ExpectationMaximum

enter image description here

2- You may have question marks in your head, especially regarding where the probabilities in the Expectation step come from. Please have a look at the explanations on this maths stack exchange page.

3- Look at/run this code that I wrote in Python that simulates the solution to the coin-toss problem in the EM tutorial paper of item 1:

P.S The code may be suboptimal, but it does the job.

import numpy as np
import math

#### E-M Coin Toss Example as given in the EM tutorial paper by Do and Batzoglou* #### 

def get_mn_log_likelihood(obs,probs):
    """ Return the (log)likelihood of obs, given the probs"""
    # Multinomial Distribution Log PMF
    # ln (pdf)      =             multinomial coeff            *   product of probabilities
    # ln[f(x|n, p)] = [ln(n!) - (ln(x1!)+ln(x2!)+...+ln(xk!))] + [x1*ln(p1)+x2*ln(p2)+...+xk*ln(pk)]     

    multinomial_coeff_denom= 0
    prod_probs = 0
    for x in range(0,len(obs)): # loop through state counts in each observation
        multinomial_coeff_denom = multinomial_coeff_denom + math.log(math.factorial(obs[x]))
        prod_probs = prod_probs + obs[x]*math.log(probs[x])

multinomial_coeff = math.log(math.factorial(sum(obs))) -  multinomial_coeff_denom
likelihood = multinomial_coeff + prod_probs
return likelihood

# 1st:  Coin B, {HTTTHHTHTH}, 5H,5T
# 2nd:  Coin A, {HHHHTHHHHH}, 9H,1T
# 3rd:  Coin A, {HTHHHHHTHH}, 8H,2T
# 4th:  Coin B, {HTHTTTHHTT}, 4H,6T
# 5th:  Coin A, {THHHTHHHTH}, 7H,3T
# so, from MLE: pA(heads) = 0.80 and pB(heads)=0.45

# represent the experiments
head_counts = np.array([5,9,8,4,7])
tail_counts = 10-head_counts
experiments = zip(head_counts,tail_counts)

# initialise the pA(heads) and pB(heads)
pA_heads = np.zeros(100); pA_heads[0] = 0.60
pB_heads = np.zeros(100); pB_heads[0] = 0.50

# E-M begins!
delta = 0.001  
j = 0 # iteration counter
improvement = float('inf')
while (improvement>delta):
    expectation_A = np.zeros((5,2), dtype=float) 
    expectation_B = np.zeros((5,2), dtype=float)
    for i in range(0,len(experiments)):
        e = experiments[i] # i'th experiment
        ll_A = get_mn_log_likelihood(e,np.array([pA_heads[j],1-pA_heads[j]])) # loglikelihood of e given coin A
        ll_B = get_mn_log_likelihood(e,np.array([pB_heads[j],1-pB_heads[j]])) # loglikelihood of e given coin B

        weightA = math.exp(ll_A) / ( math.exp(ll_A) + math.exp(ll_B) ) # corresponding weight of A proportional to likelihood of A 
        weightB = math.exp(ll_B) / ( math.exp(ll_A) + math.exp(ll_B) ) # corresponding weight of B proportional to likelihood of B                            

        expectation_A[i] = np.dot(weightA, e) 
        expectation_B[i] = np.dot(weightB, e)

    pA_heads[j+1] = sum(expectation_A)[0] / sum(sum(expectation_A)); 
    pB_heads[j+1] = sum(expectation_B)[0] / sum(sum(expectation_B)); 

    improvement = max( abs(np.array([pA_heads[j+1],pB_heads[j+1]]) - np.array([pA_heads[j],pB_heads[j]]) ))
    j = j+1

Expectation-Maximization in CSharp

Jump to: navigation, search

This example requires Emgu CV 1.5.0.0

using System.Drawing;
using Emgu.CV.Structure;
using Emgu.CV.ML;
using Emgu.CV.ML.Structure;
 
...
 
int N = 4; //number of clusters
int N1 = (int)Math.Sqrt((double)4);
 
Bgr[] colors = new Bgr[] { 
   new Bgr(0, 0, 255), 
   new Bgr(0, 255, 0),
   new Bgr(0, 255, 255),
   new Bgr(255, 255, 0)};
 
int nSamples = 100;
 
Matrix<float> samples = new Matrix<float>(nSamples, 2);
Matrix<Int32> labels = new Matrix<int>(nSamples, 1);
Image<Bgr, Byte> img = new Image<Bgr,byte>(500, 500);
Matrix<float> sample = new Matrix<float>(1, 2);
 
CvInvoke.cvReshape(samples.Ptr, samples.Ptr, 2, 0);
for (int i = 0; i < N; i++)
{
   Matrix<float> rows = samples.GetRows(i * nSamples / N, (i + 1) * nSamples / N, 1);
   double scale = ((i % N1) + 1.0) / (N1 + 1);
   MCvScalar mean = new MCvScalar(scale * img.Width, scale * img.Height);
   MCvScalar sigma = new MCvScalar(30, 30);
   ulong seed = (ulong)DateTime.Now.Ticks;
   CvInvoke.cvRandArr(ref seed, rows.Ptr, Emgu.CV.CvEnum.RAND_TYPE.CV_RAND_NORMAL, mean, sigma);
}
CvInvoke.cvReshape(samples.Ptr, samples.Ptr, 1, 0);
 
using (EM emModel1 = new EM())
using (EM emModel2 = new EM())
{
   EMParams parameters1 = new EMParams();
   parameters1.Nclusters = N;
   parameters1.CovMatType = Emgu.CV.ML.MlEnum.EM_COVARIAN_MATRIX_TYPE.COV_MAT_DIAGONAL;
   parameters1.StartStep = Emgu.CV.ML.MlEnum.EM_INIT_STEP_TYPE.START_AUTO_STEP;
   parameters1.TermCrit = new MCvTermCriteria(10, 0.01);
   emModel1.Train(samples, null, parameters1, labels);
 
   EMParams parameters2 = new EMParams();
   parameters2.Nclusters = N;
   parameters2.CovMatType = Emgu.CV.ML.MlEnum.EM_COVARIAN_MATRIX_TYPE.COV_MAT_GENERIC;
   parameters2.StartStep = Emgu.CV.ML.MlEnum.EM_INIT_STEP_TYPE.START_E_STEP;
   parameters2.TermCrit = new MCvTermCriteria(100, 1.0e-6);
   parameters2.Means = emModel1.GetMeans();
   parameters2.Covs = emModel1.GetCovariances();
   parameters2.Weights = emModel1.GetWeights();
 
   emModel2.Train(samples, null, parameters2, labels);
 
   #region Classify every image pixel
   for (int i = 0; i < img.Height; i++)
      for (int j = 0; j < img.Width; j++)
      {
         sample.Data[0, 0] = i;
         sample.Data[0, 1] = j;
         int response = (int) emModel2.Predict(sample, null);
 
         Bgr color = colors[response];
 
         img.Draw(
            new CircleF(new PointF(i, j), 1), 
            new Bgr(color.Blue*0.5, color.Green * 0.5, color.Red * 0.5 ), 
            0);
      }
   #endregion 
 
   #region draw the clustered samples
   for (int i = 0; i < nSamples; i++)
   {
      img.Draw(new CircleF(new PointF(samples.Data[i, 0], samples.Data[i, 1]), 1), colors[labels.Data[i, 0]], 0);
   }
   #endregion 
 
   Emgu.CV.UI.ImageViewer.Show(img);
}
原文地址:https://www.cnblogs.com/zeroone/p/3731832.html