{ "cells": [ { "cell_type": "code", "execution_count": 1, "metadata": { "papermill": { "duration": 0.670894, "end_time": "2019-10-23T16:01:12.210448", "exception": false, "start_time": "2019-10-23T16:01:11.539554", "status": "completed" }, "tags": [] }, "outputs": [], "source": [ "%matplotlib inline\n", "\n", "import matplotlib.pyplot as plt\n", "\n", "import gym\n", "from gym.envs.registration import register" ] }, { "cell_type": "markdown", "metadata": { "papermill": { "duration": 0.016798, "end_time": "2019-10-23T16:01:12.244111", "exception": false, "start_time": "2019-10-23T16:01:12.227313", "status": "completed" }, "tags": [] }, "source": [ "# ACS2 in Frozen Lake\n", "\n", "About the environment\n", "> The agent controls the movement of a character in a grid world. Some tiles of the grid are walkable, and others lead to the agent falling into the water. Additionally, the movement direction of the agent is uncertain and only partially depends on the chosen direction. The agent is rewarded for finding a walkable path to a goal tile." ] }, { "cell_type": "code", "execution_count": 2, "metadata": { "papermill": { "duration": 0.72504, "end_time": "2019-10-23T16:01:12.990433", "exception": false, "start_time": "2019-10-23T16:01:12.265393", "status": "completed" }, "tags": [] }, "outputs": [ { "name": "stdout", "output_type": "stream", "text": [ "\n", "\u001b[41mS\u001b[0mFFF\n", "FHFH\n", "FFFH\n", "HFFG\n" ] } ], "source": [ "fl_env = gym.make('FrozenLake-v0')\n", "\n", "# Reset the state\n", "state = fl_env.reset()\n", "\n", "# Render the environment\n", "fl_env.render()" ] }, { "cell_type": "markdown", "metadata": { "papermill": { "duration": 0.022612, "end_time": "2019-10-23T16:01:13.030417", "exception": false, "start_time": "2019-10-23T16:01:13.007805", "status": "completed" }, "tags": [] }, "source": [ "Each state might get following possible values: `{S, F, H, G}` which, refers to\n", "```\n", "SFFF (S: starting point, safe)\n", "FHFH (F: frozen surface, safe)\n", "FFFH (H: hole, fall to your doom)\n", "HFFG (G: goal, where the frisbee is located)\n", "```\n", "\n", "In case of interacting with environment agent cant perform 4 action which map as follow:\n", "- 0 - left\n", "- 1 - down\n", "- 2 - right\n", "- 3 - up\n", "\n", "> FrozenLake-v0 defines \"solving\" as getting average reward of 0.78 over 100 consecutive trials.\n", "\n", "We will also define a second version of the same environment but with `slippery=False` parameters. That make it more deterministic." ] }, { "cell_type": "code", "execution_count": 3, "metadata": { "papermill": { "duration": 0.027484, "end_time": "2019-10-23T16:01:13.075413", "exception": false, "start_time": "2019-10-23T16:01:13.047929", "status": "completed" }, "tags": [] }, "outputs": [ { "name": "stdout", "output_type": "stream", "text": [ "\n", "\u001b[41mS\u001b[0mFFF\n", "FHFH\n", "FFFH\n", "HFFG\n" ] } ], "source": [ "register(\n", " id='FrozenLakeNotSlippery-v0',\n", " entry_point='gym.envs.toy_text:FrozenLakeEnv',\n", " kwargs={'map_name': '4x4', 'is_slippery': False},\n", " max_episode_steps=100,\n", " reward_threshold=0.78, # optimum = .8196\n", ")\n", "\n", "fl_ns_env = gym.make('FrozenLakeNotSlippery-v0')\n", "\n", "# Reset the state\n", "state = fl_ns_env.reset()\n", "\n", "# Render the environment\n", "fl_ns_env.render()" ] }, { "cell_type": "markdown", "metadata": { "papermill": { "duration": 0.017077, "end_time": "2019-10-23T16:01:13.109843", "exception": false, "start_time": "2019-10-23T16:01:13.092766", "status": "completed" }, "tags": [] }, "source": [ "## ACS2" ] }, { "cell_type": "code", "execution_count": 4, "metadata": { "papermill": { "duration": 0.076895, "end_time": "2019-10-23T16:01:13.203862", "exception": false, "start_time": "2019-10-23T16:01:13.126967", "status": "completed" }, "tags": [] }, "outputs": [], "source": [ "# Import PyALCS code from local path\n", "import sys, os\n", "sys.path.append(os.path.abspath('../'))\n", "\n", "from lcs.agents import EnvironmentAdapter\n", "from lcs.agents.acs2 import ACS2, Configuration\n", "\n", "# Enable automatic module reload\n", "%load_ext autoreload\n", "%autoreload 2\n", "\n", "CLASSIFIER_LENGTH = 16 # Because we are operating in 4x4 grid\n", "POSSIBLE_ACTIONS = fl_env.action_space.n # 4" ] }, { "cell_type": "markdown", "metadata": { "papermill": { "duration": 0.016884, "end_time": "2019-10-23T16:01:13.237853", "exception": false, "start_time": "2019-10-23T16:01:13.220969", "status": "completed" }, "tags": [] }, "source": [ "### Encoding perception\n", "The only information back from the environment is the current agent position (not it's perception). Therefore our agent task will be to predicit where it will land after executing each action.\n", "\n", "To do so we will represent state as a one-hot encoded vector." ] }, { "cell_type": "code", "execution_count": 5, "metadata": { "papermill": { "duration": 0.047926, "end_time": "2019-10-23T16:01:13.303017", "exception": false, "start_time": "2019-10-23T16:01:13.255091", "status": "completed" }, "tags": [] }, "outputs": [], "source": [ "class FrozenLakeAdapter(EnvironmentAdapter):\n", " @classmethod\n", " def to_genotype(cls, phenotype):\n", " genotype = ['0' for i in range(CLASSIFIER_LENGTH)]\n", " genotype[phenotype] = 'X'\n", " return ''.join(genotype)" ] }, { "cell_type": "markdown", "metadata": { "papermill": { "duration": 0.017365, "end_time": "2019-10-23T16:01:13.337377", "exception": false, "start_time": "2019-10-23T16:01:13.320012", "status": "completed" }, "tags": [] }, "source": [ "`X` corresponds to current agent position. State 4 is encoded as follows:" ] }, { "cell_type": "code", "execution_count": 6, "metadata": { "papermill": { "duration": 0.043785, "end_time": "2019-10-23T16:01:13.398491", "exception": false, "start_time": "2019-10-23T16:01:13.354706", "status": "completed" }, "tags": [] }, "outputs": [ { "data": { "text/plain": [ "'0000X00000000000'" ] }, "execution_count": 6, "metadata": {}, "output_type": "execute_result" } ], "source": [ "FrozenLakeAdapter().to_genotype(4)" ] }, { "cell_type": "markdown", "metadata": { "papermill": { "duration": 0.018894, "end_time": "2019-10-23T16:01:13.434852", "exception": false, "start_time": "2019-10-23T16:01:13.415958", "status": "completed" }, "tags": [] }, "source": [ "### Environment metrics\n", "We will also need a function for evaluating if agent finished succesfuly a trial" ] }, { "cell_type": "code", "execution_count": 7, "metadata": { "papermill": { "duration": 0.043406, "end_time": "2019-10-23T16:01:13.499052", "exception": false, "start_time": "2019-10-23T16:01:13.455646", "status": "completed" }, "tags": [] }, "outputs": [], "source": [ "from lcs.metrics import population_metrics\n", "\n", "\n", "# We assume if the final state was with number 15 that the algorithm found the reward. Otherwise not\n", "def fl_metrics(pop, env):\n", " metrics = {\n", " 'found_reward': env.env.s == 15,\n", " }\n", " \n", " # Add basic population metrics\n", " metrics.update(population_metrics(pop, env))\n", " \n", " return metrics" ] }, { "cell_type": "markdown", "metadata": { "papermill": { "duration": 0.018531, "end_time": "2019-10-23T16:01:13.535569", "exception": false, "start_time": "2019-10-23T16:01:13.517038", "status": "completed" }, "tags": [] }, "source": [ "### Performance evaluation" ] }, { "cell_type": "code", "execution_count": 8, "metadata": { "papermill": { "duration": 0.04482, "end_time": "2019-10-23T16:01:13.601254", "exception": false, "start_time": "2019-10-23T16:01:13.556434", "status": "completed" }, "tags": [] }, "outputs": [], "source": [ "def print_performance(population, metrics):\n", " population.sort(key=lambda cl: -cl.fitness)\n", " population_count = len(population)\n", " reliable_count = len([cl for cl in population if cl.is_reliable()])\n", " successful_trials = sum(m['found_reward'] for m in metrics)\n", "\n", " print(\"Number of classifiers: {}\".format(population_count))\n", " print(\"Number of reliable classifiers: {}\".format(reliable_count))\n", " print(\"Percentage of successul trials: {:.2f}%\".format(successful_trials / EXPLOIT_TRIALS * 100))\n", " print(\"\\nTop 10 classifiers:\")\n", " for cl in population[:10]:\n", " print(\"{!r} \\tq: {:.2f} \\tr: {:.2f} \\tir: {:.2f} \\texp: {}\".format(cl, cl.q, cl.r, cl.ir, cl.exp))" ] }, { "cell_type": "code", "execution_count": 9, "metadata": { "papermill": { "duration": 0.046577, "end_time": "2019-10-23T16:01:13.665121", "exception": false, "start_time": "2019-10-23T16:01:13.618544", "status": "completed" }, "tags": [] }, "outputs": [], "source": [ "def plot_success_trials(metrics, ax=None):\n", " if ax is None:\n", " ax = plt.gca()\n", " \n", " trials = [m['trial'] for m in metrics]\n", " success = [m['found_reward'] for m in metrics]\n", "\n", " ax.plot(trials, success)\n", " ax.set_title(\"Successful Trials\")\n", " ax.set_xlabel(\"Trial\")\n", " ax.set_ylabel(\"Agent found reward\")" ] }, { "cell_type": "code", "execution_count": 10, "metadata": { "papermill": { "duration": 0.044081, "end_time": "2019-10-23T16:01:13.726645", "exception": false, "start_time": "2019-10-23T16:01:13.682564", "status": "completed" }, "tags": [] }, "outputs": [], "source": [ "def plot_population(metrics, ax=None):\n", " if ax is None:\n", " ax = plt.gca()\n", " \n", " trials = [m['trial'] for m in metrics]\n", " \n", " population_size = [m['numerosity'] for m in metrics]\n", " reliable_size = [m['reliable'] for m in metrics]\n", " \n", " ax.plot(trials, population_size, 'b', label='all')\n", " ax.plot(trials, reliable_size, 'r', label='reliable')\n", " \n", " ax.set_title(\"Population size\")\n", " ax.set_xlabel(\"Trial\")\n", " ax.set_ylabel(\"Number of macroclassifiers\")\n", " ax.legend(loc='best')" ] }, { "cell_type": "code", "execution_count": 11, "metadata": { "papermill": { "duration": 0.045741, "end_time": "2019-10-23T16:01:13.789752", "exception": false, "start_time": "2019-10-23T16:01:13.744011", "status": "completed" }, "tags": [] }, "outputs": [], "source": [ "def plot_performance(metrics):\n", " plt.figure(figsize=(13, 10), dpi=100)\n", " plt.suptitle('Performance Visualization')\n", " \n", " ax1 = plt.subplot(221)\n", " plot_success_trials(metrics, ax1)\n", " \n", " ax2 = plt.subplot(222)\n", " plot_population(metrics, ax2)\n", " \n", " plt.show()" ] }, { "cell_type": "markdown", "metadata": { "papermill": { "duration": 0.017168, "end_time": "2019-10-23T16:01:13.824416", "exception": false, "start_time": "2019-10-23T16:01:13.807248", "status": "completed" }, "tags": [] }, "source": [ "### Default ACS2 configuration\n", "Right now we are ready to configure the ACS2 agent providing some defaults" ] }, { "cell_type": "code", "execution_count": 12, "metadata": { "papermill": { "duration": 0.04294, "end_time": "2019-10-23T16:01:13.884941", "exception": false, "start_time": "2019-10-23T16:01:13.842001", "status": "completed" }, "tags": [] }, "outputs": [ { "name": "stdout", "output_type": "stream", "text": [ "ACS2Configuration:\n", "\t- Classifier length: [16]\n", "\t- Number of possible actions: [4]\n", "\t- Classifier wildcard: [#]\n", "\t- Environment adapter function: [<__main__.FrozenLakeAdapter object at 0x117bd0b00>]\n", "\t- Fitness function: [None]\n", "\t- Do GA: [False]\n", "\t- Do subsumption: [True]\n", "\t- Do Action Planning: [False]\n", "\t- Beta: [0.05]\n", "\t- ...\n", "\t- Epsilon: [0.7]\n", "\t- U_max: [100000]\n" ] } ], "source": [ "cfg = Configuration(\n", " classifier_length=CLASSIFIER_LENGTH,\n", " number_of_possible_actions=POSSIBLE_ACTIONS,\n", " environment_adapter=FrozenLakeAdapter(),\n", " metrics_trial_frequency=1,\n", " user_metrics_collector_fcn=fl_metrics,\n", " theta_i=0.3,\n", " epsilon=0.7)\n", "\n", "print(cfg)" ] }, { "cell_type": "markdown", "metadata": { "papermill": { "duration": 0.020047, "end_time": "2019-10-23T16:01:13.922588", "exception": false, "start_time": "2019-10-23T16:01:13.902541", "status": "completed" }, "tags": [] }, "source": [ "## Experiments" ] }, { "cell_type": "code", "execution_count": 13, "metadata": { "papermill": { "duration": 0.044259, "end_time": "2019-10-23T16:01:13.985648", "exception": false, "start_time": "2019-10-23T16:01:13.941389", "status": "completed" }, "tags": [] }, "outputs": [], "source": [ "EXPLORE_TRIALS = 2000\n", "EXPLOIT_TRIALS = 100\n", "\n", "\n", "def perform_experiment(cfg, env):\n", " # explore phase\n", " agent = ACS2(cfg)\n", " population_explore, metrics_explore = agent.explore(env, EXPLORE_TRIALS)\n", " \n", " # exploit phase, reinitialize agent with population above\n", " agent = ACS2(cfg, population=population_explore)\n", " population_exploit, metrics_exploit = agent.exploit(env, EXPLOIT_TRIALS)\n", " \n", " return (population_explore, metrics_explore), (population_exploit, metrics_exploit)" ] }, { "cell_type": "markdown", "metadata": { "papermill": { "duration": 0.018167, "end_time": "2019-10-23T16:01:14.022350", "exception": false, "start_time": "2019-10-23T16:01:14.004183", "status": "completed" }, "tags": [] }, "source": [ "### FrozenLake-v0 environment (baseline)" ] }, { "cell_type": "code", "execution_count": 14, "metadata": { "papermill": { "duration": 46.358754, "end_time": "2019-10-23T16:02:00.401837", "exception": false, "start_time": "2019-10-23T16:01:14.043083", "status": "completed" }, "tags": [] }, "outputs": [ { "name": "stdout", "output_type": "stream", "text": [ "CPU times: user 46.1 s, sys: 138 ms, total: 46.2 s\n", "Wall time: 46.3 s\n" ] } ], "source": [ "%%time\n", "explore_results, exploit_results = perform_experiment(cfg, fl_env) " ] }, { "cell_type": "markdown", "metadata": { "papermill": { "duration": 0.018931, "end_time": "2019-10-23T16:02:00.445959", "exception": false, "start_time": "2019-10-23T16:02:00.427028", "status": "completed" }, "tags": [] }, "source": [ "Learn some behaviour during exploration phase" ] }, { "cell_type": "code", "execution_count": 15, "metadata": { "papermill": { "duration": 0.046931, "end_time": "2019-10-23T16:02:00.514419", "exception": false, "start_time": "2019-10-23T16:02:00.467488", "status": "completed" }, "tags": [] }, "outputs": [ { "name": "stdout", "output_type": "stream", "text": [ "Number of classifiers: 441\n", "Number of reliable classifiers: 0\n", "Percentage of successul trials: 42.00%\n", "\n", "Top 10 classifiers:\n", "##############X0 1 ##############0X (00000000000000##) q: 0.602 r: 0.3693 ir: 0.2825 f: 0.2224 exp: 41 tga: 1105 talp: 15265 tav: 2.99e+02 num: 1 \tq: 0.60 \tr: 0.37 \tir: 0.28 \texp: 41\n", "#0############X0 1 ##############0X (0#000000000000##) q: 0.577 r: 0.3633 ir: 0.2765 f: 0.2096 exp: 32 tga: 3961 talp: 15265 tav: 3.26e+02 num: 1 \tq: 0.58 \tr: 0.36 \tir: 0.28 \texp: 32\n", "#0###########0X# 3 #############X0# (0#00000000000##0) q: 0.471 r: 0.3723 ir: 0.3615 f: 0.1753 exp: 43 tga: 3660 talp: 14982 tav: 2.96e+02 num: 1 \tq: 0.47 \tr: 0.37 \tir: 0.36 \texp: 43\n", "##############X0 3 ##############0X (00000000000000##) q: 0.442 r: 0.3876 ir: 0.377 f: 0.1715 exp: 45 tga: 2040 talp: 14982 tav: 3.13e+02 num: 1 \tq: 0.44 \tr: 0.39 \tir: 0.38 \texp: 45\n", "#############0X# 3 #############X0# (0000000000000##0) q: 0.446 r: 0.3829 ir: 0.3723 f: 0.1709 exp: 45 tga: 3117 talp: 14982 tav: 2.98e+02 num: 1 \tq: 0.45 \tr: 0.38 \tir: 0.37 \texp: 45\n", "#############0X# 1 #############X0# (0000000000000##0) q: 0.452 r: 0.3676 ir: 0.2807 f: 0.1661 exp: 33 tga: 3251 talp: 15265 tav: 3.33e+02 num: 1 \tq: 0.45 \tr: 0.37 \tir: 0.28 \texp: 33\n", "#0###########0X# 1 #############X0# (0#00000000000##0) q: 0.451 r: 0.3633 ir: 0.2765 f: 0.1639 exp: 33 tga: 3251 talp: 15265 tav: 3.26e+02 num: 1 \tq: 0.45 \tr: 0.36 \tir: 0.28 \texp: 33\n", "##############X# 1 ################ (00000000000000#0) q: 0.368 r: 0.3676 ir: 0.2807 f: 0.1351 exp: 40 tga: 1748 talp: 15265 tav: 2.93e+02 num: 1 \tq: 0.37 \tr: 0.37 \tir: 0.28 \texp: 40\n", "#0############X# 1 ################ (0#000000000000#0) q: 0.355 r: 0.3633 ir: 0.2765 f: 0.129 exp: 36 tga: 3119 talp: 15265 tav: 2.97e+02 num: 1 \tq: 0.36 \tr: 0.36 \tir: 0.28 \texp: 36\n", "##########0###X# 2 ##########X###0# (0000000000#000#0) q: 0.412 r: 0.2099 ir: 0.1545 f: 0.0865 exp: 18 tga: 631 talp: 14255 tav: 7.17e+02 num: 1 \tq: 0.41 \tr: 0.21 \tir: 0.15 \texp: 18\n" ] } ], "source": [ "# exploration\n", "print_performance(explore_results[0], explore_results[1])" ] }, { "cell_type": "code", "execution_count": 16, "metadata": { "papermill": { "duration": 0.75281, "end_time": "2019-10-23T16:02:01.286164", "exception": false, "start_time": "2019-10-23T16:02:00.533354", "status": "completed" }, "tags": [] }, "outputs": [ { "data": { "image/png": "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\n", "text/plain": [ "
" ] }, "metadata": { "needs_background": "light" }, "output_type": "display_data" } ], "source": [ "plot_performance(explore_results[1])" ] }, { "cell_type": "markdown", "metadata": { "papermill": { "duration": 0.021037, "end_time": "2019-10-23T16:02:01.326396", "exception": false, "start_time": "2019-10-23T16:02:01.305359", "status": "completed" }, "tags": [] }, "source": [ "Metrics from exploitation" ] }, { "cell_type": "code", "execution_count": 17, "metadata": { "papermill": { "duration": 0.044591, "end_time": "2019-10-23T16:02:01.391606", "exception": false, "start_time": "2019-10-23T16:02:01.347015", "status": "completed" }, "tags": [] }, "outputs": [ { "name": "stdout", "output_type": "stream", "text": [ "Number of classifiers: 441\n", "Number of reliable classifiers: 0\n", "Percentage of successul trials: 6.00%\n", "\n", "Top 10 classifiers:\n", "##############X0 1 ##############0X (00000000000000##) q: 0.602 r: 0.3693 ir: 0.2825 f: 0.2224 exp: 41 tga: 1105 talp: 15265 tav: 2.99e+02 num: 1 \tq: 0.60 \tr: 0.37 \tir: 0.28 \texp: 41\n", "#0############X0 1 ##############0X (0#000000000000##) q: 0.577 r: 0.3633 ir: 0.2765 f: 0.2096 exp: 32 tga: 3961 talp: 15265 tav: 3.26e+02 num: 1 \tq: 0.58 \tr: 0.36 \tir: 0.28 \texp: 32\n", "#0###########0X# 3 #############X0# (0#00000000000##0) q: 0.471 r: 0.3723 ir: 0.3615 f: 0.1753 exp: 43 tga: 3660 talp: 14982 tav: 2.96e+02 num: 1 \tq: 0.47 \tr: 0.37 \tir: 0.36 \texp: 43\n", "##############X0 3 ##############0X (00000000000000##) q: 0.442 r: 0.3876 ir: 0.377 f: 0.1715 exp: 45 tga: 2040 talp: 14982 tav: 3.13e+02 num: 1 \tq: 0.44 \tr: 0.39 \tir: 0.38 \texp: 45\n", "#############0X# 3 #############X0# (0000000000000##0) q: 0.446 r: 0.3829 ir: 0.3723 f: 0.1709 exp: 45 tga: 3117 talp: 14982 tav: 2.98e+02 num: 1 \tq: 0.45 \tr: 0.38 \tir: 0.37 \texp: 45\n", "#############0X# 1 #############X0# (0000000000000##0) q: 0.452 r: 0.3676 ir: 0.2807 f: 0.1661 exp: 33 tga: 3251 talp: 15265 tav: 3.33e+02 num: 1 \tq: 0.45 \tr: 0.37 \tir: 0.28 \texp: 33\n", "#0###########0X# 1 #############X0# (0#00000000000##0) q: 0.451 r: 0.3633 ir: 0.2765 f: 0.1639 exp: 33 tga: 3251 talp: 15265 tav: 3.26e+02 num: 1 \tq: 0.45 \tr: 0.36 \tir: 0.28 \texp: 33\n", "##############X# 1 ################ (00000000000000#0) q: 0.368 r: 0.3676 ir: 0.2807 f: 0.1351 exp: 40 tga: 1748 talp: 15265 tav: 2.93e+02 num: 1 \tq: 0.37 \tr: 0.37 \tir: 0.28 \texp: 40\n", "#0############X# 1 ################ (0#000000000000#0) q: 0.355 r: 0.3633 ir: 0.2765 f: 0.129 exp: 36 tga: 3119 talp: 15265 tav: 2.97e+02 num: 1 \tq: 0.36 \tr: 0.36 \tir: 0.28 \texp: 36\n", "##########0###X# 2 ##########X###0# (0000000000#000#0) q: 0.412 r: 0.2099 ir: 0.1545 f: 0.0865 exp: 18 tga: 631 talp: 14255 tav: 7.17e+02 num: 1 \tq: 0.41 \tr: 0.21 \tir: 0.15 \texp: 18\n" ] } ], "source": [ "# exploitation\n", "print_performance(exploit_results[0], exploit_results[1])" ] }, { "cell_type": "markdown", "metadata": { "papermill": { "duration": 0.019329, "end_time": "2019-10-23T16:02:01.430115", "exception": false, "start_time": "2019-10-23T16:02:01.410786", "status": "completed" }, "tags": [] }, "source": [ "### FrozenLakeNotSlippery-v0 environment" ] }, { "cell_type": "code", "execution_count": 18, "metadata": { "papermill": { "duration": 10.20416, "end_time": "2019-10-23T16:02:11.656905", "exception": false, "start_time": "2019-10-23T16:02:01.452745", "status": "completed" }, "tags": [] }, "outputs": [ { "name": "stdout", "output_type": "stream", "text": [ "CPU times: user 9.99 s, sys: 136 ms, total: 10.1 s\n", "Wall time: 10.2 s\n" ] } ], "source": [ "%%time\n", "explore_results_2, exploit_results_2 = perform_experiment(cfg, fl_ns_env)" ] }, { "cell_type": "code", "execution_count": 19, "metadata": { "papermill": { "duration": 0.050992, "end_time": "2019-10-23T16:02:11.727616", "exception": false, "start_time": "2019-10-23T16:02:11.676624", "status": "completed" }, "tags": [] }, "outputs": [ { "name": "stdout", "output_type": "stream", "text": [ "Number of classifiers: 89\n", "Number of reliable classifiers: 89\n", "Percentage of successul trials: 192.00%\n", "\n", "Top 10 classifiers:\n", "##############X0 2 ##############0X (empty) q: 1.0 r: 1.0 ir: 1.0 f: 1.0 exp: 191 tga: 237 talp: 14842 tav: 79.9 num: 1 \tq: 1.00 \tr: 1.00 \tir: 1.00 \texp: 191\n", "##########X###0# 1 ##########0###X# (empty) q: 1.0 r: 0.95 ir: 0.0 f: 0.95 exp: 245 tga: 324 talp: 15005 tav: 73.1 num: 1 \tq: 1.00 \tr: 0.95 \tir: 0.00 \texp: 245\n", "#############X0# 2 #############0X# (empty) q: 0.999 r: 0.9459 ir: 0.0 f: 0.9453 exp: 130 tga: 236 talp: 14842 tav: 1.38e+02 num: 1 \tq: 1.00 \tr: 0.95 \tir: 0.00 \texp: 130\n", "##############X# 1 ################ (empty) q: 0.997 r: 0.9274 ir: 0.0 f: 0.9245 exp: 87 tga: 572 talp: 15006 tav: 2.14e+02 num: 1 \tq: 1.00 \tr: 0.93 \tir: 0.00 \texp: 87\n", "#########X0##### 2 #########0X##### (empty) q: 1.0 r: 0.9025 ir: 0.0 f: 0.9025 exp: 399 tga: 117 talp: 15035 tav: 50.2 num: 1 \tq: 1.00 \tr: 0.90 \tir: 0.00 \texp: 399\n", "######X###0##### 1 ######0###X##### (empty) q: 1.0 r: 0.8974 ir: 0.0 f: 0.8971 exp: 137 tga: 26 talp: 14979 tav: 89.2 num: 1 \tq: 1.00 \tr: 0.90 \tir: 0.00 \texp: 137\n", "#########X###0## 1 #########0###X## (empty) q: 1.0 r: 0.8892 ir: 0.0 f: 0.8892 exp: 168 tga: 115 talp: 14994 tav: 93.6 num: 1 \tq: 1.00 \tr: 0.89 \tir: 0.00 \texp: 168\n", "####0#####0###X# 3 ##########X###0# (empty) q: 0.997 r: 0.8828 ir: 0.0 f: 0.8797 exp: 98 tga: 485 talp: 14543 tav: 1.62e+02 num: 1 \tq: 1.00 \tr: 0.88 \tir: 0.00 \texp: 98\n", "##########0###X# 3 ##########X###0# (empty) q: 0.997 r: 0.8828 ir: 0.0 f: 0.8797 exp: 97 tga: 485 talp: 14543 tav: 1.62e+02 num: 1 \tq: 1.00 \tr: 0.88 \tir: 0.00 \texp: 97\n", "########X0###### 2 ########0X###### (empty) q: 1.0 r: 0.8573 ir: 0.0 f: 0.8573 exp: 740 tga: 114 talp: 15034 tav: 28.7 num: 1 \tq: 1.00 \tr: 0.86 \tir: 0.00 \texp: 740\n" ] } ], "source": [ "# exploration\n", "print_performance(explore_results_2[0], explore_results_2[1])" ] }, { "cell_type": "code", "execution_count": 20, "metadata": { "papermill": { "duration": 0.742154, "end_time": "2019-10-23T16:02:12.489456", "exception": false, "start_time": "2019-10-23T16:02:11.747302", "status": "completed" }, "tags": [] }, "outputs": [ { "data": { "image/png": "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\n", "text/plain": [ "
" ] }, "metadata": { "needs_background": "light" }, "output_type": "display_data" } ], "source": [ "plot_performance(explore_results_2[1])" ] }, { "cell_type": "code", "execution_count": 21, "metadata": { "papermill": { "duration": 0.050858, "end_time": "2019-10-23T16:02:12.561596", "exception": false, "start_time": "2019-10-23T16:02:12.510738", "status": "completed" }, "tags": [] }, "outputs": [ { "name": "stdout", "output_type": "stream", "text": [ "Number of classifiers: 89\n", "Number of reliable classifiers: 89\n", "Percentage of successul trials: 100.00%\n", "\n", "Top 10 classifiers:\n", "##############X0 2 ##############0X (empty) q: 1.0 r: 1.0 ir: 1.0 f: 1.0 exp: 191 tga: 237 talp: 14842 tav: 79.9 num: 1 \tq: 1.00 \tr: 1.00 \tir: 1.00 \texp: 191\n", "##########X###0# 1 ##########0###X# (empty) q: 1.0 r: 0.95 ir: 0.0 f: 0.95 exp: 245 tga: 324 talp: 15005 tav: 73.1 num: 1 \tq: 1.00 \tr: 0.95 \tir: 0.00 \texp: 245\n", "#############X0# 2 #############0X# (empty) q: 0.999 r: 0.9459 ir: 0.0 f: 0.9453 exp: 130 tga: 236 talp: 14842 tav: 1.38e+02 num: 1 \tq: 1.00 \tr: 0.95 \tir: 0.00 \texp: 130\n", "##############X# 1 ################ (empty) q: 0.997 r: 0.9274 ir: 0.0 f: 0.9245 exp: 87 tga: 572 talp: 15006 tav: 2.14e+02 num: 1 \tq: 1.00 \tr: 0.93 \tir: 0.00 \texp: 87\n", "#########X0##### 2 #########0X##### (empty) q: 1.0 r: 0.9025 ir: 0.0 f: 0.9025 exp: 399 tga: 117 talp: 15035 tav: 50.2 num: 1 \tq: 1.00 \tr: 0.90 \tir: 0.00 \texp: 399\n", "######X###0##### 1 ######0###X##### (empty) q: 1.0 r: 0.8974 ir: 0.0 f: 0.8971 exp: 137 tga: 26 talp: 14979 tav: 89.2 num: 1 \tq: 1.00 \tr: 0.90 \tir: 0.00 \texp: 137\n", "#########X###0## 1 #########0###X## (empty) q: 1.0 r: 0.8892 ir: 0.0 f: 0.8892 exp: 168 tga: 115 talp: 14994 tav: 93.6 num: 1 \tq: 1.00 \tr: 0.89 \tir: 0.00 \texp: 168\n", "####0#####0###X# 3 ##########X###0# (empty) q: 0.997 r: 0.8828 ir: 0.0 f: 0.8797 exp: 98 tga: 485 talp: 14543 tav: 1.62e+02 num: 1 \tq: 1.00 \tr: 0.88 \tir: 0.00 \texp: 98\n", "##########0###X# 3 ##########X###0# (empty) q: 0.997 r: 0.8828 ir: 0.0 f: 0.8797 exp: 97 tga: 485 talp: 14543 tav: 1.62e+02 num: 1 \tq: 1.00 \tr: 0.88 \tir: 0.00 \texp: 97\n", "########X0###### 2 ########0X###### (empty) q: 1.0 r: 0.8573 ir: 0.0 f: 0.8573 exp: 740 tga: 114 talp: 15034 tav: 28.7 num: 1 \tq: 1.00 \tr: 0.86 \tir: 0.00 \texp: 740\n" ] } ], "source": [ "# exploitation\n", "print_performance(exploit_results_2[0], exploit_results_2[1])" ] }, { "cell_type": "markdown", "metadata": { "papermill": { "duration": 0.021458, "end_time": "2019-10-23T16:02:12.604545", "exception": false, "start_time": "2019-10-23T16:02:12.583087", "status": "completed" }, "tags": [] }, "source": [ "## Comparison" ] }, { "cell_type": "code", "execution_count": 22, "metadata": { "papermill": { "duration": 0.051526, "end_time": "2019-10-23T16:02:12.677018", "exception": false, "start_time": "2019-10-23T16:02:12.625492", "status": "completed" }, "tags": [] }, "outputs": [], "source": [ "def plot_population(metrics, ax=None):\n", " if ax is None:\n", " ax = plt.gca()\n", " \n", " trials = [m['trial'] for m in metrics]\n", " \n", " population_size = [m['numerosity'] for m in metrics]\n", " reliable_size = [m['reliable'] for m in metrics]\n", " \n", " ax.plot(trials, population_size, 'b', label='all')\n", " ax.plot(trials, reliable_size, 'r', label='reliable')\n", " \n", " ax.set_title(\"Population size\")\n", " ax.set_xlabel(\"Trial\")\n", " ax.set_ylabel(\"Number of macroclassifiers\")\n", " ax.legend(loc='best')" ] }, { "cell_type": "code", "execution_count": 23, "metadata": { "papermill": { "duration": 0.258639, "end_time": "2019-10-23T16:02:12.956342", "exception": false, "start_time": "2019-10-23T16:02:12.697703", "status": "completed" }, "tags": [] }, "outputs": [ { "data": { "image/png": "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\n", "text/plain": [ "
" ] }, "metadata": { "needs_background": "light" }, "output_type": "display_data" } ], "source": [ "original = explore_results[1]\n", "modified = explore_results_2[1]\n", "\n", "ax = plt.gca()\n", "\n", "trials = [m['trial'] for m in original]\n", "\n", "original_numerosity = [m['numerosity'] for m in original]\n", "modified_numerosity = [m['numerosity'] for m in modified]\n", "\n", "ax.plot(trials, original_numerosity, 'r')\n", "ax.text(1000, 350, \"Original environment\", color='r')\n", "\n", "ax.plot(trials, modified_numerosity, 'b')\n", "ax.text(1000, 40, 'No-slippery setting', color='b')\n", "\n", "\n", "ax.set_title('Classifier numerosity in FrozenLake environment')\n", "ax.set_xlabel('Trial')\n", "ax.set_ylabel('Number of macroclassifiers')\n", "\n", "plt.show()" ] } ], "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": "3.7.0" }, "papermill": { "duration": 62.784299, "end_time": "2019-10-23T16:02:13.395984", "environment_variables": {}, "exception": null, "input_path": "notebooks/FrozenLake.ipynb", "output_path": "docs/source/notebooks/FrozenLake.ipynb", "parameters": {}, "start_time": "2019-10-23T16:01:10.611685", "version": "1.1.0" } }, "nbformat": 4, "nbformat_minor": 4 }