mattskone

mattskone

#FlipPredictor
#A coin is drawn at random from a bag of coins of varying probabilities
#Each coin has the same chance of being drawn
#Your class FlipPredictor will be initialized with a list of the probability of 
#heads for each coin. This list of probabilities can be accessed as self.coins 
#in the functions you must write. The function update will be called after every 
#flip to enable you to update your estimate of the probability of each coin being 
#the selected coin. The function pheads may be called and any time and will 
#return your best estimate of the next flip landing on heads.

mattskone

"""A simple Markov Decision Process describing a painting robot.

This problem is inspired by the "Planning" lesson from the CS 7637 (Knowledge-
Based Artificial Intelligence) course at Georgia Institue of Technology, with
Prof. Ashok Goyel.  It was created in support of an MDP assignment in CS 7641
(Machine Learning), with Profs. Charles Isbell and Michael Littman.

A robot is tasked with painting a ladder and a ceiling.  The following
independent states are possible, for a total of 8:

mattskone

import time

import numpy as np
from pybrain.datasets import ClassificationDataSet
from pybrain.structure.modules import SigmoidLayer, SoftmaxLayer
from pybrain.supervised.trainers import BackpropTrainer
from pybrain.tools.shortcuts import buildNetwork
from pybrain.utilities import percentError