chapter_6_traffic_lights.ipynb

{
 "cells": [
  {
   "cell_type": "code",
   "execution_count": 1,
   "metadata": {},
   "outputs": [],
   "source": [
    "import numpy as np"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 2,
   "metadata": {},
   "outputs": [],
   "source": [
    "streetlights = np.array( [ [ 1, 0, 1 ],\n",
    "            [ 0, 1, 1 ],\n",
    "            [ 0, 0, 1 ],\n",
    "            [ 1, 1, 1 ],\n",
    "            [ 0, 1, 1 ],\n",
    "            [ 1, 0, 1 ] ] )\n",
    "\n",
    "walk_vs_stop = np.array( [ [ 0 ],\n",
    "            [ 1 ],\n",
    "            [ 0 ],\n",
    "            [ 1 ],\n",
    "            [ 1 ],\n",
    "            [ 0 ] ] )\n",
    "weights = np.array([0.5,0.48,-0.7])\n",
    "alpha = 0.1"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "# Building The Network"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 3,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Pred: -0.19999999999999996\n",
      "Err: [0.04] \n",
      "\n",
      "Pred: -0.19999999999999996\n",
      "Err: [1.48] \n",
      "\n",
      "Pred: -0.5599999999999999\n",
      "Err: [1.7936] \n",
      "\n",
      "Pred: 0.6160000000000001\n",
      "Err: [1.941056] \n",
      "\n",
      "Pred: 0.17279999999999995\n",
      "Err: [2.62531584] \n",
      "\n",
      "Pred: 0.17552\n",
      "Err: [2.65612311] \n",
      "\n",
      "Pred: 0.14041599999999999\n",
      "Err: [0.01971665] \n",
      "\n",
      "Pred: 0.3066464\n",
      "Err: [0.50045587] \n",
      "\n",
      "Pred: -0.34513824\n",
      "Err: [0.61957627] \n",
      "\n",
      "Pred: 1.006637344\n",
      "Err: [0.61962033] \n",
      "\n",
      "Pred: 0.4785034751999999\n",
      "Err: [0.89157895] \n",
      "\n",
      "Pred: 0.26700416768\n",
      "Err: [0.96287018] \n",
      "\n",
      "Pred: 0.213603334144\n",
      "Err: [0.04562638] \n",
      "\n",
      "Pred: 0.5347420299776\n",
      "Err: [0.26209136] \n",
      "\n",
      "Pred: -0.26067345110016\n",
      "Err: [0.33004201] \n",
      "\n",
      "Pred: 1.1319428845096962\n",
      "Err: [0.34745094] \n",
      "\n",
      "Pred: 0.6274723921901568\n",
      "Err: [0.48622775] \n",
      "\n",
      "Pred: 0.25433999330650114\n",
      "Err: [0.55091659] \n",
      "\n",
      "Pred: 0.20347199464520088\n",
      "Err: [0.04140085] \n",
      "\n",
      "Pred: 0.6561967149569552\n",
      "Err: [0.15960155] \n",
      "\n",
      "Pred: -0.221948503950995\n",
      "Err: [0.20886269] \n",
      "\n",
      "Pred: 1.166258650532124\n",
      "Err: [0.23650463] \n",
      "\n",
      "Pred: 0.7139004922542389\n",
      "Err: [0.31835756] \n",
      "\n",
      "Pred: 0.21471099528371604\n",
      "Err: [0.36445837] \n",
      "\n",
      "Pred: 0.17176879622697283\n",
      "Err: [0.02950452] \n",
      "\n",
      "Pred: 0.7324724146523222\n",
      "Err: [0.10107553] \n",
      "\n",
      "Pred: -0.19966478845083285\n",
      "Err: [0.14094156] \n",
      "\n",
      "Pred: 1.1697769945341199\n",
      "Err: [0.16976578] \n",
      "\n",
      "Pred: 0.7719890116601171\n",
      "Err: [0.22175479] \n",
      "\n",
      "Pred: 0.17297997428859369\n",
      "Err: [0.25167687] \n",
      "\n",
      "Pred: 0.13838397943087496\n",
      "Err: [0.01915013] \n",
      "\n",
      "Pred: 0.7864548139561468\n",
      "Err: [0.06475167] \n",
      "\n",
      "Pred: -0.1836567869927348\n",
      "Err: [0.09848149] \n",
      "\n",
      "Pred: 1.163248019006011\n",
      "Err: [0.1251314] \n",
      "\n",
      "Pred: 0.8148799260629888\n",
      "Err: [0.15940085] \n",
      "\n",
      "Pred: 0.1362897844408577\n",
      "Err: [0.17797575] \n",
      "\n",
      "Pred: 0.10903182755268614\n",
      "Err: [0.01188794] \n",
      "\n",
      "Pred: 0.8273717796510367\n",
      "Err: [0.04168844] \n",
      "\n",
      "Pred: -0.17037324196481937\n",
      "Err: [0.07071548] \n",
      "\n",
      "Pred: 1.1537962739591756\n",
      "Err: [0.09436878] \n",
      "\n",
      "Pred: 0.8481754931254761\n",
      "Err: [0.11741946] \n",
      "\n",
      "Pred: 0.1059488041691444\n",
      "Err: [0.12864461] \n",
      "\n",
      "Pred: 0.0847590433353155\n",
      "Err: [0.0071841] \n",
      "\n",
      "Pred: 0.859469609749935\n",
      "Err: [0.02693289] \n",
      "\n",
      "Pred: -0.1585508402022421\n",
      "Err: [0.05207125] \n",
      "\n",
      "Pred: 1.1438418857156731\n",
      "Err: [0.07276174] \n",
      "\n",
      "Pred: 0.8746623946770374\n",
      "Err: [0.08847126] \n",
      "\n",
      "Pred: 0.08148074110264475\n",
      "Err: [0.09511037] \n",
      "\n",
      "Pred: 0.06518459288211581\n",
      "Err: [0.00424903] \n",
      "\n",
      "Pred: 0.8850633823431538\n",
      "Err: [0.01745946] \n",
      "\n",
      "Pred: -0.14771905585408038\n",
      "Err: [0.03928038] \n",
      "\n",
      "Pred: 1.1341830033853888\n",
      "Err: [0.05728546] \n",
      "\n",
      "Pred: 0.8959860107828534\n",
      "Err: [0.06810437] \n",
      "\n",
      "Pred: 0.0619780399014222\n",
      "Err: [0.07194564] \n",
      "\n",
      "Pred: 0.04958243192113776\n",
      "Err: [0.00245842] \n",
      "\n",
      "Pred: 0.9056327614440267\n",
      "Err: [0.01136359] \n",
      "\n",
      "Pred: -0.13768337501215525\n",
      "Err: [0.03032031] \n",
      "\n",
      "Pred: 1.1250605910610996\n",
      "Err: [0.04596046] \n",
      "\n",
      "Pred: 0.9132624284442169\n",
      "Err: [0.05348386] \n",
      "\n",
      "Pred: 0.04653264583708144\n",
      "Err: [0.05564915] \n",
      "\n",
      "Pred: 0.03722611666966513\n",
      "Err: [0.00138578] \n",
      "\n",
      "Pred: 0.922234066504699\n",
      "Err: [0.00743332] \n",
      "\n",
      "Pred: -0.12834662236261596\n",
      "Err: [0.02390618] \n",
      "\n",
      "Pred: 1.116526024487899\n",
      "Err: [0.03748449] \n",
      "\n",
      "Pred: 0.9273167105424409\n",
      "Err: [0.04276735] \n",
      "\n",
      "Pred: 0.03435527296969987\n",
      "Err: [0.04394764] \n",
      "\n",
      "Pred: 0.027484218375759886\n",
      "Err: [0.00075538] \n",
      "\n",
      "Pred: 0.9356694192994068\n",
      "Err: [0.00489381] \n",
      "\n",
      "Pred: -0.11964712469387503\n",
      "Err: [0.01920924] \n",
      "\n",
      "Pred: 1.1085678053734553\n",
      "Err: [0.03099621] \n",
      "\n",
      "Pred: 0.9387866868342218\n",
      "Err: [0.03474328] \n",
      "\n",
      "Pred: 0.024792915481941458\n",
      "Err: [0.03535797] \n",
      "\n",
      "Pred: 0.019834332385553155\n",
      "Err: [0.0003934] \n",
      "\n",
      "Pred: 0.946566624680628\n",
      "Err: [0.00324853] \n",
      "\n",
      "Pred: -0.11153724870006754\n",
      "Err: [0.01568908] \n",
      "\n",
      "Pred: 1.1011550767549563\n",
      "Err: [0.02592143] \n",
      "\n",
      "Pred: 0.948176009263518\n",
      "Err: [0.02860716] \n",
      "\n",
      "Pred: 0.017315912033043404\n",
      "Err: [0.028907] \n",
      "\n",
      "Pred: 0.013852729626434732\n",
      "Err: [0.0001919] \n",
      "\n",
      "Pred: 0.9554239432448665\n",
      "Err: [0.00217892] \n",
      "\n",
      "Pred: -0.10397589092234266\n",
      "Err: [0.01298991] \n",
      "\n",
      "Pred: 1.0942524239871314\n",
      "Err: [0.02187343] \n",
      "\n",
      "Pred: 0.9558862588907013\n",
      "Err: [0.02381945] \n",
      "\n",
      "Pred: 0.011498267782398985\n",
      "Err: [0.02395166] \n",
      "\n",
      "Pred: 0.009198614225919194\n",
      "Err: [8.46145037e-05] \n",
      "\n",
      "Pred: 0.9626393189117293\n",
      "Err: [0.00148043] \n",
      "\n",
      "Pred: -0.09692579020989642\n",
      "Err: [0.01087504] \n",
      "\n",
      "Pred: 1.087824783849832\n",
      "Err: [0.01858824] \n",
      "\n",
      "Pred: 0.9622390773804066\n",
      "Err: [0.02001412] \n",
      "\n",
      "Pred: 0.006998674002545002\n",
      "Err: [0.02006311] \n",
      "\n",
      "Pred: 0.005598939202035996\n",
      "Err: [3.13481202e-05] \n",
      "\n",
      "Pred: 0.9685315005838672\n",
      "Err: [0.00102161] \n",
      "\n",
      "Pred: -0.09035250869077546\n",
      "Err: [0.00918519] \n",
      "\n",
      "Pred: 1.0818389613301889\n",
      "Err: [0.01588281] \n",
      "\n",
      "Pred: 0.9674926590701334\n",
      "Err: [0.01693953] \n",
      "\n",
      "Pred: 0.003544193999268516\n",
      "Err: [0.01695209] \n",
      "\n",
      "Pred: 0.0028353551994148157\n",
      "Err: [8.03923911e-06] \n",
      "\n",
      "Pred: 0.9733561723362383\n",
      "Err: [0.00071793] \n",
      "\n",
      "Pred: -0.0842239920152223\n",
      "Err: [0.00781161] \n",
      "\n",
      "Pred: 1.0762639960116431\n",
      "Err: [0.01362781] \n",
      "\n",
      "Pred: 0.9718545378681842\n",
      "Err: [0.01441998] \n",
      "\n",
      "Pred: 0.0009168131382832068\n",
      "Err: [0.01442082] \n",
      "\n",
      "Pred: 0.0007334505106265654\n",
      "Err: [5.37949652e-07] \n",
      "\n",
      "Pred: 0.9773186039296565\n",
      "Err: [0.00051498] \n",
      "\n",
      "Pred: -0.07851033295953944\n",
      "Err: [0.00667886] \n",
      "\n",
      "Pred: 1.0710711494147542\n",
      "Err: [0.01172996] \n",
      "\n",
      "Pred: 0.9754916865567282\n",
      "Err: [0.01233062] \n",
      "\n",
      "Pred: -0.0010574652271341245\n",
      "Err: [0.01233174] \n",
      "\n",
      "Pred: -0.0008459721817072885\n",
      "Err: [7.15668932e-07] \n",
      "\n",
      "Pred: 0.9805836929862668\n",
      "Err: [0.00037771] \n",
      "\n",
      "Pred: -0.07318360881847627\n",
      "Err: [0.00573355] \n",
      "\n",
      "Pred: 1.066233777045345\n",
      "Err: [0.01012046] \n",
      "\n",
      "Pred: 0.9785385598617921\n",
      "Err: [0.01058106] \n",
      "\n",
      "Pred: -0.0025173975573930946\n",
      "Err: [0.01058739] \n",
      "\n",
      "Pred: -0.002013918045914484\n",
      "Err: [4.0558659e-06] \n",
      "\n",
      "Pred: 0.9832839794497644\n",
      "Err: [0.00028348] \n",
      "\n",
      "Pred: -0.06821774801198803\n",
      "Err: [0.00493714] \n",
      "\n",
      "Pred: 1.0617271739912904\n",
      "Err: [0.00874739] \n",
      "\n",
      "Pred: 0.9811035235627523\n",
      "Err: [0.00910446] \n",
      "\n",
      "Pred: -0.0035735447350425317\n",
      "Err: [0.00911723] \n",
      "\n",
      "Pred: -0.002858835788034024\n",
      "Err: [8.17294206e-06] \n",
      "\n",
      "Pred: 0.9855260569025094\n",
      "Err: [0.00021767] \n",
      "\n",
      "Pred: -0.06358841060413677\n",
      "Err: [0.00426115] \n",
      "\n",
      "Pred: 1.05752842286588\n",
      "Err: [0.00757067] \n",
      "\n",
      "Pred: 0.9832740020092452\n",
      "Err: [0.00785043] \n",
      "\n",
      "Pred: -0.004313918034364962\n",
      "Err: [0.00786904] \n",
      "\n",
      "Pred: -0.003451134427491974\n",
      "Err: [1.19103288e-05] \n",
      "\n",
      "Pred: 0.9873957068535818\n",
      "Err: [0.00017078] \n",
      "\n",
      "Pred: -0.059272877470408075\n",
      "Err: [0.00368405] \n",
      "\n",
      "Pred: 1.0536162524729626\n",
      "Err: [0.00655876] \n",
      "\n",
      "Pred: 0.9851206027353137\n",
      "Err: [0.00678015] \n",
      "\n",
      "Pred: -0.004808501248434842\n",
      "Err: [0.00680327] \n",
      "\n",
      "Pred: -0.0038468009987478735\n",
      "Err: [1.47978779e-05] \n",
      "\n",
      "Pred: 0.9889620124129692\n",
      "Err: [0.00013664] \n",
      "\n",
      "Pred: -0.05524994626077355\n",
      "Err: [0.00318919] \n",
      "\n",
      "Pred: 1.049970908776931\n",
      "Err: [0.00568628] \n",
      "\n",
      "Pred: 0.9867004228010665\n",
      "Err: [0.00586316] \n",
      "\n",
      "Pred: -0.005112871449710697\n",
      "Err: [0.0058893] \n",
      "\n",
      "Pred: -0.004090297159768559\n",
      "Err: [1.67305309e-05] \n",
      "\n",
      "Pred: 0.9902806551018011\n",
      "Err: [0.0001112] \n",
      "\n",
      "Pred: -0.051499833441728114\n",
      "Err: [0.00276343] \n",
      "\n",
      "Pred: 1.0465740376293469\n",
      "Err: [0.00493257] \n",
      "\n",
      "Pred: 0.9880596998997442\n",
      "Err: [0.00507514] \n",
      "\n",
      "Pred: -0.0052710974096659285\n",
      "Err: [0.00510293] \n",
      "\n",
      "Pred: -0.004216877927732746\n",
      "Err: [1.77820595e-05] \n",
      "\n",
      "Pred: 0.9913965574535352\n",
      "Err: [9.18012831e-05] \n",
      "\n",
      "Pred: -0.048004082062078055\n",
      "Err: [0.00239619] \n",
      "\n",
      "Pred: 1.043408578143574\n",
      "Err: [0.0042805] \n",
      "\n",
      "Pred: 0.9892359385403211\n",
      "Err: [0.00439636] \n",
      "\n",
      "Pred: -0.005318059364078823\n",
      "Err: [0.00442464] \n",
      "\n",
      "Pred: -0.0042544474912630525\n",
      "Err: [1.81003235e-05] \n",
      "\n",
      "Pred: 0.992346001517791\n",
      "Err: [7.66840162e-05] \n",
      "\n",
      "Pred: -0.044745474990504665\n",
      "Err: [0.00207884] \n",
      "\n",
      "Pred: 1.0404586655589985\n",
      "Err: [0.00371575] \n",
      "\n",
      "Pred: 0.9902596156014837\n",
      "Err: [0.00381062] \n",
      "\n",
      "Pred: -0.005281305317687134\n",
      "Err: [0.00383851] \n",
      "\n",
      "Pred: -0.0042250442541497055\n",
      "Err: [1.78509989e-05] \n",
      "\n",
      "Pred: 0.9931583274383705\n",
      "Err: [6.46594824e-05] \n",
      "\n",
      "Pred: -0.041707953394155776\n",
      "Err: [0.00180421] \n",
      "\n",
      "Pred: 1.0377095425371112\n",
      "Err: [0.00322622] \n",
      "\n",
      "Pred: 0.9911555487826897\n",
      "Err: [0.00330445] \n",
      "\n",
      "Pred: -0.005182536193432452\n",
      "Err: [0.00333131] \n",
      "\n",
      "Pred: -0.004146028954745959\n",
      "Err: [1.71895561e-05] \n",
      "\n",
      "Pred: 0.9938572955409696\n",
      "Err: [5.49223742e-05] \n",
      "\n",
      "Pred: -0.03887654022599941\n",
      "Err: [0.00156631] \n",
      "\n",
      "Pred: 1.0351474779634813\n",
      "Err: [0.00280165] \n",
      "\n",
      "Pred: 0.9919439948626794\n",
      "Err: [0.00286655] \n",
      "\n",
      "Pred: -0.00503879377425797\n",
      "Err: [0.00289194] \n",
      "\n",
      "Pred: -0.004031035019406375\n",
      "Err: [1.62492433e-05] \n",
      "\n",
      "Pred: 0.9944621787695098\n",
      "Err: [4.69167073e-05] \n",
      "\n",
      "Pred: -0.03623726848360008\n",
      "Err: [0.00136006] \n",
      "\n",
      "Pred: 1.032759692455092\n",
      "Err: [0.00243325] \n",
      "\n",
      "Pred: 0.9926415313729495\n",
      "Err: [0.0024874] \n",
      "\n",
      "Pred: -0.004863410672429416\n",
      "Err: [0.00251105] \n",
      "\n",
      "Pred: -0.003890728537943533\n",
      "Err: [1.51377686e-05] \n",
      "\n",
      "Pred: 0.9949886390193969\n",
      "Err: [4.02515074e-05] \n",
      "\n",
      "Pred: -0.03377711399894662\n",
      "Err: [0.00118114] \n",
      "\n",
      "Pred: 1.0305342898820642\n",
      "Err: [0.00211349] \n",
      "\n",
      "Pred: 0.9932617646389992\n",
      "Err: [0.00215889] \n",
      "\n",
      "Pred: -0.004666769772712614\n",
      "Err: [0.00218067] \n",
      "\n",
      "Pred: -0.003733415818170091\n",
      "Err: [1.39383937e-05] \n",
      "\n",
      "Pred: 0.9954494302702878\n",
      "Err: [3.46460785e-05] \n",
      "\n",
      "Pred: -0.03148393251909879\n",
      "Err: [0.00102588] \n",
      "\n",
      "Pred: 1.0284601943056741\n",
      "Err: [0.00183587] \n",
      "\n",
      "Pred: 0.9938158986070053\n",
      "Err: [0.00187411] \n",
      "\n",
      "Pred: -0.004456911151490314\n",
      "Err: [0.00189397] \n",
      "\n",
      "Pred: -0.003565528921192253\n",
      "Err: [1.27129965e-05] \n",
      "\n",
      "Pred: 0.9958549628928723\n",
      "Err: [2.98943291e-05] \n",
      "\n",
      "Pred: -0.029346400840475826\n",
      "Err: [0.00089111] \n",
      "\n",
      "Pred: 1.0265270918125804\n",
      "Err: [0.00159479] \n",
      "\n",
      "Pred: 0.9943131920358295\n",
      "Err: [0.00162713] \n",
      "\n",
      "Pred: -0.004240016908292479\n",
      "Err: [0.00164511] \n",
      "\n",
      "Pred: -0.0033920135266339822\n",
      "Err: [1.15057558e-05] \n",
      "\n",
      "Pred: 0.9962137566721563\n",
      "Err: [2.58413943e-05] \n",
      "\n",
      "Pred: -0.02735396176499221\n"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Err: [0.00077408] \n",
      "\n",
      "Pred: 1.0247253767906936\n",
      "Err: [0.00138542] \n",
      "\n",
      "Pred: 0.9947613261560856\n",
      "Err: [0.00141287] \n",
      "\n",
      "Pred: -0.004020798285770878\n",
      "Err: [0.00142904] \n",
      "\n",
      "Pred: -0.003216638628616701\n",
      "Err: [1.03467641e-05] \n",
      "\n",
      "Pred: 0.9965328046163073\n",
      "Err: [2.23682079e-05] \n",
      "\n",
      "Pred: -0.025496772653362886\n",
      "Err: [0.00067245] \n",
      "\n",
      "Pred: 1.0230461022472208\n",
      "Err: [0.00120358] \n",
      "\n",
      "Pred: 0.9951667005089379\n",
      "Err: [0.00122694] \n",
      "\n",
      "Pred: -0.0038028045995257484\n",
      "Err: [0.0012414] \n",
      "\n",
      "Pred: -0.003042243679620596\n",
      "Err: [9.25524661e-06] \n",
      "\n",
      "Pred: 0.996817865235065\n",
      "Err: [1.93812283e-05] \n",
      "\n",
      "Pred: -0.023765657359234325\n",
      "Err: [0.00058419] \n",
      "\n",
      "Pred: 1.0214809338160067\n",
      "Err: [0.00104562] \n",
      "\n",
      "Pred: 0.995534671160774\n",
      "Err: [0.00106556] \n",
      "\n",
      "Pred: -0.0035886696105582767\n",
      "Err: [0.00107844] \n",
      "\n",
      "Pred: -0.0028709356884466207\n",
      "Err: [8.24227173e-06] \n",
      "\n",
      "Pred: 0.9970736974585198\n",
      "Err: [1.68055183e-05] \n",
      "\n",
      "Pred: -0.022152061336940452\n",
      "Err: [0.00050752] \n",
      "\n",
      "Pred: 1.0200221071408409\n",
      "Err: [0.0009084] \n",
      "\n",
      "Pred: 0.9958697426723416\n",
      "Err: [0.00092546] \n",
      "\n",
      "Pred: -0.0033803078583175654\n",
      "Err: [0.00093689] \n",
      "\n",
      "Pred: -0.0027042462866540516\n",
      "Err: [7.31294798e-06] \n",
      "\n",
      "Pred: 0.9973042495523706\n",
      "Err: [1.45800185e-05] \n",
      "\n",
      "Pred: -0.02064800972530455\n",
      "Err: [0.00044092] \n",
      "\n",
      "Pred: 1.018662388355171\n",
      "Err: [0.00078921] \n",
      "\n",
      "Pred: 0.9961757229433927\n",
      "Err: [0.00080383] \n",
      "\n",
      "Pred: -0.0031790709774033414\n",
      "Err: [0.00081394] \n",
      "\n",
      "Pred: -0.002543256781922673\n",
      "Err: [6.46815506e-06] \n",
      "\n",
      "Pred: 0.9975128111306469\n",
      "Err: [1.26542635e-05] \n",
      "\n",
      "Pred: -0.019246068219762574\n",
      "Err: [0.00038307] \n",
      "\n",
      "Pred: 1.0173950374076535\n",
      "Err: [0.00068565] \n",
      "\n",
      "Pred: 0.9964558482449631\n",
      "Err: [0.00069821] \n",
      "\n",
      "Pred: -0.0029858720226535913\n",
      "Err: [0.00070713] \n",
      "\n",
      "Pred: -0.002388697618122871\n",
      "Err: [5.70587631e-06] \n",
      "\n",
      "Pred: 0.9977021355600483\n",
      "Err: [1.09860573e-05] \n",
      "\n",
      "Pred: -0.01793930655497516\n",
      "Err: [0.0003328] \n",
      "\n",
      "Pred: 1.0162137740080082\n",
      "Err: [0.00059569] \n",
      "\n",
      "Pred: 0.9967128843019345\n",
      "Err: [0.0006065] \n",
      "\n",
      "Pred: -0.0028012842268006904\n",
      "Err: [0.00061434] \n",
      "\n",
      "Pred: -0.0022410273814405524\n",
      "Err: [5.02220372e-06] \n",
      "\n",
      "Pred: 0.9978745386023716\n",
      "Err: [9.53978988e-06] \n",
      "\n",
      "Pred: -0.016721264429884947\n",
      "Err: [0.00028914] \n",
      "\n",
      "Pred: 1.0151127459893812\n",
      "Err: [0.00051754] \n",
      "\n",
      "Pred: 0.9969492081270097\n",
      "Err: [0.00052684] \n",
      "\n",
      "Pred: -0.0026256193329783125\n",
      "Err: [0.00053374] \n",
      "\n"
     ]
    }
   ],
   "source": [
    "input = streetlights[0]\n",
    "goal_prediction = walk_vs_stop[0]\n",
    "for iteration in range(40):\n",
    "    error_for_all_lights = 0\n",
    "    for row_index in range(len(walk_vs_stop)):\n",
    "        input = streetlights[row_index]\n",
    "        goal_prediction = walk_vs_stop[row_index]\n",
    "        prediction = input.dot(weights)\n",
    "        error= (goal_prediction - prediction) **2\n",
    "        error_for_all_lights += error\n",
    "        delta = prediction - goal_prediction\n",
    "        weights = weights - (alpha * (input * delta))\n",
    "        print(\"Pred:\",prediction)\n",
    "        print(\"Err:\",error_for_all_lights,\"\\n\")"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "# adding backpropagation and using relu to set all the negative numbers to zero that ruins the network from layer 1"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 4,
   "metadata": {},
   "outputs": [],
   "source": [
    "np.random.seed(1)\n",
    "\n",
    "def relu(x):\n",
    "    return (x>0) *x\n",
    "alpha = 0.2\n",
    "hidden_size =4\n",
    "\n",
    "streetlights = np.array( [[ 1, 0, 1 ],\n",
    "                          [ 0, 1, 1 ],\n",
    "                          [ 0, 0, 1 ],\n",
    "                          [ 1, 1, 1 ] ] )\n",
    "walk_vs_stop = np.array([[ 1, 1, 0, 0]]).T"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 5,
   "metadata": {},
   "outputs": [],
   "source": [
    "weights_0_1 = 2* np.random.random((3,hidden_size)) -1\n",
    "weights_1_2 = 2* np.random.random((hidden_size,1)) -1\n",
    "\n",
    "layer_0 = streetlights[0]\n",
    "layer_1 = relu(np.dot(layer_0,weights_0_1))\n",
    "layer_2 = np.dot(layer_1,weights_1_2)"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "# backprobagation in code"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 6,
   "metadata": {},
   "outputs": [],
   "source": [
    "def relu2deriv(output):\n",
    "    return output>0\n",
    "\n",
    "alpha = 0.2\n",
    "hidden_size = 4"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 7,
   "metadata": {},
   "outputs": [],
   "source": [
    "weights_0_1 = 2 * np.random.random((3,hidden_size)) -1\n",
    "weight_1_2 = 2* np.random.random((hidden_size,1)) - 1"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 8,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "ERROR: 1.5917785930433102\n",
      "ERROR: 1.459807623530446\n",
      "ERROR: 1.406615166969481\n",
      "ERROR: 1.370936154089727\n",
      "ERROR: 1.3171814077744681\n",
      "ERROR: 1.1676938947594402\n"
     ]
    }
   ],
   "source": [
    "for iteration in range(60):\n",
    "    layer_2_error = 0\n",
    "    for i in range(len(streetlights)):\n",
    "        layer_0 = streetlights[i: i+1]\n",
    "        layer_1 = relu(np.dot(layer_0, weights_0_1))\n",
    "        layer_2 = np.dot(layer_1, weight_1_2)\n",
    "        \n",
    "        layer_2_error += np.sum((layer_2 - walk_vs_stop[i:i+1]) ** 2)\n",
    "        layer_2_delta = (walk_vs_stop[i:i+1] - layer_2)\n",
    "        layer_1_delta = layer_2_delta.dot(weight_1_2.T)*relu2deriv(layer_1)\n",
    "        \n",
    "        weight_1_2 += alpha * layer_1.T.dot(layer_2_delta)\n",
    "        weights_0_1 += alpha * layer_0.T.dot(layer_1_delta)\n",
    "    if(iteration % 10 == 9):\n",
    "        print(\"ERROR:\",layer_2_error)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 9,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "array([[0.52316203]])"
      ]
     },
     "execution_count": 9,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": []
  }
 ],
 "metadata": {
  "kernelspec": {
   "display_name": "Python 3",
   "language": "python",
   "name": "python3"
  },
  "language_info": {
   "codemirror_mode": {
    "name": "ipython",
    "version": 3
   },
   "file_extension": ".py",
   "mimetype": "text/x-python",
   "name": "python",
   "nbconvert_exporter": "python",
   "pygments_lexer": "ipython3",
   "version": "3.6.7"
  }
 },
 "nbformat": 4,
 "nbformat_minor": 2
}