DQN_cartpole.ipynb

{
 "cells": [
  {
   "cell_type": "code",
   "execution_count": 21,
   "metadata": {},
   "outputs": [],
   "source": [
    "'''\n",
    "Code adapted from\n",
    "https://github.com/berkeleydeeprlcourse/homework/blob/master/hw3/dqn.py\n",
    "'''\n",
    "import tensorflow as tf\n",
    "from keras.models import Sequential\n",
    "from keras.layers import Dense\n",
    "from keras.optimizers import Adam\n",
    "import matplotlib.pyplot as plt\n",
    "import numpy as np\n",
    "import random\n",
    "import gym\n",
    "import math\n",
    "import matplotlib.pyplot as plt\n",
    "from gym.wrappers.monitor import Monitor\n",
    "import time\n",
    "import sys\n",
    "import random\n",
    "from collections import deque"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 22,
   "metadata": {},
   "outputs": [],
   "source": [
    "class DQNAgent:\n",
    "    def __init__(self, state_size, action_size):\n",
    "        self.state_size = state_size\n",
    "        self.action_size = action_size\n",
    "        \n",
    "        self.gamma = 0.99\n",
    "        self.epsilon = 1.0\n",
    "        self.epsilon_min = 0.01\n",
    "        self.epsilon_decay = 0.999\n",
    "        self.learning_rate = 0.001\n",
    "        self.batch_size = 64\n",
    "        self.train_start = 1000\n",
    "        self.render = False\n",
    "        \n",
    "        self.memory = deque([], 2000)\n",
    "        self.score_history = []\n",
    "        self.model = self.DQNNet()\n",
    "        self.target_model = self.DQNNet()\n",
    "        \n",
    "        self.updateTarget()\n",
    "        \n",
    "    def DQNNet(self):\n",
    "        model = Sequential()\n",
    "        \n",
    "        model.add(Dense(24, input_dim=self.state_size, activation='relu'))\n",
    "        model.add(Dense(24, activation='relu'))\n",
    "        model.add(Dense(self.action_size, activation='linear'))\n",
    "        model.compile(loss='mse', optimizer=Adam(lr=self.learning_rate))\n",
    "        \n",
    "        return model\n",
    "    \n",
    "    def updateTarget(self):\n",
    "        self.target_model.set_weights(self.model.get_weights())\n",
    "    \n",
    "    def remember(self, state, action, reward, next_state, done):\n",
    "        self.memory.append((state, action, reward, next_state, done))\n",
    "        if self.epsilon > self.epsilon_min:\n",
    "            self.epsilon *= self.epsilon_decay\n",
    "        \n",
    "    def getAction(self, state):\n",
    "        r = np.random.uniform(low=0, high=1)\n",
    "\n",
    "        if r <= self.epsilon:\n",
    "            action = random.randrange(self.action_size)\n",
    "        else:\n",
    "            actionValues = self.model.predict(state)\n",
    "            action = np.argmax(actionValues[0])\n",
    "        return action\n",
    "                \n",
    "    def train(self):\n",
    "        if len(self.memory) < self.train_start:\n",
    "            return\n",
    "        \n",
    "        batch_size = min(self.batch_size, len(self.memory))\n",
    "        mb = random.sample(self.memory, batch_size)\n",
    "        \n",
    "        new = np.zeros((batch_size, self.state_size))\n",
    "        newTarget = np.zeros((batch_size, self.state_size))\n",
    "        action = []\n",
    "        reward = []\n",
    "        done = []\n",
    "        \n",
    "        for i in range(self.batch_size):\n",
    "            new[i] = mb[i][0]\n",
    "            action.append(mb[i][1])\n",
    "            reward.append(mb[i][2])\n",
    "            newTarget[i] = mb[i][3]\n",
    "            done.append(mb[i][4])\n",
    "            \n",
    "        target = self.model.predict(new)\n",
    "        target_val = self.target_model.predict(newTarget)\n",
    "        \n",
    "        for i in range(self.batch_size):\n",
    "            if done[i]:\n",
    "                target[i][action[i]] = reward[i]\n",
    "            else:\n",
    "                target[i][action[i]] = reward[i] + self.gamma * (np.amax(target_val[i]))\n",
    "        \n",
    "        self.model.fit(new, target, batch_size=self.batch_size, epochs=1, verbose=0)\n"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 23,
   "metadata": {},
   "outputs": [],
   "source": [
    "env = gym.make('CartPole-v1')\n",
    "state_size = env.observation_space.shape[0]\n",
    "action_size = env.action_space.n\n",
    "agent = DQNAgent(state_size, action_size)\n",
    "numEpisodes = 200"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 24,
   "metadata": {},
   "outputs": [],
   "source": [
    "scores = []\n",
    "episodes = []"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 25,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "0 23.0\n",
      "1 33.0\n",
      "2 9.0\n",
      "3 28.0\n",
      "4 59.0\n",
      "5 25.0\n",
      "6 16.0\n",
      "7 26.0\n",
      "8 10.0\n",
      "9 38.0\n",
      "10 24.0\n",
      "11 23.0\n",
      "12 17.0\n",
      "13 44.0\n",
      "14 41.0\n",
      "15 37.0\n",
      "16 62.0\n",
      "17 67.0\n",
      "18 14.0\n",
      "19 44.0\n",
      "20 26.0\n",
      "21 67.0\n",
      "22 93.0\n",
      "23 64.0\n",
      "24 34.0\n",
      "25 44.0\n",
      "26 37.0\n",
      "27 12.0\n",
      "28 12.0\n",
      "29 11.0\n",
      "30 11.0\n",
      "31 9.0\n",
      "32 9.0\n",
      "33 13.0\n",
      "34 30.0\n",
      "35 31.0\n",
      "36 23.0\n",
      "37 79.0\n",
      "38 21.0\n",
      "39 52.0\n",
      "40 40.0\n",
      "41 94.0\n",
      "42 27.0\n",
      "43 18.0\n",
      "44 22.0\n",
      "45 18.0\n",
      "46 17.0\n",
      "47 19.0\n",
      "48 19.0\n",
      "49 14.0\n",
      "50 17.0\n",
      "51 30.0\n",
      "52 19.0\n",
      "53 45.0\n",
      "54 120.0\n",
      "55 79.0\n",
      "56 83.0\n",
      "57 85.0\n",
      "58 219.0\n",
      "59 115.0\n",
      "60 134.0\n",
      "61 137.0\n",
      "62 148.0\n",
      "63 210.0\n",
      "64 139.0\n",
      "65 135.0\n",
      "66 156.0\n",
      "67 334.0\n",
      "68 267.0\n",
      "69 241.0\n",
      "70 245.0\n",
      "71 275.0\n",
      "72 237.0\n",
      "73 191.0\n",
      "74 420.0\n",
      "75 305.0\n",
      "76 306.0\n",
      "77 230.0\n",
      "78 366.0\n",
      "79 239.0\n",
      "80 263.0\n",
      "81 178.0\n",
      "82 277.0\n",
      "83 332.0\n",
      "84 197.0\n",
      "85 300.0\n",
      "86 410.0\n",
      "87 500.0\n",
      "88 196.0\n",
      "89 402.0\n",
      "90 331.0\n",
      "91 167.0\n",
      "92 126.0\n",
      "93 352.0\n",
      "94 500.0\n",
      "95 212.0\n",
      "96 149.0\n",
      "97 467.0\n",
      "98 271.0\n",
      "99 325.0\n",
      "100 336.0\n",
      "101 415.0\n",
      "102 273.0\n",
      "103 147.0\n",
      "104 478.0\n",
      "105 203.0\n",
      "106 466.0\n",
      "107 192.0\n",
      "108 195.0\n",
      "109 342.0\n",
      "110 405.0\n",
      "111 312.0\n",
      "112 349.0\n",
      "113 500.0\n",
      "114 199.0\n",
      "115 343.0\n",
      "116 400.0\n",
      "117 370.0\n",
      "118 209.0\n",
      "119 500.0\n",
      "120 239.0\n",
      "121 401.0\n",
      "122 365.0\n",
      "123 500.0\n",
      "124 210.0\n",
      "125 287.0\n",
      "126 407.0\n",
      "127 276.0\n",
      "128 198.0\n",
      "129 308.0\n",
      "130 300.0\n",
      "131 500.0\n",
      "132 422.0\n",
      "133 329.0\n",
      "134 275.0\n",
      "135 486.0\n",
      "136 500.0\n",
      "137 246.0\n",
      "138 484.0\n",
      "139 450.0\n",
      "140 500.0\n",
      "141 500.0\n",
      "142 500.0\n",
      "143 500.0\n",
      "144 500.0\n",
      "145 435.0\n",
      "146 500.0\n",
      "147 500.0\n",
      "148 500.0\n",
      "149 500.0\n",
      "150 500.0\n",
      "151 342.0\n",
      "152 377.0\n",
      "153 266.0\n",
      "154 233.0\n",
      "155 306.0\n",
      "156 500.0\n",
      "157 500.0\n",
      "158 500.0\n",
      "159 500.0\n",
      "160 500.0\n",
      "161 500.0\n",
      "162 500.0\n",
      "163 500.0\n",
      "164 202.0\n",
      "165 205.0\n",
      "166 154.0\n",
      "167 137.0\n",
      "168 138.0\n",
      "169 500.0\n",
      "170 500.0\n",
      "171 500.0\n",
      "172 500.0\n",
      "173 500.0\n",
      "174 500.0\n",
      "175 500.0\n",
      "176 500.0\n",
      "177 500.0\n",
      "178 500.0\n",
      "179 500.0\n",
      "180 500.0\n",
      "181 500.0\n",
      "182 500.0\n",
      "183 500.0\n",
      "184 500.0\n",
      "185 500.0\n",
      "186 500.0\n",
      "187 211.0\n",
      "188 410.0\n",
      "189 500.0\n",
      "190 426.0\n",
      "191 186.0\n",
      "192 290.0\n",
      "193 398.0\n",
      "194 288.0\n",
      "195 213.0\n",
      "196 219.0\n",
      "197 278.0\n",
      "198 314.0\n",
      "199 265.0\n"
     ]
    }
   ],
   "source": [
    "for e in range(numEpisodes):\n",
    "    done = False\n",
    "    score = 0    \n",
    "    state = env.reset()\n",
    "    state = np.reshape(state, [1,state_size])\n",
    "    \n",
    "    while not done:\n",
    "        if agent.render:\n",
    "            env.render()\n",
    "            \n",
    "        action = agent.getAction(state)\n",
    "        next_state, reward, done, _ = env.step(action)\n",
    "        next_state = np.reshape(next_state, [1, state_size])\n",
    "        reward = reward if not done or score == 499 else -100\n",
    "        \n",
    "        agent.remember(state, action, reward, next_state, done)\n",
    "        agent.train()\n",
    "        score += reward\n",
    "        state = next_state\n",
    "        \n",
    "        if done:\n",
    "            agent.updateTarget()\n",
    "            \n",
    "            score = score if score == 500 else score + 100\n",
    "            scores.append(score)\n",
    "            episodes.append(e)\n",
    "            print(e, score)\n",
    "            \n",
    "#             if np.mean(scores[-min(10, len(scores)):]) > 490:\n",
    "#                 sys.exit()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 26,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "[<matplotlib.lines.Line2D at 0x7fb13406c630>]"
      ]
     },
     "execution_count": 26,
     "metadata": {},
     "output_type": "execute_result"
    },
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<Figure size 432x288 with 1 Axes>"
      ]
     },
     "metadata": {
      "needs_background": "light"
     },
     "output_type": "display_data"
    }
   ],
   "source": [
    "plt.plot(episodes, scores)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {},
   "outputs": [],
   "source": []
  }
 ],
 "metadata": {
  "kernelspec": {
   "display_name": "Python 3",
   "language": "python",
   "name": "python3"
  },
  "language_info": {
   "codemirror_mode": {
    "name": "ipython",
    "version": 3
   },
   "file_extension": ".py",
   "mimetype": "text/x-python",
   "name": "python",
   "nbconvert_exporter": "python",
   "pygments_lexer": "ipython3",
   "version": "3.6.8"
  }
 },
 "nbformat": 4,
 "nbformat_minor": 2
}