From bf4c4dc3e7f496e03767659674c44446044ac94c Mon Sep 17 00:00:00 2001 From: Ryan Enderby Date: Mon, 1 Oct 2018 14:12:58 -0400 Subject: [PATCH 1/2] adding k-bandit implementation (a one state MPD) --- MDP/k-bandit.ipynb | 352 +++++++++++++++++++++++++++++++++++++++++++++ 1 file changed, 352 insertions(+) create mode 100644 MDP/k-bandit.ipynb diff --git a/MDP/k-bandit.ipynb b/MDP/k-bandit.ipynb new file mode 100644 index 000000000..87dc87003 --- /dev/null +++ b/MDP/k-bandit.ipynb @@ -0,0 +1,352 @@ +{ + "cells": [ + { + "cell_type": "markdown", + "metadata": {}, + "source": [ + "### Types of gym_bandits\n", + "`BanditTwoArmedDeterministicFixed-v0`: Simplest case where one bandit always pays, and the other always doesn't\n", + "\n", + "`BanditTwoArmedHighLowFixed-v0`: Stochastic version with a large difference between which bandit pays out of two choices\n", + "\n", + "`BanditTwoArmedHighHighFixed-v0`: Stochastic version with a small difference between which bandit pays where both are good\n", + "\n", + "`BanditTwoArmedLowLowFixed-v0`: Stochastic version with a small difference between which bandit pays where both are bad\n", + "\n", + "`BanditTenArmedRandomFixed-v0`: 10 armed bandit with random probabilities assigned to payouts\n", + "\n", + "`BanditTenArmedRandomRandom-v0`: 10 armed bandit with random probabilities assigned to both payouts and rewards\n", + "\n", + "`BanditTenArmedUniformDistributedReward-v0`: 10 armed bandit with that always pays out with a reward selected from a uniform distribution\n", + "\n", + "`BanditTenArmedGaussian-v0`: 10 armed bandit mentioned on page 30 of Reinforcement Learning: An Introduction (Sutton and Barto)\n", + "\n", + "----------\n", + "\n", + "#### References\n", + "- gym_bandits library from https://github.com/JKCooper2/gym-bandits\n", + "- moving average function https://github.com/klangner/rl-examples/blob/master/notebooks/openai-gym/%5Bnot%20solved%5D%20CartPole.ipynb" + ] + }, + { + "cell_type": "code", + "execution_count": 352, + "metadata": { + "collapsed": true + }, + "outputs": [], + "source": [ + "import gym_bandits\n", + "import gym\n", + "import matplotlib.pyplot as plt\n", + "import numpy as np" + ] + }, + { + "cell_type": "code", + "execution_count": 353, + "metadata": {}, + "outputs": [ + { + "name": "stderr", + "output_type": "stream", + "text": [ + "[2018-01-24 14:06:40,096] Making new env: BanditTenArmedGaussian-v0\n" + ] + }, + { + "data": { + "text/plain": [ + "0" + ] + }, + "execution_count": 353, + "metadata": {}, + "output_type": "execute_result" + } + ], + "source": [ + "env = gym.make(\"BanditTenArmedGaussian-v0\")\n", + "env.reset()" + ] + }, + { + "cell_type": "code", + "execution_count": 354, + "metadata": { + "collapsed": true + }, + "outputs": [], + "source": [ + "# Helper functions\n", + "\n", + "def moving_average(xs, n=100):\n", + " ret = np.cumsum(xs, dtype=float)\n", + " ret[n:] = ret[n:] - ret[:-n]\n", + " return ret[n - 1:] / n\n", + "\n", + "def divide_by_index(x): \n", + " output = np.zeros(len(x))\n", + " for n in range(len(x)):\n", + " output[n] = float(x[n]) / (n + 1)\n", + " return output\n", + "\n", + "def avg_across_arrays(xs): # where xs is an array of arrays, all of same length\n", + " length = len(xs[0])\n", + " num_arrays = len(xs)\n", + " output = np.zeros(length)\n", + " for j in range(length):\n", + " xsum = 0.\n", + " for i in range(num_arrays):\n", + " xsum += xs[i][j]\n", + " output[j] = xsum / num_arrays\n", + " return output\n", + "\n", + "def softmax(x):\n", + " return np.exp(x - x.max()) / np.sum(np.exp(x - x.max())) # slightly modified to mitigate chances of overflow" + ] + }, + { + "cell_type": "code", + "execution_count": 355, + "metadata": { + "collapsed": true + }, + "outputs": [], + "source": [ + "# Defining the agents\n", + "\n", + "class SimpleAgent(object):\n", + " def __init__(self, num_actions, Q_init=0):\n", + " self.num_actions = num_actions\n", + " self.A = range(num_actions)\n", + " self.N = np.zeros(num_actions) # the number of times N(a) has been called\n", + " self.Q = np.full(num_action, Q_init) # the expected value of choosing Q(a)\n", + " \n", + " def choose_action(self, epsilon): # choose random action with probability epsilon \n", + " rand = np.random.rand()\n", + " if (rand < epsilon):\n", + " choice = np.random.choice(self.A)\n", + " return choice\n", + " else:\n", + " choice = np.argmax(self.Q)\n", + " return choice\n", + " \n", + " def learn(self, action, reward): # update Q and N\n", + " self.N[a] += 1\n", + " self.Q[a] += (1 / self.N[a]) * (reward - self.Q[a])\n", + " \n", + " def alpha_learn(self, action, reward, alpha=0.1): # update Q and N\n", + " self.N[a] += 1\n", + " self.Q[a] += alpha * (reward - self.Q[a])\n", + "\n", + " \n", + "class GradientAgent(SimpleAgent):\n", + " def __init__(self, num_actions, Q_init=0):\n", + " self.num_actions = num_actions\n", + " self.A = range(num_actions)\n", + " self.H = softmax(np.zeros(num_actions))\n", + " self.avg_reward = 0\n", + " self.timestep = 0\n", + " \n", + " def choose_action(self): \n", + " return np.random.choice(self.num_actions, p=softmax(self.H))\n", + " \n", + " def learn(self, action, reward, alpha=0.4):\n", + " self.timestep += 1\n", + " self.avg_reward = self.avg_reward + (1. / self.timestep) * (reward - self.avg_reward)\n", + " for a in self.A:\n", + " if a == action:\n", + " self.H[a] = self.H[a] + alpha * (reward - self.avg_reward) * (1 - softmax(self.H)[a])\n", + " else:\n", + " self.H[a] = self.H[a] - alpha * (reward - self.avg_reward) * (softmax(self.H)[a])\n", + " " + ] + }, + { + "cell_type": "code", + "execution_count": 356, + "metadata": {}, + "outputs": [ + { + "name": "stdout", + "output_type": "stream", + "text": [ + "optimal choice is action at index: 4\n", + "reward distribution: [-1.2717412660464487, 0.8109740983565349, -1.6815563579207893, -1.0818436030941483, 1.492005935762376, -1.269804656864728, 0.5953374965613057, 0.2654389056279407, 1.0576079326748915, -0.8055287044820053]\n" + ] + } + ], + "source": [ + "# Get reward distribution\n", + "\n", + "r_dist = list(map(lambda x: x[0], env.env.r_dist))\n", + " \n", + "optimal_choice = np.argmax(r_dist)\n", + "print(\"optimal choice is action at index:\", optimal_choice)\n", + "print(\"reward distribution:\", r_dist)\n" + ] + }, + { + "cell_type": "code", + "execution_count": 357, + "metadata": { + "collapsed": true + }, + "outputs": [], + "source": [ + "# Event loop for i episodes each with t timesteps\n", + "\n", + "timesteps = 1000\n", + "episodes = 100\n", + "\n", + "agent = GradientAgent(env.action_space.n)\n", + "e = 0.1 # epsilon factor for exploration\n", + "\n", + "all_optimal_trackers = []\n", + "all_total_rewards = []\n", + "all_rewards = []\n", + "\n", + "for i in range(episodes):\n", + " \n", + " optimal_picks = 0\n", + " optimal_tracker = []\n", + " \n", + " total_reward = 0\n", + " total_rewards = []\n", + " rewards = []\n", + " \n", + " for t in range(timesteps):\n", + " a = agent.choose_action()\n", + "\n", + " if a == optimal_choice: optimal_picks += 1\n", + " optimal_tracker.append(optimal_picks)\n", + "\n", + " observation, reward, done, info = env.step(a)\n", + "\n", + " total_reward += reward\n", + " rewards.append(reward)\n", + " total_rewards.append(total_reward)\n", + " agent.learn(a, reward)\n", + " \n", + " all_optimal_trackers.append(divide_by_index(optimal_tracker))\n", + " all_total_rewards.append(divide_by_index(total_rewards))\n", + " all_rewards.append(moving_average(rewards))\n" + ] + }, + { + "cell_type": "code", + "execution_count": 358, + "metadata": { + "collapsed": true + }, + "outputs": [], + "source": [ + "# Get averages across all episodes\n", + "\n", + "avg_rewards = avg_across_arrays(all_rewards)\n", + "avg_total_rewards = avg_across_arrays(all_total_rewards)\n", + "avg_optimal = avg_across_arrays(all_optimal_trackers)" + ] + }, + { + "cell_type": "code", + "execution_count": 359, + "metadata": {}, + "outputs": [ + { + "data": { + "image/png": "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\n", + "text/plain": [ + "" + ] + }, + "metadata": {}, + "output_type": "display_data" + } + ], + "source": [ + "# Plotting\n", + "\n", + "plt.figure(1, figsize=(15,10))\n", + "\n", + "# probability distribution\n", + "plt.subplot(321)\n", + "plt.plot(env.env.p_dist, 'o')\n", + "plt.title('probability distribution of bandit arms')\n", + "plt.xlabel('action')\n", + "plt.ylabel('reward')\n", + "plt.grid(True)\n", + "\n", + "\n", + "# reward distribution\n", + "plt.subplot(322)\n", + "plt.plot(r_dist, 'ro')\n", + "plt.title('mean of reward distribution by action')\n", + "plt.xlabel('action')\n", + "plt.ylabel('reward')\n", + "plt.grid(True)\n", + "\n", + "\n", + "# moving average of rewards\n", + "plt.subplot(323)\n", + "plt.plot(avg_rewards)\n", + "plt.title('moving average of rewards')\n", + "plt.xlabel('timestep')\n", + "plt.ylabel('reward')\n", + "plt.grid(True)\n", + "\n", + "# moving average of total rewards\n", + "plt.subplot(324)\n", + "plt.plot(avg_total_rewards) # to get average reward\n", + "plt.title('average of total rewards')\n", + "plt.xlabel('timestep')\n", + "plt.ylabel('reward')\n", + "plt.grid(True)\n", + "\n", + "# % of optimal choices\n", + "plt.subplot(325)\n", + "plt.plot(avg_optimal) # to get % times optimal choice was chosen\n", + "plt.title('% of optimal choices')\n", + "plt.xlabel('timestep')\n", + "plt.ylabel('% optimal choice')\n", + "plt.grid(True)\n", + "\n", + "plt.subplots_adjust(top=0.99, bottom=.0, left=0.10, right=0.95, hspace=0.5,\n", + " wspace=0.5)\n", + "\n", + "plt.show()" + ] + }, + { + "cell_type": "code", + "execution_count": null, + "metadata": { + "collapsed": true + }, + "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": "3.6.3" + } + }, + "nbformat": 4, + "nbformat_minor": 2 +} From 45c8c2310202143baaacc95984e465a8b1c0dd3c Mon Sep 17 00:00:00 2001 From: Ryan Enderby Date: Mon, 1 Oct 2018 14:14:49 -0400 Subject: [PATCH 2/2] moving notebook to appropriate folder --- {MDP => Bandits}/k-bandit.ipynb | 0 1 file changed, 0 insertions(+), 0 deletions(-) rename {MDP => Bandits}/k-bandit.ipynb (100%) diff --git a/MDP/k-bandit.ipynb b/Bandits/k-bandit.ipynb similarity index 100% rename from MDP/k-bandit.ipynb rename to Bandits/k-bandit.ipynb