0.00000000e+00],\n",
       "       [0.00000000e+00, 5.46230674e-01, 0.00000000e+00, 0.00000000e+00,\n",
       "        1.79390430e-01, 5.27280629e-01, 0.00000000e+00, 0.00000000e+00,\n",
       "        0.00000000e+00, 9.12733614e-01, 1.06908143e+00, 1.89638883e-02,\n",
       "        0.00000000e+00, 3.28856409e-02, 1.01672947e-01, 1.17982602e+00,\n",
       "        1.29137611e+00, 8.44556093e-03, 0.00000000e+00, 2.65366524e-01,\n",
       "        1.91728652e-01, 0.00000000e+00, 0.00000000e+00, 0.00000000e+00,\n",
       "        0.00000000e+00, 1.90732390e-01, 0.00000000e+00, 2.21963334e+00,\n",
       "        0.00000000e+00, 0.00000000e+00, 2.06298232e-02, 1.14849973e+00,\n",
       "        8.31767559e-01, 1.85497478e-01, 0.00000000e+00, 0.00000000e+00,\n",
       "        0.00000000e+00, 0.00000000e+00, 7.17890710e-02, 0.00000000e+00,\n",
       "        0.00000000e+00, 3.20504427e-01, 1.42362761e+00, 1.92894411e+00,\n",
       "        0.00000000e+00, 1.09523833e+00, 0.00000000e+00, 0.00000000e+00,\n",
       "        5.32516837e-01, 0.00000000e+00, 0.00000000e+00, 1.08365214e+00,\n",
       "        0.00000000e+00, 0.00000000e+00],\n",
       "       [0.00000000e+00, 8.39797616e-01, 0.00000000e+00, 0.00000000e+00,\n",
       "        5.76273203e-02, 3.49319577e-01, 1.57188550e-01, 0.00000000e+00,\n",
       "        0.00000000e+00, 8.68118405e-01, 9.73502517e-01, 4.27236021e-01,\n",
       "        0.00000000e+00, 0.00000000e+00, 0.00000000e+00, 1.33476996e+00,\n",
       "        1.60889006e+00, 0.00000000e+00, 0.00000000e+00, 6.41082764e-01,\n",
       "        1.03452027e-01, 0.00000000e+00, 0.00000000e+00, 0.00000000e+00,\n",
       "        0.00000000e+00, 3.27197880e-01, 0.00000000e+00, 1.39850128e+00,\n",
       "        0.00000000e+00, 0.00000000e+00, 0.00000000e+00, 6.45429730e-01,\n",
       "        6.33484244e-01, 2.07169265e-01, 3.27319950e-01, 0.00000000e+00,\n",
       "        3.50996405e-01, 0.00000000e+00, 0.00000000e+00, 5.05937636e-02,\n",
       "        5.47833741e-02, 9.78787839e-01, 1.45578611e+00, 1.62757516e+00,\n",
       "        0.00000000e+00, 8.95021558e-01, 0.00000000e+00, 8.77291113e-02,\n",
       "        0.00000000e+00, 4.64615822e-01, 0.00000000e+00, 1.12476146e+00,\n",
       "        3.15269858e-01, 1.55882090e-02],\n",
       "       [0.00000000e+00, 8.70699644e-01, 2.17348292e-01, 0.00000000e+00,\n",
       "        3.88890147e-01, 4.70277011e-01, 0.00000000e+00, 0.00000000e+00,\n",
       "        0.00000000e+00, 3.73602390e-01, 8.59932184e-01, 2.12532133e-02,\n",
       "        0.00000000e+00, 0.00000000e+00, 0.00000000e+00, 8.04888070e-01,\n",
       "        7.64947414e-01, 0.00000000e+00, 4.38416898e-02, 2.75202572e-01,\n",
       "        3.50575864e-01, 0.00000000e+00, 0.00000000e+00, 0.00000000e+00,\n",
       "        2.97814608e-03, 0.00000000e+00, 0.00000000e+00, 1.73746669e+00,\n",
       "        0.00000000e+00, 1.90198720e-01, 0.00000000e+00, 6.51982069e-01,\n",
       "        8.23402405e-01, 9.74051505e-02, 0.00000000e+00, 0.00000000e+00,\n",
       "        4.15621400e-01, 0.00000000e+00, 0.00000000e+00, 2.62369275e-01,\n",
       "        0.00000000e+00, 2.63897061e-01, 1.48563445e+00, 1.38981354e+00,\n",
       "        0.00000000e+00, 2.58531541e-01, 0.00000000e+00, 4.61724937e-01,\n",
       "        0.00000000e+00, 0.00000000e+00, 0.00000000e+00, 7.91998267e-01,\n",
       "        0.00000000e+00, 0.00000000e+00],\n",
       "       [0.00000000e+00, 7.10762382e-01, 0.00000000e+00, 0.00000000e+00,\n",
       "        5.89598119e-01, 2.57393599e-01, 1.14184633e-01, 0.00000000e+00,\n",
       "        0.00000000e+00, 0.00000000e+00, 2.13524371e-01, 0.00000000e+00,\n",
       "        0.00000000e+00, 0.00000000e+00, 0.00000000e+00, 0.00000000e+00,\n",
       "        9.63818073e-01, 2.86768317e-01, 0.00000000e+00, 2.19703421e-01,\n",
       "        0.00000000e+00, 0.00000000e+00, 0.00000000e+00, 0.00000000e+00,\n",
       "        0.00000000e+00, 0.00000000e+00, 0.00000000e+00, 2.06452584e+00,\n",
       "        0.00000000e+00, 1.76046997e-01, 0.00000000e+00, 9.74110842e-01,\n",
       "        0.00000000e+00, 2.72775054e-01, 0.00000000e+00, 0.00000000e+00,\n",
       "        0.00000000e+00, 0.00000000e+00, 3.25259149e-01, 0.00000000e+00,\n",
       "        0.00000000e+00, 1.15649617e+00, 8.04665804e-01, 1.42646337e+00,\n",
       "        3.11011732e-01, 6.41824305e-01, 0.00000000e+00, 1.54549599e-01,\n",
       "        2.31058016e-01, 2.84501612e-01, 0.00000000e+00, 3.55988264e-01,\n",
       "        0.00000000e+00, 1.66547000e-02],\n",
       "       [0.00000000e+00, 4.59400028e-01, 0.00000000e+00, 0.00000000e+00,\n",
       "        0.00000000e+00, 8.62170815e-01, 0.00000000e+00, 2.93012202e-01,\n",
       "        0.00000000e+00, 2.44278610e-01, 0.00000000e+00, 2.94885725e-01,\n",
       "        0.00000000e+00, 7.00599551e-01, 0.00000000e+00, 4.09678400e-01,\n",
       "        8.47944915e-01, 0.00000000e+00, 0.00000000e+00, 0.00000000e+00,\n",
       "        9.70434844e-02, 0.00000000e+00, 0.00000000e+00, 5.40999055e-01,\n",
       "        0.00000000e+00, 0.00000000e+00, 2.94447184e-01, 1.36614156e+00,\n",
       "        0.00000000e+00, 4.77541536e-02, 0.00000000e+00, 8.49524140e-02,\n",
       "        3.15096200e-01, 3.79861295e-01, 5.83317429e-02, 0.00000000e+00,\n",
       "        0.00000000e+00, 0.00000000e+00, 4.85889196e-01, 0.00000000e+00,\n",
       "        0.00000000e+00, 0.00000000e+00, 4.13369387e-01, 1.58468857e-01,\n",
       "        0.00000000e+00, 4.07588661e-01, 0.00000000e+00, 2.69160807e-01,\n",
       "        5.77704549e-01, 0.00000000e+00, 0.00000000e+00, 3.99622828e-01,\n",
       "        2.26868540e-02, 6.68567866e-02],\n",
       "       [0.00000000e+00, 0.00000000e+00, 0.00000000e+00, 0.00000000e+00,\n",
       "        3.31883132e-02, 4.79679286e-01, 7.02383518e-01, 3.13475370e-01,\n",
       "        0.00000000e+00, 1.07965946e+00, 1.24655530e-01, 7.56445378e-02,\n",
       "        0.00000000e+00, 3.81928146e-01, 2.20087767e-01, 1.36567879e+00,\n",
       "        1.01440763e+00, 1.66498691e-01, 0.00000000e+00, 5.68127036e-01,\n",
       "        0.00000000e+00, 0.00000000e+00, 3.05912048e-02, 1.85367465e-03,\n",
       "        0.00000000e+00, 5.58835626e-01, 3.15349340e-01, 2.74532944e-01,\n",
       "        0.00000000e+00, 0.00000000e+00, 0.00000000e+00, 6.95086718e-01,\n",
       "        9.84776258e-01, 0.00000000e+00, 5.93979955e-02, 0.00000000e+00,\n",
       "        0.00000000e+00, 0.00000000e+00, 0.00000000e+00, 2.17954144e-01,\n",
       "        1.19862616e-01, 3.39611173e-02, 6.23646140e-01, 5.64466894e-01,\n",
       "        0.00000000e+00, 1.24840677e-01, 0.00000000e+00, 3.99574637e-01,\n",
       "        0.00000000e+00, 0.00000000e+00, 7.66486526e-02, 3.59743863e-01,\n",
       "        0.00000000e+00, 4.55534339e-01],\n",
       "       [3.79917920e-02, 5.99945188e-01, 1.43087640e-01, 0.00000000e+00,\n",
       "        1.27424097e+00, 0.00000000e+00, 7.23830163e-01, 0.00000000e+00,\n",
       "        0.00000000e+00, 5.95397234e-01, 1.19306087e-01, 7.42862046e-01,\n",
       "        5.47072530e-01, 0.00000000e+00, 5.35174370e-01, 1.01423168e+00,\n",
       "        6.16238236e-01, 0.00000000e+00, 3.81099492e-01, 7.90973246e-01,\n",
       "        0.00000000e+00, 4.86597121e-02, 0.00000000e+00, 0.00000000e+00,\n",
       "        3.14935803e-01, 4.60512549e-01, 0.00000000e+00, 0.00000000e+00,\n",
       "        0.00000000e+00, 1.88098460e-01, 0.00000000e+00, 5.71393788e-01,\n",
       "        7.88905621e-02, 0.00000000e+00, 2.78822660e-01, 2.52552956e-01,\n",
       "        6.13377273e-01, 1.07059047e-01, 0.00000000e+00, 5.34752309e-02,\n",
       "        1.63929522e-01, 4.71786112e-02, 1.42013955e+00, 1.71750546e+00,\n",
       "        4.10765707e-02, 5.92642426e-01, 6.14109933e-01, 6.88156962e-01,\n",
       "        0.00000000e+00, 1.26047343e-01, 0.00000000e+00, 3.00867289e-01,\n",
       "        4.06625390e-01, 0.00000000e+00]], dtype=float32)"
      ]
     },
     "execution_count": 11,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "hidden_layer_output"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {
    "collapsed": true
   },
   "outputs": [],
   "source": []
  },
  {
   "cell_type": "code",
   "execution_count": 12,
   "metadata": {
    "collapsed": false
   },
   "outputs": [
    {
     "data": {
      "image/png": "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",
      "text/plain": [
       "<Figure size 1152x288 with 1 Axes>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    },
    {
     "data": {
      "image/png": "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",
      "text/plain": [
       "<Figure size 1152x288 with 1 Axes>"
      ]
     },
     "metadata": {
      "needs_background": "light"
     },
     "output_type": "display_data"
    },
    {
     "data": {
      "image/png": "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",
      "text/plain": [
       "<Figure size 1152x288 with 1 Axes>"
      ]
     },
     "metadata": {
      "needs_background": "light"
     },
     "output_type": "display_data"
    },
    {
     "data": {
      "image/png": "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",
      "text/plain": [
       "<Figure size 1152x288 with 1 Axes>"
      ]
     },
     "metadata": {
      "needs_background": "light"
     },
     "output_type": "display_data"
    },
    {
     "data": {
      "image/png": "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",
      "text/plain": [
       "<Figure size 1152x288 with 1 Axes>"
      ]
     },
     "metadata": {
      "needs_background": "light"
     },
     "output_type": "display_data"
    },
    {
     "data": {
      "image/png": "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",
      "text/plain": [
       "<Figure size 1152x288 with 1 Axes>"
      ]
     },
     "metadata": {
      "needs_background": "light"
     },
     "output_type": "display_data"
    },
    {
     "data": {
      "image/png": "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",
      "text/plain": [
       "<Figure size 1152x288 with 1 Axes>"
      ]
     },
     "metadata": {
      "needs_background": "light"
     },
     "output_type": "display_data"
    },
    {
     "data": {
      "image/png": "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",
      "text/plain": [
       "<Figure size 1152x288 with 1 Axes>"
      ]
     },
     "metadata": {
      "needs_background": "light"
     },
     "output_type": "display_data"
    },
    {
     "data": {
      "image/png": "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",
      "text/plain": [
       "<Figure size 1152x288 with 1 Axes>"
      ]
     },
     "metadata": {
      "needs_background": "light"
     },
     "output_type": "display_data"
    },
    {
     "data": {
      "image/png": "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",
      "text/plain": [
       "<Figure size 1152x288 with 1 Axes>"
      ]
     },
     "metadata": {
      "needs_background": "light"
     },
     "output_type": "display_data"
    },
    {
     "data": {
      "image/png": "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",
      "text/plain": [
       "<Figure size 1152x288 with 1 Axes>"
      ]
     },
     "metadata": {
      "needs_background": "light"
     },
     "output_type": "display_data"
    }
   ],
   "source": [
    "for i in range(10):\n",
    "    history = model.fit(at, at, epochs=50, batch_size=64, shuffle=True, verbose=False)\n",
    "    #print(history.history['acc'][-1])\n",
    "    pt = model.predict(at)\n",
    "    p = pt.transpose()\n",
    "    fig, axes = plt.subplots(1,1,figsize=(16,4))\n",
    "    axes.imshow(p,cmap=\"gray\", interpolation=\"nearest\");\n",
    "    axes.set_xlim(0,75)\n",
    "    display.clear_output(wait=True)\n",
    "    display.display(plt.gcf());\n",
    "    #axes[1].imshow(a,cmap=\"gray\", interpolation=\"nearest\")"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 13,
   "metadata": {
    "collapsed": false
   },
   "outputs": [
    {
     "data": {
      "text/plain": [
       "584.5897"
      ]
     },
     "execution_count": 13,
     "metadata": {},
     "output_type": "execute_result"
    },
    {
     "data": {
      "image/png": "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",
      "text/plain": [
       "<Figure size 7200x432 with 3 Axes>"
      ]
     },
     "metadata": {
      "needs_background": "light"
     },
     "output_type": "display_data"
    }
   ],
   "source": [
    "test_a = gm.getMapsRangeWindowed(4, 0, 5)\n",
    "test_at = test_a.transpose()\n",
    "test_pt = model.predict(test_at)\n",
    "test_p = test_pt.transpose()\n",
    "fig, axes = plt.subplots(3,figsize=(100,6))\n",
    "axes[0].imshow(test_p,cmap=\"gray\", interpolation=\"nearest\")\n",
    "axes[1].imshow(test_a,cmap=\"gray\", interpolation=\"nearest\")\n",
    "axes[2].imshow(test_a-test_p, interpolation=\"nearest\", cmap=\"hot\", vmin=-1, vmax=1)\n",
    "np.sum(pow(test_a-test_p,2))"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 14,
   "metadata": {
    "collapsed": false
   },
   "outputs": [
    {
     "data": {
      "text/plain": [
       "<matplotlib.image.AxesImage at 0x19d40730eb0>"
      ]
     },
     "execution_count": 14,
     "metadata": {},
     "output_type": "execute_result"
    },
    {
     "data": {
      "image/png": "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",
      "text/plain": [
       "<Figure size 1152x288 with 2 Axes>"
      ]
     },
     "metadata": {
      "needs_background": "light"
     },
     "output_type": "display_data"
    }
   ],
   "source": [
    "pt = model.predict(at)\n",
    "p = pt.transpose()\n",
    "fig, axes = plt.subplots(2,1,figsize=(16,4))\n",
    "axes[0].imshow(p,cmap=\"gray\", interpolation=\"nearest\")\n",
    "axes[1].imshow(a,cmap=\"gray\", interpolation=\"nearest\")"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 15,
   "metadata": {
    "collapsed": false
   },
   "outputs": [
    {
     "data": {
      "text/plain": [
       "<matplotlib.image.AxesImage at 0x19d409f3e80>"
      ]
     },
     "execution_count": 15,
     "metadata": {},
     "output_type": "execute_result"
    },
    {
     "data": {
      "image/png": "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",
      "text/plain": [
       "<Figure size 1152x288 with 1 Axes>"
      ]
     },
     "metadata": {
      "needs_background": "light"
     },
     "output_type": "display_data"
    }
   ],
   "source": [
    "pt = model.predict(at)\n",
    "p = pt.transpose()\n",
    "fig, axes = plt.subplots(1,1,figsize=(16,4))\n",
    "axes.imshow(a-p,cmap=\"gray\", interpolation=\"nearest\")\n"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {
    "collapsed": true
   },
   "outputs": [],
   "source": []
  },
  {
   "cell_type": "code",
   "execution_count": 16,
   "metadata": {
    "collapsed": false
   },
   "outputs": [
    {
     "data": {
      "text/plain": [
       "(0.0, 75.0)"
      ]
     },
     "execution_count": 16,
     "metadata": {},
     "output_type": "execute_result"
    },
    {
     "data": {
      "image/png": "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",
      "text/plain": [
       "<Figure size 504x360 with 2 Axes>"
      ]
     },
     "metadata": {
      "needs_background": "light"
     },
     "output_type": "display_data"
    }
   ],
   "source": [
    "fig, axes = plt.subplots(2,figsize=(7,5))\n",
    "axes[0].imshow(test_a,cmap=\"gray\", interpolation=\"nearest\")\n",
    "axes[1].plot(hidden_layer_output[:])\n",
    "axes[1].set_xlim(0,75)\n",
    "#axes[1].scatter(hidd en_layer_output[:,0], hidden_layer_output[:,1])\n",
    "#axes[1].imshow(hidden_layer_output.transpose(),cmap=\"gray\", interpolation=\"nearest\")"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 17,
   "metadata": {
    "collapsed": false
   },
   "outputs": [
    {
     "data": {
      "text/plain": [
       "[<keras.layers.core.Dense at 0x19d35b98670>,\n",
       " <keras.layers.core.Activation at 0x19d4b3b85e0>,\n",
       " <keras.layers.core.Dense at 0x19d356c9400>,\n",
       " <keras.layers.core.Activation at 0x19d356f34f0>,\n",
       " <keras.layers.core.Dense at 0x19d35bd0640>,\n",
       " <keras.layers.core.Activation at 0x19d35b78a60>,\n",
       " <keras.layers.core.Dense at 0x19d35b25c70>,\n",
       " <keras.layers.core.Activation at 0x19d35b6ed00>,\n",
       " <keras.layers.core.Dense at 0x19d35bdc2e0>,\n",
       " <keras.layers.core.Activation at 0x19d35bdc310>,\n",
       " <keras.layers.core.Dense at 0x19d35bb99d0>,\n",
       " <keras.layers.core.Activation at 0x19d35bd83d0>,\n",
       " <keras.layers.core.Dense at 0x19d35be6d30>,\n",
       " <keras.layers.core.Activation at 0x19d35be25e0>,\n",
       " <keras.layers.core.Dense at 0x19d35bdc1c0>,\n",
       " <keras.layers.core.Activation at 0x19d35be2e80>,\n",
       " <keras.layers.core.Dense at 0x19d35bf2f70>,\n",
       " <keras.layers.core.Activation at 0x19d35bf2160>,\n",
       " <keras.layers.core.Dense at 0x19d35beb340>,\n",
       " <keras.layers.core.Activation at 0x19d35bffd60>,\n",
       " <keras.layers.core.Dense at 0x19d35be2a00>,\n",
       " <keras.layers.core.Activation at 0x19d35c0a0a0>,\n",
       " <keras.layers.core.Dense at 0x19d35c0a250>,\n",
       " <keras.layers.core.Activation at 0x19d35c0a790>,\n",
       " <keras.layers.core.Dense at 0x19d35c1b940>,\n",
       " <keras.layers.core.Activation at 0x19d35c1b730>]"
      ]
     },
     "execution_count": 17,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "model.layers"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 18,
   "metadata": {
    "collapsed": false
   },
   "outputs": [
    {
     "data": {
      "image/png": "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",
      "text/plain": [
       "<Figure size 504x360 with 1 Axes>"
      ]
     },
     "metadata": {
      "needs_background": "light"
     },
     "output_type": "display_data"
    }
   ],
   "source": [
    "fig, axes = plt.subplots(1,figsize=(7,5))\n",
    "\n",
    "get_hidden_layer_output = K.function([model.layers[0].input],\n",
    "                                  [model.layers[4].output])\n",
    "\n",
    "if True:\n",
    "    test_a = gm.getMapsRangeWindowed(4, 0, 1)\n",
    "    test_at = test_a.transpose()\n",
    "    hidden_layer_output_A = get_hidden_layer_output([test_at])[0]\n",
    "    plt.scatter(hidden_layer_output_A[:,0], hidden_layer_output_A[:,1], alpha=1)\n",
    "\n",
    "if True:\n",
    "    test_a = gm.getMapsRangeWindowed(4, 1, 2)\n",
    "    test_at = test_a.transpose()\n",
    "    hidden_layer_output_B = get_hidden_layer_output([test_at])[0]\n",
    "    plt.scatter(hidden_layer_output_B[:,0], hidden_layer_output_B[:,1], color=\"green\", alpha=1)\n",
    "\n",
    "if True:\n",
    "    test_a = gm.getMapsRangeWindowed(4, 2, 3)\n",
    "    test_at = test_a.transpose()\n",
    "    hidden_layer_output = get_hidden_layer_output([test_at])[0]\n",
    "    plt.scatter(hidden_layer_output[:,0], hidden_layer_output[:,1], color=\"red\", alpha=1)\n",
    "\n",
    "if False:\n",
    "    test_a = gm.getMapsRangeWindowed(4, 3, 4)\n",
    "    test_at = test_a.transpose()\n",
    "    hidden_layer_output = get_hidden_layer_output([test_at])[0]\n",
    "    plt.scatter(hidden_layer_output[:,0], hidden_layer_output[:,1], color=\"orange\", alpha=.5)\n",
    "\n"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 19,
   "metadata": {
    "collapsed": false
   },
   "outputs": [
    {
     "data": {
      "text/plain": [
       "array([[ 0.05473968, -0.14120965, -0.47438407,  0.13577187, -0.9391118 ,\n",
       "        -0.73181546, -0.06138577, -0.5656669 ,  0.10538684, -0.17031376,\n",
       "        -0.38350454, -0.9597273 , -0.40733078, -0.48703477,  0.5527987 ,\n",
       "        -0.01573296,  0.13175076, -0.86638784],\n",
       "       [-0.22425331, -0.05604824, -0.5265307 ,  0.07652354, -0.34824362,\n",
       "        -1.0894794 ,  0.24653208, -0.81301147,  0.15201563, -0.356752  ,\n",
       "        -0.44542018, -1.2452903 , -0.40680867, -0.98710847,  0.64364105,\n",
       "        -0.28345615,  0.1962702 , -0.70494866],\n",
       "       [-0.11616445, -0.34391144, -0.6025249 ,  0.27452555, -0.30767033,\n",
       "        -0.42573565,  0.15877156, -0.7799236 ,  0.0052478 , -0.6576821 ,\n",
       "        -0.3694914 , -0.83124524, -0.7069916 , -0.729063  ,  0.5677379 ,\n",
       "        -0.10959845, -0.12840226, -0.7283452 ],\n",
       "       [-0.09810477, -0.35672137, -0.6063158 ,  0.38106635, -0.7627249 ,\n",
       "        -0.7098891 , -0.20235538, -0.26235196, -0.04225278, -0.5297934 ,\n",
       "        -0.562308  , -1.2788335 , -0.58824486, -0.72650576,  0.6268138 ,\n",
       "         0.2228282 ,  0.19825092, -1.0072796 ],\n",
       "       [-0.02782625, -0.2783536 , -0.46400845,  0.1447938 , -0.7097844 ,\n",
       "        -0.82817566,  0.09645667, -0.6850797 ,  0.20261553, -0.6165188 ,\n",
       "        -0.46691796, -1.0336838 , -0.6957533 , -0.7305143 ,  0.6351467 ,\n",
       "        -0.11858052,  0.22474296, -0.8203693 ],\n",
       "       [-0.21120805, -0.43347043, -0.6486863 ,  0.19290353, -0.6815183 ,\n",
       "        -1.089564  ,  0.1818308 , -0.9702751 , -0.19520508, -0.67549247,\n",
       "        -0.48625496, -1.4008471 , -0.69469017, -1.0626303 ,  0.6172376 ,\n",
       "         0.05326208,  0.24743351, -1.3232336 ],\n",
       "       [-0.23790962,  0.0237988 , -0.128445  , -0.2336531 , -0.41765797,\n",
       "        -0.86318713,  0.18785661, -0.7888295 , -0.12753075, -0.5028435 ,\n",
       "        -0.21445556, -1.012782  , -0.46450892, -0.82644606,  0.47203648,\n",
       "        -0.2429079 ,  0.49375   , -0.8509975 ],\n",
       "       [-0.04351101,  0.03258992, -0.08096514, -0.0783623 ,  0.04365854,\n",
       "        -0.65579605,  0.41951638, -0.4757294 ,  0.10355657, -0.4660797 ,\n",
       "        -0.24534257, -0.91372186, -0.48416004, -0.65047157,  0.5056681 ,\n",
       "        -0.01776172,  0.19389693, -0.49034566],\n",
       "       [-0.22387053, -0.08686222, -0.28842264, -0.08294132, -0.09674045,\n",
       "        -0.547534  ,  0.24074367, -0.62895304,  0.05258131, -0.55986065,\n",
       "        -0.16513398, -0.89714986, -0.4739024 , -0.6236005 ,  0.46398467,\n",
       "        -0.10355721,  0.13860792, -0.62406516],\n",
       "       [-0.01644273,  0.18501651,  0.07582963, -0.04151297,  0.06583323,\n",
       "        -0.8874838 ,  0.65403444, -0.4318363 , -0.01779025, -0.31069088,\n",
       "        -0.1392533 , -0.51618314, -0.07488342, -0.8555435 ,  0.28007528,\n",
       "        -0.24197835, -0.10624034, -0.6245771 ],\n",
       "       [-0.11863788, -0.20921001, -0.08297876,  0.10472743,  0.3613586 ,\n",
       "        -0.39043516,  0.32965785, -0.2751169 ,  0.04700372, -0.32800066,\n",
       "        -0.07964443, -0.5135623 ,  0.01372168, -0.60328704,  0.29469576,\n",
       "        -0.02250904, -0.06291146, -0.30766916],\n",
       "       [-0.07677024,  0.15623803, -0.14741841,  0.0105388 , -0.20286399,\n",
       "        -0.30165192, -0.06242453, -0.1291326 , -0.05201832, -0.00945629,\n",
       "        -0.2528077 , -0.5230158 , -0.10677122, -0.23395556,  0.38510635,\n",
       "         0.0226214 , -0.01670401, -0.49368232],\n",
       "       [ 0.3139607 ,  0.053349  , -0.29747638,  0.46872553, -0.68989944,\n",
       "        -0.48821193, -0.09095667, -0.24761169,  0.2800276 , -0.23384535,\n",
       "        -0.8511877 , -0.45091575, -0.6713463 , -0.06812919,  0.4827193 ,\n",
       "         0.3110291 , -0.08978092, -1.0846479 ]], dtype=float32)"
      ]
     },
     "execution_count": 19,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "hidden_layer_output_A"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 20,
   "metadata": {
    "collapsed": false
   },
   "outputs": [
    {
     "data": {
      "text/plain": [
       "[<keras.layers.core.Dense at 0x19d35b98670>,\n",
       " <keras.layers.core.Activation at 0x19d4b3b85e0>,\n",
       " <keras.layers.core.Dense at 0x19d356c9400>,\n",
       " <keras.layers.core.Activation at 0x19d356f34f0>,\n",
       " <keras.layers.core.Dense at 0x19d35bd0640>,\n",
       " <keras.layers.core.Activation at 0x19d35b78a60>,\n",
       " <keras.layers.core.Dense at 0x19d35b25c70>,\n",
       " <keras.layers.core.Activation at 0x19d35b6ed00>,\n",
       " <keras.layers.core.Dense at 0x19d35bdc2e0>,\n",
       " <keras.layers.core.Activation at 0x19d35bdc310>,\n",
       " <keras.layers.core.Dense at 0x19d35bb99d0>,\n",
       " <keras.layers.core.Activation at 0x19d35bd83d0>,\n",
       " <keras.layers.core.Dense at 0x19d35be6d30>,\n",
       " <keras.layers.core.Activation at 0x19d35be25e0>,\n",
       " <keras.layers.core.Dense at 0x19d35bdc1c0>,\n",
       " <keras.layers.core.Activation at 0x19d35be2e80>,\n",
       " <keras.layers.core.Dense at 0x19d35bf2f70>,\n",
       " <keras.layers.core.Activation at 0x19d35bf2160>,\n",
       " <keras.layers.core.Dense at 0x19d35beb340>,\n",
       " <keras.layers.core.Activation at 0x19d35bffd60>,\n",
       " <keras.layers.core.Dense at 0x19d35be2a00>,\n",
       " <keras.layers.core.Activation at 0x19d35c0a0a0>,\n",
       " <keras.layers.core.Dense at 0x19d35c0a250>,\n",
       " <keras.layers.core.Activation at 0x19d35c0a790>,\n",
       " <keras.layers.core.Dense at 0x19d35c1b940>,\n",
       " <keras.layers.core.Activation at 0x19d35c1b730>]"
      ]
     },
     "execution_count": 20,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "model.layers"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 21,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "(None, 54)\n",
      "(None, 54)\n",
      "(None, 28)\n",
      "(None, 28)\n",
      "(None, 18)\n",
      "(None, 18)\n",
      "(None, 9)\n",
      "(None, 9)\n",
      "(None, 8)\n",
      "(None, 8)\n",
      "(None, 4)\n",
      "(None, 4)\n",
      "(None, 2)\n",
      "(None, 2)\n",
      "(None, 4)\n",
      "(None, 4)\n",
      "(None, 8)\n",
      "(None, 8)\n",
      "(None, 18)\n",
      "(None, 18)\n",
      "(None, 28)\n",
      "(None, 28)\n",
      "(None, 54)\n",
      "(None, 54)\n",
      "(None, 36)\n",
      "(None, 36)\n"
     ]
    }
   ],
   "source": [
    "for layer in model.layers:\n",
    "    print(layer.output_shape)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 22,
   "metadata": {
    "collapsed": false
   },
   "outputs": [
    {
     "data": {
      "text/plain": [
       "(-0.30717552, 0.13836479)"
      ]
     },
     "execution_count": 22,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "# make some fake input!\n",
    "h = np.ndarray(shape=(150,2))\n",
    "h[:,0] = np.random.normal(loc=hidden_layer_output_B[8,0], scale=2, size=150)\n",
    "h[:,1] = np.random.normal(loc=hidden_layer_output_B[8,1], scale=2, size=150)\n",
    "#h[:,0].fill(hidden_layer_output_B[8,0])\n",
    "#h[:,1].fill(hidden_layer_output_B[8,1])\n",
    "hidden_layer_output_B[8,0], hidden_layer_output_B[8,1]"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 23,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "[<keras.layers.core.Dense at 0x19d35b98670>,\n",
       " <keras.layers.core.Activation at 0x19d4b3b85e0>,\n",
       " <keras.layers.core.Dense at 0x19d356c9400>,\n",
       " <keras.layers.core.Activation at 0x19d356f34f0>,\n",
       " <keras.layers.core.Dense at 0x19d35bd0640>,\n",
       " <keras.layers.core.Activation at 0x19d35b78a60>,\n",
       " <keras.layers.core.Dense at 0x19d35b25c70>,\n",
       " <keras.layers.core.Activation at 0x19d35b6ed00>,\n",
       " <keras.layers.core.Dense at 0x19d35bdc2e0>,\n",
       " <keras.layers.core.Activation at 0x19d35bdc310>,\n",
       " <keras.layers.core.Dense at 0x19d35bb99d0>,\n",
       " <keras.layers.core.Activation at 0x19d35bd83d0>,\n",
       " <keras.layers.core.Dense at 0x19d35be6d30>,\n",
       " <keras.layers.core.Activation at 0x19d35be25e0>,\n",
       " <keras.layers.core.Dense at 0x19d35bdc1c0>,\n",
       " <keras.layers.core.Activation at 0x19d35be2e80>,\n",
       " <keras.layers.core.Dense at 0x19d35bf2f70>,\n",
       " <keras.layers.core.Activation at 0x19d35bf2160>,\n",
       " <keras.layers.core.Dense at 0x19d35beb340>,\n",
       " <keras.layers.core.Activation at 0x19d35bffd60>,\n",
       " <keras.layers.core.Dense at 0x19d35be2a00>,\n",
       " <keras.layers.core.Activation at 0x19d35c0a0a0>,\n",
       " <keras.layers.core.Dense at 0x19d35c0a250>,\n",
       " <keras.layers.core.Activation at 0x19d35c0a790>,\n",
       " <keras.layers.core.Dense at 0x19d35c1b940>,\n",
       " <keras.layers.core.Activation at 0x19d35c1b730>]"
      ]
     },
     "execution_count": 23,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "model.layers"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 24,
   "metadata": {
    "collapsed": false
   },
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "WARNING:tensorflow:Functional model inputs must come from `tf.keras.Input` (thus holding past layer metadata), they cannot be the output of a previous non-Input layer. Here, a tensor specified as input to \"model_2\" was not an Input tensor, it was generated by layer dense_15.\n",
      "Note that input tensors are instantiated via `tensor = tf.keras.Input(shape)`.\n",
      "The tensor that caused the issue was: dense_15/BiasAdd:0\n"
     ]
    },
    {
     "ename": "ValueError",
     "evalue": "Graph disconnected: cannot obtain value for tensor KerasTensor(type_spec=TensorSpec(shape=(None, 36), dtype=tf.float32, name='dense_13_input'), name='dense_13_input', description=\"created by layer 'dense_13_input'\") at layer \"dense_13\". The following previous layers were accessed without issue: []",
     "output_type": "error",
     "traceback": [
      "\u001b[1;31m---------------------------------------------------------------------------\u001b[0m",
      "\u001b[1;31mValueError\u001b[0m                                Traceback (most recent call last)",
      "\u001b[1;32m<ipython-input-24-db682d22ef92>\u001b[0m in \u001b[0;36m<module>\u001b[1;34m\u001b[0m\n\u001b[1;32m----> 1\u001b[1;33m inject_fake_hidden = K.function([model.layers[5].input],\n\u001b[0m\u001b[0;32m      2\u001b[0m                                 [model.layers[-1].output])\n\u001b[0;32m      3\u001b[0m \u001b[1;33m\u001b[0m\u001b[0m\n\u001b[0;32m      4\u001b[0m \u001b[1;31m#X = inject_fake_hidden([hidden_layer_output])[0].transpose()\u001b[0m\u001b[1;33m\u001b[0m\u001b[1;33m\u001b[0m\u001b[1;33m\u001b[0m\u001b[0m\n\u001b[0;32m      5\u001b[0m \u001b[0mX\u001b[0m \u001b[1;33m=\u001b[0m \u001b[0minject_fake_hidden\u001b[0m\u001b[1;33m(\u001b[0m\u001b[1;33m[\u001b[0m\u001b[0mh\u001b[0m\u001b[1;33m]\u001b[0m\u001b[1;33m)\u001b[0m\u001b[1;33m[\u001b[0m\u001b[1;36m0\u001b[0m\u001b[1;33m]\u001b[0m\u001b[1;33m.\u001b[0m\u001b[0mtranspose\u001b[0m\u001b[1;33m(\u001b[0m\u001b[1;33m)\u001b[0m\u001b[1;33m\u001b[0m\u001b[1;33m\u001b[0m\u001b[0m\n",
      "\u001b[1;32m~\\anaconda3\\lib\\site-packages\\keras\\backend.py\u001b[0m in \u001b[0;36mfunction\u001b[1;34m(inputs, outputs, updates, name, **kwargs)\u001b[0m\n\u001b[0;32m   4060\u001b[0m     \u001b[1;32mfrom\u001b[0m \u001b[0mkeras\u001b[0m \u001b[1;32mimport\u001b[0m \u001b[0mmodels\u001b[0m  \u001b[1;31m# pylint: disable=g-import-not-at-top\u001b[0m\u001b[1;33m\u001b[0m\u001b[1;33m\u001b[0m\u001b[0m\n\u001b[0;32m   4061\u001b[0m     \u001b[1;32mfrom\u001b[0m \u001b[0mkeras\u001b[0m\u001b[1;33m.\u001b[0m\u001b[0mutils\u001b[0m \u001b[1;32mimport\u001b[0m \u001b[0mtf_utils\u001b[0m  \u001b[1;31m# pylint: disable=g-import-not-at-top\u001b[0m\u001b[1;33m\u001b[0m\u001b[1;33m\u001b[0m\u001b[0m\n\u001b[1;32m-> 4062\u001b[1;33m     \u001b[0mmodel\u001b[0m \u001b[1;33m=\u001b[0m \u001b[0mmodels\u001b[0m\u001b[1;33m.\u001b[0m\u001b[0mModel\u001b[0m\u001b[1;33m(\u001b[0m\u001b[0minputs\u001b[0m\u001b[1;33m=\u001b[0m\u001b[0minputs\u001b[0m\u001b[1;33m,\u001b[0m \u001b[0moutputs\u001b[0m\u001b[1;33m=\u001b[0m\u001b[0moutputs\u001b[0m\u001b[1;33m)\u001b[0m\u001b[1;33m\u001b[0m\u001b[1;33m\u001b[0m\u001b[0m\n\u001b[0m\u001b[0;32m   4063\u001b[0m \u001b[1;33m\u001b[0m\u001b[0m\n\u001b[0;32m   4064\u001b[0m     \u001b[0mwrap_outputs\u001b[0m \u001b[1;33m=\u001b[0m \u001b[0misinstance\u001b[0m\u001b[1;33m(\u001b[0m\u001b[0moutputs\u001b[0m\u001b[1;33m,\u001b[0m \u001b[0mlist\u001b[0m\u001b[1;33m)\u001b[0m \u001b[1;32mand\u001b[0m \u001b[0mlen\u001b[0m\u001b[1;33m(\u001b[0m\u001b[0moutputs\u001b[0m\u001b[1;33m)\u001b[0m \u001b[1;33m==\u001b[0m \u001b[1;36m1\u001b[0m\u001b[1;33m\u001b[0m\u001b[1;33m\u001b[0m\u001b[0m\n",
      "\u001b[1;32m~\\anaconda3\\lib\\site-packages\\tensorflow\\python\\training\\tracking\\base.py\u001b[0m in \u001b[0;36m_method_wrapper\u001b[1;34m(self, *args, **kwargs)\u001b[0m\n\u001b[0;32m    520\u001b[0m     \u001b[0mself\u001b[0m\u001b[1;33m.\u001b[0m\u001b[0m_self_setattr_tracking\u001b[0m \u001b[1;33m=\u001b[0m \u001b[1;32mFalse\u001b[0m  \u001b[1;31m# pylint: disable=protected-access\u001b[0m\u001b[1;33m\u001b[0m\u001b[1;33m\u001b[0m\u001b[0m\n\u001b[0;32m    521\u001b[0m     \u001b[1;32mtry\u001b[0m\u001b[1;33m:\u001b[0m\u001b[1;33m\u001b[0m\u001b[1;33m\u001b[0m\u001b[0m\n\u001b[1;32m--> 522\u001b[1;33m       \u001b[0mresult\u001b[0m \u001b[1;33m=\u001b[0m \u001b[0mmethod\u001b[0m\u001b[1;33m(\u001b[0m\u001b[0mself\u001b[0m\u001b[1;33m,\u001b[0m \u001b[1;33m*\u001b[0m\u001b[0margs\u001b[0m\u001b[1;33m,\u001b[0m \u001b[1;33m**\u001b[0m\u001b[0mkwargs\u001b[0m\u001b[1;33m)\u001b[0m\u001b[1;33m\u001b[0m\u001b[1;33m\u001b[0m\u001b[0m\n\u001b[0m\u001b[0;32m    523\u001b[0m     \u001b[1;32mfinally\u001b[0m\u001b[1;33m:\u001b[0m\u001b[1;33m\u001b[0m\u001b[1;33m\u001b[0m\u001b[0m\n\u001b[0;32m    524\u001b[0m       \u001b[0mself\u001b[0m\u001b[1;33m.\u001b[0m\u001b[0m_self_setattr_tracking\u001b[0m \u001b[1;33m=\u001b[0m \u001b[0mprevious_value\u001b[0m  \u001b[1;31m# pylint: disable=protected-access\u001b[0m\u001b[1;33m\u001b[0m\u001b[1;33m\u001b[0m\u001b[0m\n",
      "\u001b[1;32m~\\anaconda3\\lib\\site-packages\\keras\\engine\\functional.py\u001b[0m in \u001b[0;36m__init__\u001b[1;34m(self, inputs, outputs, name, trainable, **kwargs)\u001b[0m\n\u001b[0;32m    108\u001b[0m     \u001b[0mgeneric_utils\u001b[0m\u001b[1;33m.\u001b[0m\u001b[0mvalidate_kwargs\u001b[0m\u001b[1;33m(\u001b[0m\u001b[0mkwargs\u001b[0m\u001b[1;33m,\u001b[0m \u001b[1;33m{\u001b[0m\u001b[1;33m}\u001b[0m\u001b[1;33m)\u001b[0m\u001b[1;33m\u001b[0m\u001b[1;33m\u001b[0m\u001b[0m\n\u001b[0;32m    109\u001b[0m     \u001b[0msuper\u001b[0m\u001b[1;33m(\u001b[0m\u001b[0mFunctional\u001b[0m\u001b[1;33m,\u001b[0m \u001b[0mself\u001b[0m\u001b[1;33m)\u001b[0m\u001b[1;33m.\u001b[0m\u001b[0m__init__\u001b[0m\u001b[1;33m(\u001b[0m\u001b[0mname\u001b[0m\u001b[1;33m=\u001b[0m\u001b[0mname\u001b[0m\u001b[1;33m,\u001b[0m \u001b[0mtrainable\u001b[0m\u001b[1;33m=\u001b[0m\u001b[0mtrainable\u001b[0m\u001b[1;33m)\u001b[0m\u001b[1;33m\u001b[0m\u001b[1;33m\u001b[0m\u001b[0m\n\u001b[1;32m--> 110\u001b[1;33m     \u001b[0mself\u001b[0m\u001b[1;33m.\u001b[0m\u001b[0m_init_graph_network\u001b[0m\u001b[1;33m(\u001b[0m\u001b[0minputs\u001b[0m\u001b[1;33m,\u001b[0m \u001b[0moutputs\u001b[0m\u001b[1;33m)\u001b[0m\u001b[1;33m\u001b[0m\u001b[1;33m\u001b[0m\u001b[0m\n\u001b[0m\u001b[0;32m    111\u001b[0m \u001b[1;33m\u001b[0m\u001b[0m\n\u001b[0;32m    112\u001b[0m   \u001b[1;33m@\u001b[0m\u001b[0mtrackable\u001b[0m\u001b[1;33m.\u001b[0m\u001b[0mno_automatic_dependency_tracking\u001b[0m\u001b[1;33m\u001b[0m\u001b[1;33m\u001b[0m\u001b[0m\n",
      "\u001b[1;32m~\\anaconda3\\lib\\site-packages\\tensorflow\\python\\training\\tracking\\base.py\u001b[0m in \u001b[0;36m_method_wrapper\u001b[1;34m(self, *args, **kwargs)\u001b[0m\n\u001b[0;32m    520\u001b[0m     \u001b[0mself\u001b[0m\u001b[1;33m.\u001b[0m\u001b[0m_self_setattr_tracking\u001b[0m \u001b[1;33m=\u001b[0m \u001b[1;32mFalse\u001b[0m  \u001b[1;31m# pylint: disable=protected-access\u001b[0m\u001b[1;33m\u001b[0m\u001b[1;33m\u001b[0m\u001b[0m\n\u001b[0;32m    521\u001b[0m     \u001b[1;32mtry\u001b[0m\u001b[1;33m:\u001b[0m\u001b[1;33m\u001b[0m\u001b[1;33m\u001b[0m\u001b[0m\n\u001b[1;32m--> 522\u001b[1;33m       \u001b[0mresult\u001b[0m \u001b[1;33m=\u001b[0m \u001b[0mmethod\u001b[0m\u001b[1;33m(\u001b[0m\u001b[0mself\u001b[0m\u001b[1;33m,\u001b[0m \u001b[1;33m*\u001b[0m\u001b[0margs\u001b[0m\u001b[1;33m,\u001b[0m \u001b[1;33m**\u001b[0m\u001b[0mkwargs\u001b[0m\u001b[1;33m)\u001b[0m\u001b[1;33m\u001b[0m\u001b[1;33m\u001b[0m\u001b[0m\n\u001b[0m\u001b[0;32m    523\u001b[0m     \u001b[1;32mfinally\u001b[0m\u001b[1;33m:\u001b[0m\u001b[1;33m\u001b[0m\u001b[1;33m\u001b[0m\u001b[0m\n\u001b[0;32m    524\u001b[0m       \u001b[0mself\u001b[0m\u001b[1;33m.\u001b[0m\u001b[0m_self_setattr_tracking\u001b[0m \u001b[1;33m=\u001b[0m \u001b[0mprevious_value\u001b[0m  \u001b[1;31m# pylint: disable=protected-access\u001b[0m\u001b[1;33m\u001b[0m\u001b[1;33m\u001b[0m\u001b[0m\n",
      "\u001b[1;32m~\\anaconda3\\lib\\site-packages\\keras\\engine\\functional.py\u001b[0m in \u001b[0;36m_init_graph_network\u001b[1;34m(self, inputs, outputs)\u001b[0m\n\u001b[0;32m    191\u001b[0m \u001b[1;33m\u001b[0m\u001b[0m\n\u001b[0;32m    192\u001b[0m     \u001b[1;31m# Keep track of the network's nodes and layers.\u001b[0m\u001b[1;33m\u001b[0m\u001b[1;33m\u001b[0m\u001b[1;33m\u001b[0m\u001b[0m\n\u001b[1;32m--> 193\u001b[1;33m     nodes, nodes_by_depth, layers, _ = _map_graph_network(\n\u001b[0m\u001b[0;32m    194\u001b[0m         self.inputs, self.outputs)\n\u001b[0;32m    195\u001b[0m     \u001b[0mself\u001b[0m\u001b[1;33m.\u001b[0m\u001b[0m_network_nodes\u001b[0m \u001b[1;33m=\u001b[0m \u001b[0mnodes\u001b[0m\u001b[1;33m\u001b[0m\u001b[1;33m\u001b[0m\u001b[0m\n",
      "\u001b[1;32m~\\anaconda3\\lib\\site-packages\\keras\\engine\\functional.py\u001b[0m in \u001b[0;36m_map_graph_network\u001b[1;34m(inputs, outputs)\u001b[0m\n\u001b[0;32m    978\u001b[0m         \u001b[1;32mfor\u001b[0m \u001b[0mx\u001b[0m \u001b[1;32min\u001b[0m \u001b[0mtf\u001b[0m\u001b[1;33m.\u001b[0m\u001b[0mnest\u001b[0m\u001b[1;33m.\u001b[0m\u001b[0mflatten\u001b[0m\u001b[1;33m(\u001b[0m\u001b[0mnode\u001b[0m\u001b[1;33m.\u001b[0m\u001b[0mkeras_inputs\u001b[0m\u001b[1;33m)\u001b[0m\u001b[1;33m:\u001b[0m\u001b[1;33m\u001b[0m\u001b[1;33m\u001b[0m\u001b[0m\n\u001b[0;32m    979\u001b[0m           \u001b[1;32mif\u001b[0m \u001b[0mid\u001b[0m\u001b[1;33m(\u001b[0m\u001b[0mx\u001b[0m\u001b[1;33m)\u001b[0m \u001b[1;32mnot\u001b[0m \u001b[1;32min\u001b[0m \u001b[0mcomputable_tensors\u001b[0m\u001b[1;33m:\u001b[0m\u001b[1;33m\u001b[0m\u001b[1;33m\u001b[0m\u001b[0m\n\u001b[1;32m--> 980\u001b[1;33m             raise ValueError('Graph disconnected: '\n\u001b[0m\u001b[0;32m    981\u001b[0m                              \u001b[1;34m'cannot obtain value for tensor '\u001b[0m \u001b[1;33m+\u001b[0m \u001b[0mstr\u001b[0m\u001b[1;33m(\u001b[0m\u001b[0mx\u001b[0m\u001b[1;33m)\u001b[0m \u001b[1;33m+\u001b[0m\u001b[1;33m\u001b[0m\u001b[1;33m\u001b[0m\u001b[0m\n\u001b[0;32m    982\u001b[0m                              \u001b[1;34m' at layer \"'\u001b[0m \u001b[1;33m+\u001b[0m \u001b[0mlayer\u001b[0m\u001b[1;33m.\u001b[0m\u001b[0mname\u001b[0m \u001b[1;33m+\u001b[0m \u001b[1;34m'\". '\u001b[0m\u001b[1;33m\u001b[0m\u001b[1;33m\u001b[0m\u001b[0m\n",
      "\u001b[1;31mValueError\u001b[0m: Graph disconnected: cannot obtain value for tensor KerasTensor(type_spec=TensorSpec(shape=(None, 36), dtype=tf.float32, name='dense_13_input'), name='dense_13_input', description=\"created by layer 'dense_13_input'\") at layer \"dense_13\". The following previous layers were accessed without issue: []"
     ]
    }
   ],
   "source": [
    "inject_fake_hidden = K.function([model.layers[5].input],\n",
    "                                [model.layers[-1].output])\n",
    "\n",
    "#X = inject_fake_hidden([hidden_layer_output])[0].transpose()\n",
    "X = inject_fake_hidden([h])[0].transpose()\n",
    "\n",
    "#th = 0.1\n",
    "#X[ X <= th] = 0\n",
    "#X[ X > th] = 1\n",
    "\n",
    "plt.imshow(X[:14,:], cmap=\"gray\", interpolation=\"nearest\")"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 83,
   "metadata": {
    "collapsed": true
   },
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "WARNING:tensorflow:Functional inputs must come from `tf.keras.Input` (thus holding past layer metadata), they cannot be the output of a previous non-Input layer. Here, a tensor specified as input to \"functional_30\" was not an Input tensor, it was generated by layer dense_8.\n",
      "Note that input tensors are instantiated via `tensor = tf.keras.Input(shape)`.\n",
      "The tensor that caused the issue was: dense_8/BiasAdd:0\n"
     ]
    },
    {
     "ename": "ValueError",
     "evalue": "Graph disconnected: cannot obtain value for tensor Tensor(\"dense_7_input:0\", shape=(None, 36), dtype=float32) at layer \"dense_7\". The following previous layers were accessed without issue: []",
     "output_type": "error",
     "traceback": [
      "\u001b[1;31m---------------------------------------------------------------------------\u001b[0m",
      "\u001b[1;31mValueError\u001b[0m                                Traceback (most recent call last)",
      "\u001b[1;32m<ipython-input-83-d886a116f871>\u001b[0m in \u001b[0;36m<module>\u001b[1;34m\u001b[0m\n\u001b[0;32m      1\u001b[0m \u001b[0mfig\u001b[0m\u001b[1;33m,\u001b[0m \u001b[0maxes\u001b[0m \u001b[1;33m=\u001b[0m \u001b[0mplt\u001b[0m\u001b[1;33m.\u001b[0m\u001b[0msubplots\u001b[0m\u001b[1;33m(\u001b[0m\u001b[1;36m1\u001b[0m\u001b[1;33m,\u001b[0m\u001b[0mfigsize\u001b[0m\u001b[1;33m=\u001b[0m\u001b[1;33m(\u001b[0m\u001b[1;36m7\u001b[0m\u001b[1;33m,\u001b[0m\u001b[1;36m5\u001b[0m\u001b[1;33m)\u001b[0m\u001b[1;33m)\u001b[0m\u001b[1;33m\u001b[0m\u001b[1;33m\u001b[0m\u001b[0m\n\u001b[1;32m----> 2\u001b[1;33m inject_fake_hidden = K.function([model.layers[4].input],\n\u001b[0m\u001b[0;32m      3\u001b[0m                                 [model.layers[0].output])\n\u001b[0;32m      4\u001b[0m \u001b[1;33m\u001b[0m\u001b[0m\n\u001b[0;32m      5\u001b[0m \u001b[0minject_fake_hidden\u001b[0m\u001b[1;33m(\u001b[0m\u001b[1;33m[\u001b[0m\u001b[0mhidden_layer_output\u001b[0m\u001b[1;33m]\u001b[0m\u001b[1;33m)\u001b[0m\u001b[1;33m\u001b[0m\u001b[1;33m\u001b[0m\u001b[0m\n",
      "\u001b[1;32mD:\\Users\\HP\\anaconda3\\lib\\site-packages\\tensorflow\\python\\keras\\backend.py\u001b[0m in \u001b[0;36mfunction\u001b[1;34m(inputs, outputs, updates, name, **kwargs)\u001b[0m\n\u001b[0;32m   3934\u001b[0m     \u001b[1;32mfrom\u001b[0m \u001b[0mtensorflow\u001b[0m\u001b[1;33m.\u001b[0m\u001b[0mpython\u001b[0m\u001b[1;33m.\u001b[0m\u001b[0mkeras\u001b[0m \u001b[1;32mimport\u001b[0m \u001b[0mmodels\u001b[0m  \u001b[1;31m# pylint: disable=g-import-not-at-top\u001b[0m\u001b[1;33m\u001b[0m\u001b[1;33m\u001b[0m\u001b[0m\n\u001b[0;32m   3935\u001b[0m     \u001b[1;32mfrom\u001b[0m \u001b[0mtensorflow\u001b[0m\u001b[1;33m.\u001b[0m\u001b[0mpython\u001b[0m\u001b[1;33m.\u001b[0m\u001b[0mkeras\u001b[0m\u001b[1;33m.\u001b[0m\u001b[0mutils\u001b[0m \u001b[1;32mimport\u001b[0m \u001b[0mtf_utils\u001b[0m  \u001b[1;31m# pylint: disable=g-import-not-at-top\u001b[0m\u001b[1;33m\u001b[0m\u001b[1;33m\u001b[0m\u001b[0m\n\u001b[1;32m-> 3936\u001b[1;33m     \u001b[0mmodel\u001b[0m \u001b[1;33m=\u001b[0m \u001b[0mmodels\u001b[0m\u001b[1;33m.\u001b[0m\u001b[0mModel\u001b[0m\u001b[1;33m(\u001b[0m\u001b[0minputs\u001b[0m\u001b[1;33m=\u001b[0m\u001b[0minputs\u001b[0m\u001b[1;33m,\u001b[0m \u001b[0moutputs\u001b[0m\u001b[1;33m=\u001b[0m\u001b[0moutputs\u001b[0m\u001b[1;33m)\u001b[0m\u001b[1;33m\u001b[0m\u001b[1;33m\u001b[0m\u001b[0m\n\u001b[0m\u001b[0;32m   3937\u001b[0m \u001b[1;33m\u001b[0m\u001b[0m\n\u001b[0;32m   3938\u001b[0m     \u001b[0mwrap_outputs\u001b[0m \u001b[1;33m=\u001b[0m \u001b[0misinstance\u001b[0m\u001b[1;33m(\u001b[0m\u001b[0moutputs\u001b[0m\u001b[1;33m,\u001b[0m \u001b[0mlist\u001b[0m\u001b[1;33m)\u001b[0m \u001b[1;32mand\u001b[0m \u001b[0mlen\u001b[0m\u001b[1;33m(\u001b[0m\u001b[0moutputs\u001b[0m\u001b[1;33m)\u001b[0m \u001b[1;33m==\u001b[0m \u001b[1;36m1\u001b[0m\u001b[1;33m\u001b[0m\u001b[1;33m\u001b[0m\u001b[0m\n",
      "\u001b[1;32mD:\\Users\\HP\\anaconda3\\lib\\site-packages\\tensorflow\\python\\keras\\engine\\training.py\u001b[0m in \u001b[0;36m__new__\u001b[1;34m(cls, *args, **kwargs)\u001b[0m\n\u001b[0;32m    240\u001b[0m       \u001b[1;31m# Functional model\u001b[0m\u001b[1;33m\u001b[0m\u001b[1;33m\u001b[0m\u001b[1;33m\u001b[0m\u001b[0m\n\u001b[0;32m    241\u001b[0m       \u001b[1;32mfrom\u001b[0m \u001b[0mtensorflow\u001b[0m\u001b[1;33m.\u001b[0m\u001b[0mpython\u001b[0m\u001b[1;33m.\u001b[0m\u001b[0mkeras\u001b[0m\u001b[1;33m.\u001b[0m\u001b[0mengine\u001b[0m \u001b[1;32mimport\u001b[0m \u001b[0mfunctional\u001b[0m  \u001b[1;31m# pylint: disable=g-import-not-at-top\u001b[0m\u001b[1;33m\u001b[0m\u001b[1;33m\u001b[0m\u001b[0m\n\u001b[1;32m--> 242\u001b[1;33m       \u001b[1;32mreturn\u001b[0m \u001b[0mfunctional\u001b[0m\u001b[1;33m.\u001b[0m\u001b[0mFunctional\u001b[0m\u001b[1;33m(\u001b[0m\u001b[1;33m*\u001b[0m\u001b[0margs\u001b[0m\u001b[1;33m,\u001b[0m \u001b[1;33m**\u001b[0m\u001b[0mkwargs\u001b[0m\u001b[1;33m)\u001b[0m\u001b[1;33m\u001b[0m\u001b[1;33m\u001b[0m\u001b[0m\n\u001b[0m\u001b[0;32m    243\u001b[0m     \u001b[1;32melse\u001b[0m\u001b[1;33m:\u001b[0m\u001b[1;33m\u001b[0m\u001b[1;33m\u001b[0m\u001b[0m\n\u001b[0;32m    244\u001b[0m       \u001b[1;32mreturn\u001b[0m \u001b[0msuper\u001b[0m\u001b[1;33m(\u001b[0m\u001b[0mModel\u001b[0m\u001b[1;33m,\u001b[0m \u001b[0mcls\u001b[0m\u001b[1;33m)\u001b[0m\u001b[1;33m.\u001b[0m\u001b[0m__new__\u001b[0m\u001b[1;33m(\u001b[0m\u001b[0mcls\u001b[0m\u001b[1;33m,\u001b[0m \u001b[1;33m*\u001b[0m\u001b[0margs\u001b[0m\u001b[1;33m,\u001b[0m \u001b[1;33m**\u001b[0m\u001b[0mkwargs\u001b[0m\u001b[1;33m)\u001b[0m\u001b[1;33m\u001b[0m\u001b[1;33m\u001b[0m\u001b[0m\n",
      "\u001b[1;32mD:\\Users\\HP\\anaconda3\\lib\\site-packages\\tensorflow\\python\\training\\tracking\\base.py\u001b[0m in \u001b[0;36m_method_wrapper\u001b[1;34m(self, *args, **kwargs)\u001b[0m\n\u001b[0;32m    455\u001b[0m     \u001b[0mself\u001b[0m\u001b[1;33m.\u001b[0m\u001b[0m_self_setattr_tracking\u001b[0m \u001b[1;33m=\u001b[0m \u001b[1;32mFalse\u001b[0m  \u001b[1;31m# pylint: disable=protected-access\u001b[0m\u001b[1;33m\u001b[0m\u001b[1;33m\u001b[0m\u001b[0m\n\u001b[0;32m    456\u001b[0m     \u001b[1;32mtry\u001b[0m\u001b[1;33m:\u001b[0m\u001b[1;33m\u001b[0m\u001b[1;33m\u001b[0m\u001b[0m\n\u001b[1;32m--> 457\u001b[1;33m       \u001b[0mresult\u001b[0m \u001b[1;33m=\u001b[0m \u001b[0mmethod\u001b[0m\u001b[1;33m(\u001b[0m\u001b[0mself\u001b[0m\u001b[1;33m,\u001b[0m \u001b[1;33m*\u001b[0m\u001b[0margs\u001b[0m\u001b[1;33m,\u001b[0m \u001b[1;33m**\u001b[0m\u001b[0mkwargs\u001b[0m\u001b[1;33m)\u001b[0m\u001b[1;33m\u001b[0m\u001b[1;33m\u001b[0m\u001b[0m\n\u001b[0m\u001b[0;32m    458\u001b[0m     \u001b[1;32mfinally\u001b[0m\u001b[1;33m:\u001b[0m\u001b[1;33m\u001b[0m\u001b[1;33m\u001b[0m\u001b[0m\n\u001b[0;32m    459\u001b[0m       \u001b[0mself\u001b[0m\u001b[1;33m.\u001b[0m\u001b[0m_self_setattr_tracking\u001b[0m \u001b[1;33m=\u001b[0m \u001b[0mprevious_value\u001b[0m  \u001b[1;31m# pylint: disable=protected-access\u001b[0m\u001b[1;33m\u001b[0m\u001b[1;33m\u001b[0m\u001b[0m\n",
      "\u001b[1;32mD:\\Users\\HP\\anaconda3\\lib\\site-packages\\tensorflow\\python\\keras\\engine\\functional.py\u001b[0m in \u001b[0;36m__init__\u001b[1;34m(self, inputs, outputs, name, trainable)\u001b[0m\n\u001b[0;32m    113\u001b[0m     \u001b[1;31m#     'arguments during initialization. Got an unexpected argument:')\u001b[0m\u001b[1;33m\u001b[0m\u001b[1;33m\u001b[0m\u001b[1;33m\u001b[0m\u001b[0m\n\u001b[0;32m    114\u001b[0m     \u001b[0msuper\u001b[0m\u001b[1;33m(\u001b[0m\u001b[0mFunctional\u001b[0m\u001b[1;33m,\u001b[0m \u001b[0mself\u001b[0m\u001b[1;33m)\u001b[0m\u001b[1;33m.\u001b[0m\u001b[0m__init__\u001b[0m\u001b[1;33m(\u001b[0m\u001b[0mname\u001b[0m\u001b[1;33m=\u001b[0m\u001b[0mname\u001b[0m\u001b[1;33m,\u001b[0m \u001b[0mtrainable\u001b[0m\u001b[1;33m=\u001b[0m\u001b[0mtrainable\u001b[0m\u001b[1;33m)\u001b[0m\u001b[1;33m\u001b[0m\u001b[1;33m\u001b[0m\u001b[0m\n\u001b[1;32m--> 115\u001b[1;33m     \u001b[0mself\u001b[0m\u001b[1;33m.\u001b[0m\u001b[0m_init_graph_network\u001b[0m\u001b[1;33m(\u001b[0m\u001b[0minputs\u001b[0m\u001b[1;33m,\u001b[0m \u001b[0moutputs\u001b[0m\u001b[1;33m)\u001b[0m\u001b[1;33m\u001b[0m\u001b[1;33m\u001b[0m\u001b[0m\n\u001b[0m\u001b[0;32m    116\u001b[0m \u001b[1;33m\u001b[0m\u001b[0m\n\u001b[0;32m    117\u001b[0m   \u001b[1;33m@\u001b[0m\u001b[0mtrackable\u001b[0m\u001b[1;33m.\u001b[0m\u001b[0mno_automatic_dependency_tracking\u001b[0m\u001b[1;33m\u001b[0m\u001b[1;33m\u001b[0m\u001b[0m\n",
      "\u001b[1;32mD:\\Users\\HP\\anaconda3\\lib\\site-packages\\tensorflow\\python\\training\\tracking\\base.py\u001b[0m in \u001b[0;36m_method_wrapper\u001b[1;34m(self, *args, **kwargs)\u001b[0m\n\u001b[0;32m    455\u001b[0m     \u001b[0mself\u001b[0m\u001b[1;33m.\u001b[0m\u001b[0m_self_setattr_tracking\u001b[0m \u001b[1;33m=\u001b[0m \u001b[1;32mFalse\u001b[0m  \u001b[1;31m# pylint: disable=protected-access\u001b[0m\u001b[1;33m\u001b[0m\u001b[1;33m\u001b[0m\u001b[0m\n\u001b[0;32m    456\u001b[0m     \u001b[1;32mtry\u001b[0m\u001b[1;33m:\u001b[0m\u001b[1;33m\u001b[0m\u001b[1;33m\u001b[0m\u001b[0m\n\u001b[1;32m--> 457\u001b[1;33m       \u001b[0mresult\u001b[0m \u001b[1;33m=\u001b[0m \u001b[0mmethod\u001b[0m\u001b[1;33m(\u001b[0m\u001b[0mself\u001b[0m\u001b[1;33m,\u001b[0m \u001b[1;33m*\u001b[0m\u001b[0margs\u001b[0m\u001b[1;33m,\u001b[0m \u001b[1;33m**\u001b[0m\u001b[0mkwargs\u001b[0m\u001b[1;33m)\u001b[0m\u001b[1;33m\u001b[0m\u001b[1;33m\u001b[0m\u001b[0m\n\u001b[0m\u001b[0;32m    458\u001b[0m     \u001b[1;32mfinally\u001b[0m\u001b[1;33m:\u001b[0m\u001b[1;33m\u001b[0m\u001b[1;33m\u001b[0m\u001b[0m\n\u001b[0;32m    459\u001b[0m       \u001b[0mself\u001b[0m\u001b[1;33m.\u001b[0m\u001b[0m_self_setattr_tracking\u001b[0m \u001b[1;33m=\u001b[0m \u001b[0mprevious_value\u001b[0m  \u001b[1;31m# pylint: disable=protected-access\u001b[0m\u001b[1;33m\u001b[0m\u001b[1;33m\u001b[0m\u001b[0m\n",
      "\u001b[1;32mD:\\Users\\HP\\anaconda3\\lib\\site-packages\\tensorflow\\python\\keras\\engine\\functional.py\u001b[0m in \u001b[0;36m_init_graph_network\u001b[1;34m(self, inputs, outputs)\u001b[0m\n\u001b[0;32m    188\u001b[0m \u001b[1;33m\u001b[0m\u001b[0m\n\u001b[0;32m    189\u001b[0m     \u001b[1;31m# Keep track of the network's nodes and layers.\u001b[0m\u001b[1;33m\u001b[0m\u001b[1;33m\u001b[0m\u001b[1;33m\u001b[0m\u001b[0m\n\u001b[1;32m--> 190\u001b[1;33m     nodes, nodes_by_depth, layers, _ = _map_graph_network(\n\u001b[0m\u001b[0;32m    191\u001b[0m         self.inputs, self.outputs)\n\u001b[0;32m    192\u001b[0m     \u001b[0mself\u001b[0m\u001b[1;33m.\u001b[0m\u001b[0m_network_nodes\u001b[0m \u001b[1;33m=\u001b[0m \u001b[0mnodes\u001b[0m\u001b[1;33m\u001b[0m\u001b[1;33m\u001b[0m\u001b[0m\n",
      "\u001b[1;32mD:\\Users\\HP\\anaconda3\\lib\\site-packages\\tensorflow\\python\\keras\\engine\\functional.py\u001b[0m in \u001b[0;36m_map_graph_network\u001b[1;34m(inputs, outputs)\u001b[0m\n\u001b[0;32m    924\u001b[0m         \u001b[1;32mfor\u001b[0m \u001b[0mx\u001b[0m \u001b[1;32min\u001b[0m \u001b[0mnest\u001b[0m\u001b[1;33m.\u001b[0m\u001b[0mflatten\u001b[0m\u001b[1;33m(\u001b[0m\u001b[0mnode\u001b[0m\u001b[1;33m.\u001b[0m\u001b[0mkeras_inputs\u001b[0m\u001b[1;33m)\u001b[0m\u001b[1;33m:\u001b[0m\u001b[1;33m\u001b[0m\u001b[1;33m\u001b[0m\u001b[0m\n\u001b[0;32m    925\u001b[0m           \u001b[1;32mif\u001b[0m \u001b[0mid\u001b[0m\u001b[1;33m(\u001b[0m\u001b[0mx\u001b[0m\u001b[1;33m)\u001b[0m \u001b[1;32mnot\u001b[0m \u001b[1;32min\u001b[0m \u001b[0mcomputable_tensors\u001b[0m\u001b[1;33m:\u001b[0m\u001b[1;33m\u001b[0m\u001b[1;33m\u001b[0m\u001b[0m\n\u001b[1;32m--> 926\u001b[1;33m             raise ValueError('Graph disconnected: '\n\u001b[0m\u001b[0;32m    927\u001b[0m                              \u001b[1;34m'cannot obtain value for tensor '\u001b[0m \u001b[1;33m+\u001b[0m \u001b[0mstr\u001b[0m\u001b[1;33m(\u001b[0m\u001b[0mx\u001b[0m\u001b[1;33m)\u001b[0m \u001b[1;33m+\u001b[0m\u001b[1;33m\u001b[0m\u001b[1;33m\u001b[0m\u001b[0m\n\u001b[0;32m    928\u001b[0m                              \u001b[1;34m' at layer \"'\u001b[0m \u001b[1;33m+\u001b[0m \u001b[0mlayer\u001b[0m\u001b[1;33m.\u001b[0m\u001b[0mname\u001b[0m \u001b[1;33m+\u001b[0m \u001b[1;34m'\". '\u001b[0m\u001b[1;33m\u001b[0m\u001b[1;33m\u001b[0m\u001b[0m\n",
      "\u001b[1;31mValueError\u001b[0m: Graph disconnected: cannot obtain value for tensor Tensor(\"dense_7_input:0\", shape=(None, 36), dtype=float32) at layer \"dense_7\". The following previous layers were accessed without issue: []"
     ]
    }
   ],
   "source": [
    "fig, axes = plt.subplots(1,figsize=(7,5))\n",
    "inject_fake_hidden = K.function([model.layers[4].input],\n",
    "                                [model.layers[0].output])\n",
    "\n",
    "inject_fake_hidden([hidden_layer_output])\n",
    "\n",
    "if True:\n",
    "    test_a = gm.getMapsRangeWindowed(4, 0, 1)\n",
    "    test_at = test_a.transpose()\n",
    "    hidden_layer_output = get_hidden_layer_output([test_at])[0]\n",
    "    plt.scatter(hidden_layer_output[:,0], hidden_layer_output[:,1], alpha=1)\n",
    "\n",
    "if True:\n",
    "    test_a = gm.getMapsRangeWindowed(4, 1, 2)\n",
    "    test_at = test_a.transpose()\n",
    "    hidden_layer_output = get_hidden_layer_output([test_at])[0]\n",
    "    plt.scatter(hidden_layer_output[:,0], hidden_layer_output[:,1], color=\"green\", alpha=1)\n",
    "\n",
    "if True:\n",
    "    test_a = gm.getMapsRangeWindowed(4, 2, 3)\n",
    "    test_at = test_a.transpose()\n",
    "    hidden_layer_output = get_hidden_layer_output([test_at])[0]\n",
    "    plt.scatter(hidden_layer_output[:,0], hidden_layer_output[:,1], color=\"red\", alpha=1)\n",
    "\n",
    "if False:\n",
    "    test_a = gm.getMapsRangeWindowed(4, 3, 4)\n",
    "    test_at = test_a.transpose()\n",
    "    hidden_layer_output = get_hidden_layer_output([test_at])[0]\n",
    "    plt.scatter(hidden_layer_output[:,0], hidden_layer_output[:,1], color=\"orange\", alpha=.5)\n",
    "\n"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 29,
   "metadata": {
    "collapsed": false
   },
   "outputs": [
    {
     "data": {
      "text/plain": [
       "<matplotlib.collections.PathCollection at 0x19d4084d8b0>"
      ]
     },
     "execution_count": 29,
     "metadata": {},
     "output_type": "execute_result"
    },
    {
     "data": {
      "image/png": "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",
      "text/plain": [
       "<Figure size 504x360 with 1 Axes>"
      ]
     },
     "metadata": {
      "needs_background": "light"
     },
     "output_type": "display_data"
    }
   ],
   "source": [
    "fig, axes = plt.subplots(1,figsize=(7,5))\r\n",
    "\r\n",
    "get_hidden_layer_output = K.function([model.layers[0].input],\r\n",
    "                                  [model.layers[4].output])\r\n",
    "\r\n",
    "test_a = gm.getMapsRangeWindowed(4, 2, 3)\r\n",
    "test_at = test_a.transpose()\r\n",
    "hidden_layer_output = get_hidden_layer_output([test_at])[0]\r\n",
    "plt.scatter(hidden_layer_output[:,0], hidden_layer_output[:,1], alpha=.5)\r\n",
    "\r\n",
    "if False:\r\n",
    "    test_a = gm.getMapsRangeWindowed(4, 1, 2)\r\n",
    "    test_at = test_a.transpose()\r\n",
    "    hidden_layer_output = get_hidden_layer_output([test_at])[0]\r\n",
    "    plt.scatter(hidden_layer_output[:,0], hidden_layer_output[:,1], color=\"green\", alpha=.5)\r\n",
    "\r\n",
    "if False:\r\n",
    "    test_a = gm.getMapsRangeWindowed(4, 2, 3)\r\n",
    "    test_at = test_a.transpose()\r\n",
    "    hidden_layer_output = get_hidden_layer_output([test_at])[0]\r\n",
    "    plt.scatter(hidden_layer_output[:,0], hidden_layer_output[:,1], color=\"red\", alpha=.5)\r\n",
    "\r\n",
    "\r\n",
    "test_a = gm.getMapsRangeWindowed(4, 3, 4)\r\n",
    "test_at = test_a.transpose()\r\n",
    "hidden_layer_output = get_hidden_layer_output([test_at])[0]\r\n",
    "plt.scatter(hidden_layer_output[:,0], hidden_layer_output[:,1], color=\"orange\", alpha=.5)\r\n",
    "\r\n"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 33,
   "metadata": {
    "collapsed": false
   },
   "outputs": [
    {
     "data": {
      "text/plain": [
       "<matplotlib.collections.PathCollection at 0x19d4314c6a0>"
      ]
     },
     "execution_count": 33,
     "metadata": {},
     "output_type": "execute_result"
    },
    {
     "data": {
      "image/png": "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",
      "text/plain": [
       "<Figure size 504x360 with 1 Axes>"
      ]
     },
     "metadata": {
      "needs_background": "light"
     },
     "output_type": "display_data"
    }
   ],
   "source": [
    "fig, axes = plt.subplots(1,figsize=(7,5))\r\n",
    "\r\n",
    "get_hidden_layer_output = K.function([model.layers[0].input],\r\n",
    "                                  [model.layers[1].output])\r\n",
    "\r\n",
    "test_a = gm.getMapsRangeWindowed(4, 1, 2)\r\n",
    "test_at = test_a.transpose()\r\n",
    "hidden_layer_output = get_hidden_layer_output([test_at])[0]\r\n",
    "plt.scatter(hidden_layer_output[:,0], hidden_layer_output[:,1], color=\"black\", alpha=.5)\r\n",
    "\r\n",
    "if True:\r\n",
    "    test_a = gm.getMapsRangeWindowed(4, 0, 1)\r\n",
    "    test_at = test_a.transpose()\r\n",
    "    hidden_layer_output = get_hidden_layer_output([test_at])[0]\r\n",
    "    plt.scatter(hidden_layer_output[:,0], hidden_layer_output[:,1], color=\"green\", alpha=.5)\r\n",
    "\r\n",
    "if True:\r\n",
    "    test_a = gm.getMapsRangeWindowed(4, 3, 4)\r\n",
    "    test_at = test_a.transpose()\r\n",
    "    hidden_layer_output = get_hidden_layer_output([test_at])[0]\r\n",
    "    plt.scatter(hidden_layer_output[:,0], hidden_layer_output[:,1], color=\"red\", alpha=.5)\r\n",
    "\r\n",
    "\r\n",
    "test_a = gm.getMapsRangeWindowed(4, 4, 5)\r\n",
    "test_at = test_a.transpose()\r\n",
    "hidden_layer_output = get_hidden_layer_output([test_at])[0]\r\n",
    "plt.scatter(hidden_layer_output[:,0], hidden_layer_output[:,1], color=\"orange\", alpha=.5)\r\n",
    "\r\n"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 38,
   "metadata": {
    "collapsed": false
   },
   "outputs": [
    {
     "data": {
      "text/plain": [
       "<matplotlib.collections.PathCollection at 0x19d405fcbb0>"
      ]
     },
     "execution_count": 38,
     "metadata": {},
     "output_type": "execute_result"
    },
    {
     "data": {
      "image/png": "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",
      "text/plain": [
       "<Figure size 504x360 with 1 Axes>"
      ]
     },
     "metadata": {
      "needs_background": "light"
     },
     "output_type": "display_data"
    }
   ],
   "source": [
    "fig, axes = plt.subplots(1,figsize=(7,5))\r\n",
    "\r\n",
    "get_hidden_layer_output = K.function([model.layers[0].input],\r\n",
    "                                  [model.layers[1].output])\r\n",
    "\r\n",
    "test_a = gm.getMapsRangeWindowed(4, 2, 3)\r\n",
    "test_at = test_a.transpose()\r\n",
    "hidden_layer_output = get_hidden_layer_output([test_at])[0]\r\n",
    "plt.scatter(hidden_layer_output[:,0], hidden_layer_output[:,1], color=\"black\",alpha=.5)\r\n",
    "\r\n",
    "if False:\r\n",
    "    test_a = gm.getMapsRangeWindowed(4, 1, 2)\r\n",
    "    test_at = test_a.transpose()\r\n",
    "    hidden_layer_output = get_hidden_layer_output([test_at])[0]\r\n",
    "    plt.scatter(hidden_layer_output[:,0], hidden_layer_output[:,1], color=\"green\", alpha=.5)\r\n",
    "\r\n",
    "if False:\r\n",
    "    test_a = gm.getMapsRangeWindowed(4, 2, 3)\r\n",
    "    test_at = test_a.transpose()\r\n",
    "    hidden_layer_output = get_hidden_layer_output([test_at])[0]\r\n",
    "    plt.scatter(hidden_layer_output[:,0], hidden_layer_output[:,1], color=\"red\", alpha=.5)\r\n",
    "\r\n",
    "\r\n",
    "test_a = gm.getMapsRangeWindowed(4, 4, 5)\r\n",
    "test_at = test_a.transpose()\r\n",
    "hidden_layer_output = get_hidden_layer_output([test_at])[0]\r\n",
    "plt.scatter(hidden_layer_output[:,0], hidden_layer_output[:,1], color=\"orange\", alpha=.5)\r\n",
    "\r\n"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {
    "collapsed": true
   },
   "outputs": [],
   "source": []
  }
 ],
 "metadata": {
  "interpreter": {
   "hash": "71ff158078b9094704a9ca0f88dd88feb63a8763717434d1ce66e321efc9e249"
  },
  "kernelspec": {
   "display_name": "Python 3.8.8 64-bit ('base': conda)",
   "name": "python3"
  },
  "language_info": {
   "name": "python",
   "version": ""
  }
 },
 "nbformat": 4,
 "nbformat_minor": 0
}