learn_identity_recurrent_with_input.ipynb

{
 "metadata": {
  "name": "",
  "signature": "sha256:24374023008cacb97228d52969fdb363913e1c89fdf58e3c406b3eeafe05d60c"
 },
 "nbformat": 3,
 "nbformat_minor": 0,
 "worksheets": [
  {
   "cells": [
    {
     "cell_type": "code",
     "collapsed": false,
     "input": [
      "# Import here\n",
      "%matplotlib inline\n",
      "import matplotlib.pylab as pylab\n",
      "import matplotlib.pyplot as plt\n",
      "import numpy as np\n",
      "import theano\n",
      "import theano.tensor as T\n",
      "\n",
      "pylab.rcParams['figure.figsize'] = (9.0, 6.0)"
     ],
     "language": "python",
     "metadata": {},
     "outputs": [],
     "prompt_number": 1
    },
    {
     "cell_type": "code",
     "collapsed": false,
     "input": [
      "n_steps = 100\n",
      "x = T.fmatrix(name='x')\n",
      "S = T.fmatrix(name='S')\n",
      "W = T.fmatrix(name='W')\n",
      "b_eff = T.dot(W, x)\n",
      "def one_step(y, S, b_eff):\n",
      "    return T.dot(S, y) + b_eff\n",
      "y_seq, updates = theano.scan(fn=one_step,\n",
      "                             outputs_info=T.zeros_like(b_eff),\n",
      "                             non_sequences=[S, b_eff],\n",
      "                             n_steps=n_steps)\n",
      "y=y_seq[-1]\n",
      "cost = T.mean(T.sum((x - y) ** 2, axis=0, keepdims=True))\n",
      "dS = T.grad(cost, S)\n",
      "dW = T.grad(cost, W)\n",
      "f = theano.function(inputs=[x, S, W],\n",
      "                    outputs=[cost, dS, dW, y],\n",
      "                    updates=updates)"
     ],
     "language": "python",
     "metadata": {},
     "outputs": [],
     "prompt_number": 17
    },
    {
     "cell_type": "code",
     "collapsed": false,
     "input": [
      "dim_x = 64\n",
      "dim_batch = 256\n",
      "val_S = np.eye(dim_x, dtype=np.float32)\n",
      "val_W = np.random.randn(dim_x, dim_x).astype(np.float32) / np.sqrt(dim_x) / n_steps"
     ],
     "language": "python",
     "metadata": {},
     "outputs": [],
     "prompt_number": 22
    },
    {
     "cell_type": "code",
     "collapsed": false,
     "input": [
      "eta_W = 0.1 / dim_batch\n",
      "eta_S = 0.1 / dim_batch / n_steps\n",
      "\n",
      "for t in range(1000):\n",
      "    val_x = np.random.randn(dim_x, dim_batch).astype(np.float32)\n",
      "    val_cost, val_dS, val_dW, val_y = f(val_x, val_S, val_W)\n",
      "    if (t + 1) % 10 == 0:\n",
      "        print (t + 1), val_cost\n",
      "    val_S -= eta_S * val_dS\n",
      "    val_W -= eta_W * val_dW"
     ],
     "language": "python",
     "metadata": {},
     "outputs": [
      {
       "output_type": "stream",
       "stream": "stdout",
       "text": [
        "10 70.2283554077\n",
        "20"
       ]
      },
      {
       "output_type": "stream",
       "stream": "stdout",
       "text": [
        " 58.7585639954\n",
        "30"
       ]
      },
      {
       "output_type": "stream",
       "stream": "stdout",
       "text": [
        " 51.4939842224\n",
        "40"
       ]
      },
      {
       "output_type": "stream",
       "stream": "stdout",
       "text": [
        " 43.9743728638\n",
        "50"
       ]
      },
      {
       "output_type": "stream",
       "stream": "stdout",
       "text": [
        " 39.5721511841\n",
        "60"
       ]
      },
      {
       "output_type": "stream",
       "stream": "stdout",
       "text": [
        " 34.2540283203\n",
        "70"
       ]
      },
      {
       "output_type": "stream",
       "stream": "stdout",
       "text": [
        " 28.9766349792\n",
        "80"
       ]
      },
      {
       "output_type": "stream",
       "stream": "stdout",
       "text": [
        " 26.5210094452\n",
        "90"
       ]
      },
      {
       "output_type": "stream",
       "stream": "stdout",
       "text": [
        " 23.2932872772\n",
        "100"
       ]
      },
      {
       "output_type": "stream",
       "stream": "stdout",
       "text": [
        " 21.7147140503\n",
        "110"
       ]
      },
      {
       "output_type": "stream",
       "stream": "stdout",
       "text": [
        " 19.7919521332\n",
        "120"
       ]
      },
      {
       "output_type": "stream",
       "stream": "stdout",
       "text": [
        " 18.0458621979\n",
        "130"
       ]
      },
      {
       "output_type": "stream",
       "stream": "stdout",
       "text": [
        " 15.7576503754\n",
        "140"
       ]
      },
      {
       "output_type": "stream",
       "stream": "stdout",
       "text": [
        " 14.2348108292\n",
        "150"
       ]
      },
      {
       "output_type": "stream",
       "stream": "stdout",
       "text": [
        " 12.7208366394\n",
        "160"
       ]
      },
      {
       "output_type": "stream",
       "stream": "stdout",
       "text": [
        " 11.2104034424\n",
        "170"
       ]
      },
      {
       "output_type": "stream",
       "stream": "stdout",
       "text": [
        " 11.1228265762\n",
        "180"
       ]
      },
      {
       "output_type": "stream",
       "stream": "stdout",
       "text": [
        " 10.1094770432\n",
        "190"
       ]
      },
      {
       "output_type": "stream",
       "stream": "stdout",
       "text": [
        " 8.88823795319\n",
        "200"
       ]
      },
      {
       "output_type": "stream",
       "stream": "stdout",
       "text": [
        " 8.09349822998\n",
        "210"
       ]
      },
      {
       "output_type": "stream",
       "stream": "stdout",
       "text": [
        " 7.41362094879\n",
        "220"
       ]
      },
      {
       "output_type": "stream",
       "stream": "stdout",
       "text": [
        " 6.91442012787\n",
        "230"
       ]
      },
      {
       "output_type": "stream",
       "stream": "stdout",
       "text": [
        " 6.06851100922\n",
        "240"
       ]
      },
      {
       "output_type": "stream",
       "stream": "stdout",
       "text": [
        " 5.5574092865\n",
        "250"
       ]
      },
      {
       "output_type": "stream",
       "stream": "stdout",
       "text": [
        " 4.96092796326\n",
        "260"
       ]
      },
      {
       "output_type": "stream",
       "stream": "stdout",
       "text": [
        " 4.79374551773\n",
        "270"
       ]
      },
      {
       "output_type": "stream",
       "stream": "stdout",
       "text": [
        " 4.8021364212\n",
        "280"
       ]
      },
      {
       "output_type": "stream",
       "stream": "stdout",
       "text": [
        " 4.51977014542\n",
        "290"
       ]
      },
      {
       "output_type": "stream",
       "stream": "stdout",
       "text": [
        " 4.24153661728\n",
        "300"
       ]
      },
      {
       "output_type": "stream",
       "stream": "stdout",
       "text": [
        " 3.95704865456\n",
        "310"
       ]
      },
      {
       "output_type": "stream",
       "stream": "stdout",
       "text": [
        " 3.62325525284\n",
        "320"
       ]
      },
      {
       "output_type": "stream",
       "stream": "stdout",
       "text": [
        " 3.59461092949\n",
        "330"
       ]
      },
      {
       "output_type": "stream",
       "stream": "stdout",
       "text": [
        " 3.16009163857\n",
        "340"
       ]
      },
      {
       "output_type": "stream",
       "stream": "stdout",
       "text": [
        " 2.90127444267\n",
        "350"
       ]
      },
      {
       "output_type": "stream",
       "stream": "stdout",
       "text": [
        " 2.45963668823\n",
        "360"
       ]
      },
      {
       "output_type": "stream",
       "stream": "stdout",
       "text": [
        " 2.49890470505\n",
        "370"
       ]
      },
      {
       "output_type": "stream",
       "stream": "stdout",
       "text": [
        " 2.35923743248\n",
        "380"
       ]
      },
      {
       "output_type": "stream",
       "stream": "stdout",
       "text": [
        " 2.15628027916\n",
        "390"
       ]
      },
      {
       "output_type": "stream",
       "stream": "stdout",
       "text": [
        " 2.08569288254\n",
        "400"
       ]
      },
      {
       "output_type": "stream",
       "stream": "stdout",
       "text": [
        " 2.00370645523\n",
        "410"
       ]
      },
      {
       "output_type": "stream",
       "stream": "stdout",
       "text": [
        " 1.69234800339\n",
        "420"
       ]
      },
      {
       "output_type": "stream",
       "stream": "stdout",
       "text": [
        " 1.95856952667\n",
        "430"
       ]
      },
      {
       "output_type": "stream",
       "stream": "stdout",
       "text": [
        " 1.63814485073\n",
        "440"
       ]
      },
      {
       "output_type": "stream",
       "stream": "stdout",
       "text": [
        " 1.58379006386\n",
        "450"
       ]
      },
      {
       "output_type": "stream",
       "stream": "stdout",
       "text": [
        " 1.54001557827\n",
        "460"
       ]
      },
      {
       "output_type": "stream",
       "stream": "stdout",
       "text": [
        " 1.41342389584\n",
        "470"
       ]
      },
      {
       "output_type": "stream",
       "stream": "stdout",
       "text": [
        " 1.24092912674\n",
        "480"
       ]
      },
      {
       "output_type": "stream",
       "stream": "stdout",
       "text": [
        " 1.19672417641\n",
        "490"
       ]
      },
      {
       "output_type": "stream",
       "stream": "stdout",
       "text": [
        " 1.10880458355\n",
        "500"
       ]
      },
      {
       "output_type": "stream",
       "stream": "stdout",
       "text": [
        " 1.07253813744\n",
        "510"
       ]
      },
      {
       "output_type": "stream",
       "stream": "stdout",
       "text": [
        " 1.03193199635\n",
        "520"
       ]
      },
      {
       "output_type": "stream",
       "stream": "stdout",
       "text": [
        " 1.08108055592\n",
        "530"
       ]
      },
      {
       "output_type": "stream",
       "stream": "stdout",
       "text": [
        " 0.955024003983\n",
        "540"
       ]
      },
      {
       "output_type": "stream",
       "stream": "stdout",
       "text": [
        " 0.866100728512\n",
        "550"
       ]
      },
      {
       "output_type": "stream",
       "stream": "stdout",
       "text": [
        " 0.861440658569\n",
        "560"
       ]
      },
      {
       "output_type": "stream",
       "stream": "stdout",
       "text": [
        " 0.696216404438\n",
        "570"
       ]
      },
      {
       "output_type": "stream",
       "stream": "stdout",
       "text": [
        " 0.733282744884\n",
        "580"
       ]
      },
      {
       "output_type": "stream",
       "stream": "stdout",
       "text": [
        " 0.767969310284\n",
        "590"
       ]
      },
      {
       "output_type": "stream",
       "stream": "stdout",
       "text": [
        " 0.59978967905\n",
        "600"
       ]
      },
      {
       "output_type": "stream",
       "stream": "stdout",
       "text": [
        " 0.645955085754\n",
        "610"
       ]
      },
      {
       "output_type": "stream",
       "stream": "stdout",
       "text": [
        " 0.628410041332\n",
        "620"
       ]
      },
      {
       "output_type": "stream",
       "stream": "stdout",
       "text": [
        " 0.528648495674\n",
        "630"
       ]
      },
      {
       "output_type": "stream",
       "stream": "stdout",
       "text": [
        " 0.514638841152\n",
        "640"
       ]
      },
      {
       "output_type": "stream",
       "stream": "stdout",
       "text": [
        " 0.477892398834\n",
        "650"
       ]
      },
      {
       "output_type": "stream",
       "stream": "stdout",
       "text": [
        " 0.445438295603\n",
        "660"
       ]
      },
      {
       "output_type": "stream",
       "stream": "stdout",
       "text": [
        " 0.516854822636\n",
        "670"
       ]
      },
      {
       "output_type": "stream",
       "stream": "stdout",
       "text": [
        " 0.46590834856\n",
        "680"
       ]
      },
      {
       "output_type": "stream",
       "stream": "stdout",
       "text": [
        " 0.402379214764\n",
        "690"
       ]
      },
      {
       "output_type": "stream",
       "stream": "stdout",
       "text": [
        " 0.391501277685\n",
        "700"
       ]
      },
      {
       "output_type": "stream",
       "stream": "stdout",
       "text": [
        " 0.379371553659\n",
        "710"
       ]
      },
      {
       "output_type": "stream",
       "stream": "stdout",
       "text": [
        " 0.411156445742\n",
        "720"
       ]
      },
      {
       "output_type": "stream",
       "stream": "stdout",
       "text": [
        " 0.332494288683\n",
        "730"
       ]
      },
      {
       "output_type": "stream",
       "stream": "stdout",
       "text": [
        " 0.268220216036\n",
        "740"
       ]
      },
      {
       "output_type": "stream",
       "stream": "stdout",
       "text": [
        " 0.288221120834\n",
        "750"
       ]
      },
      {
       "output_type": "stream",
       "stream": "stdout",
       "text": [
        " 0.266344666481\n",
        "760"
       ]
      },
      {
       "output_type": "stream",
       "stream": "stdout",
       "text": [
        " 0.277551949024\n",
        "770"
       ]
      },
      {
       "output_type": "stream",
       "stream": "stdout",
       "text": [
        " 0.27564522624\n",
        "780"
       ]
      },
      {
       "output_type": "stream",
       "stream": "stdout",
       "text": [
        " 0.227135062218\n",
        "790"
       ]
      },
      {
       "output_type": "stream",
       "stream": "stdout",
       "text": [
        " 0.257936567068\n",
        "800"
       ]
      },
      {
       "output_type": "stream",
       "stream": "stdout",
       "text": [
        " 0.224724963307\n",
        "810"
       ]
      },
      {
       "output_type": "stream",
       "stream": "stdout",
       "text": [
        " 0.210002377629\n",
        "820"
       ]
      },
      {
       "output_type": "stream",
       "stream": "stdout",
       "text": [
        " 0.200717478991\n",
        "830"
       ]
      },
      {
       "output_type": "stream",
       "stream": "stdout",
       "text": [
        " 0.178926870227\n",
        "840"
       ]
      },
      {
       "output_type": "stream",
       "stream": "stdout",
       "text": [
        " 0.180963680148\n",
        "850"
       ]
      },
      {
       "output_type": "stream",
       "stream": "stdout",
       "text": [
        " 0.176275461912\n",
        "860"
       ]
      },
      {
       "output_type": "stream",
       "stream": "stdout",
       "text": [
        " 0.152346223593\n",
        "870"
       ]
      },
      {
       "output_type": "stream",
       "stream": "stdout",
       "text": [
        " 0.147061794996\n",
        "880"
       ]
      },
      {
       "output_type": "stream",
       "stream": "stdout",
       "text": [
        " 0.170192927122\n",
        "890"
       ]
      },
      {
       "output_type": "stream",
       "stream": "stdout",
       "text": [
        " 0.131817892194\n",
        "900"
       ]
      },
      {
       "output_type": "stream",
       "stream": "stdout",
       "text": [
        " 0.151087611914\n",
        "910"
       ]
      },
      {
       "output_type": "stream",
       "stream": "stdout",
       "text": [
        " 0.133023500443\n",
        "920"
       ]
      },
      {
       "output_type": "stream",
       "stream": "stdout",
       "text": [
        " 0.137734174728\n",
        "930"
       ]
      },
      {
       "output_type": "stream",
       "stream": "stdout",
       "text": [
        " 0.101545706391\n",
        "940"
       ]
      },
      {
       "output_type": "stream",
       "stream": "stdout",
       "text": [
        " 0.106590211391\n",
        "950"
       ]
      },
      {
       "output_type": "stream",
       "stream": "stdout",
       "text": [
        " 0.124645128846\n",
        "960"
       ]
      },
      {
       "output_type": "stream",
       "stream": "stdout",
       "text": [
        " 0.0900602787733\n",
        "970"
       ]
      },
      {
       "output_type": "stream",
       "stream": "stdout",
       "text": [
        " 0.0960389003158\n",
        "980"
       ]
      },
      {
       "output_type": "stream",
       "stream": "stdout",
       "text": [
        " 0.0834612026811\n",
        "990"
       ]
      },
      {
       "output_type": "stream",
       "stream": "stdout",
       "text": [
        " 0.0895466953516\n",
        "1000"
       ]
      },
      {
       "output_type": "stream",
       "stream": "stdout",
       "text": [
        " 0.0854084566236\n"
       ]
      }
     ],
     "prompt_number": 24
    },
    {
     "cell_type": "code",
     "collapsed": false,
     "input": [
      "print val_S\n",
      "plt.imshow(val_S, interpolation='None')"
     ],
     "language": "python",
     "metadata": {},
     "outputs": [
      {
       "output_type": "stream",
       "stream": "stdout",
       "text": [
        "[[  7.91551828e-01   1.09368796e-02   8.90446454e-03 ...,   9.46536602e-04\n",
        "   -1.09681813e-02  -2.88536400e-02]\n",
        " [ -4.43507422e-04   7.99930394e-01  -9.05676279e-03 ...,  -4.71686348e-02\n",
        "    2.56220158e-02  -1.41058501e-03]\n",
        " [  2.22858787e-03  -2.90666968e-02   8.28683913e-01 ...,  -1.49826519e-03\n",
        "   -1.80445565e-03  -1.26732010e-02]\n",
        " ..., \n",
        " [  3.53699573e-03  -3.98051888e-02   1.74171582e-03 ...,   8.09756994e-01\n",
        "    9.91128106e-03  -9.01053566e-03]\n",
        " [ -1.42429871e-02   3.22850645e-02  -6.38222974e-03 ...,   6.68446906e-03\n",
        "    8.07649314e-01  -2.28243377e-02]\n",
        " [ -2.03520227e-02   1.53767061e-03  -3.90229234e-03 ...,  -9.68658552e-03\n",
        "   -2.98266788e-03   8.06548715e-01]]\n"
       ]
      },
      {
       "metadata": {},
       "output_type": "pyout",
       "prompt_number": 25,
       "text": [
        "<matplotlib.image.AxesImage at 0x114877f50>"
       ]
      },
      {
       "metadata": {},
       "output_type": "display_data",
       "png": "iVBORw0KGgoAAAANSUhEUgAAAWwAAAFtCAYAAAAjw1AcAAAABHNCSVQICAgIfAhkiAAAAAlwSFlz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\nRK5CYII=\n",
       "text": [
        "<matplotlib.figure.Figure at 0x113a20e90>"
       ]
      }
     ],
     "prompt_number": 25
    },
    {
     "cell_type": "code",
     "collapsed": false,
     "input": [
      "print val_W\n",
      "plt.imshow(val_W, interpolation='None')"
     ],
     "language": "python",
     "metadata": {},
     "outputs": [
      {
       "output_type": "stream",
       "stream": "stdout",
       "text": [
        "[[  1.94576979e-01  -9.00759269e-03  -8.84541078e-04 ...,   1.66932924e-03\n",
        "    1.23289805e-02   2.62517035e-02]\n",
        " [  4.65040193e-06   1.98400542e-01   7.54676294e-03 ...,   4.57942858e-02\n",
        "   -2.46644337e-02   1.70087279e-03]\n",
        " [ -4.66752192e-03   2.87849307e-02   1.72156021e-01 ...,   1.45561306e-03\n",
        "    2.42715818e-03   1.21286856e-02]\n",
        " ..., \n",
        " [ -5.88082150e-03   3.92963067e-02  -1.02771446e-03 ...,   1.89997867e-01\n",
        "   -9.23397858e-03   8.56697652e-03]\n",
        " [  8.21293332e-03  -3.07865757e-02   1.08245527e-02 ...,  -4.96891467e-03\n",
        "    1.92571193e-01   2.13590879e-02]\n",
        " [  8.59491248e-03  -2.48973083e-04   1.06977308e-02 ...,   1.16584627e-02\n",
        "    4.31708433e-03   1.91000864e-01]]\n"
       ]
      },
      {
       "metadata": {},
       "output_type": "pyout",
       "prompt_number": 26,
       "text": [
        "<matplotlib.image.AxesImage at 0x11bc978d0>"
       ]
      },
      {
       "metadata": {},
       "output_type": "display_data",
       "png": "iVBORw0KGgoAAAANSUhEUgAAAWwAAAFtCAYAAAAjw1AcAAAABHNCSVQICAgIfAhkiAAAAAlwSFlz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\n8P8BFdNs4xlZETwAAAAASUVORK5CYII=\n",
       "text": [
        "<matplotlib.figure.Figure at 0x11bc25a10>"
       ]
      }
     ],
     "prompt_number": 26
    },
    {
     "cell_type": "code",
     "collapsed": false,
     "input": [],
     "language": "python",
     "metadata": {},
     "outputs": []
    }
   ],
   "metadata": {}
  }
 ]
}