update

Author: bob7783

unsupervised_class3/bayes_classifier_gaussian.py

@@ -11,6 +11,12 @@
 from scipy.stats import multivariate_normal as mvn
 
 
+def clamp_sample(x):
+  x = np.minimum(x, 1)
+  x = np.maximum(x, 0)
+  return x
+
+
 class BayesClassifier:
   def fit(self, X, Y):
     # assume classes are numbered 0...K-1
@@ -30,11 +36,11 @@ def fit(self, X, Y):
 
   def sample_given_y(self, y):
     g = self.gaussians[y]
-    return mvn.rvs(mean=g['m'], cov=g['c'])
+    return clamp_sample( mvn.rvs(mean=g['m'], cov=g['c']) )
 
   def sample(self):
     y = np.random.choice(self.K, p=self.p_y)
-    return self.sample_given_y(y)
+    return clamp_sample( self.sample_given_y(y) )
 
 
 if __name__ == '__main__':

unsupervised_class3/bayes_classifier_gmm.py

@@ -11,6 +11,12 @@
 from sklearn.mixture import BayesianGaussianMixture
 
 
+def clamp_sample(x):
+  x = np.minimum(x, 1)
+  x = np.maximum(x, 0)
+  return x
+
+
 class BayesClassifier:
   def fit(self, X, Y):
     # assume classes are numbered 0...K-1
@@ -39,11 +45,11 @@ def sample_given_y(self, y):
     # we cheat by looking at "non-public" params in
     # the sklearn source code
     mean = gmm.means_[sample[1]]
-    return sample[0].reshape(28, 28), mean.reshape(28, 28)
+    return clamp_sample( sample[0].reshape(28, 28) ), mean.reshape(28, 28)
 
   def sample(self):
     y = np.random.choice(self.K, p=self.p_y)
-    return self.sample_given_y(y)
+    return clamp_sample( self.sample_given_y(y) )
 
 
 if __name__ == '__main__':

unsupervised_class3/util.py

@@ -14,6 +14,7 @@
 from scipy.misc import imread, imsave, imresize
 from glob import glob
 from tqdm import tqdm
+from sklearn.utils import shuffle
 
 
 def get_mnist(limit=None):
@@ -26,10 +27,11 @@ def get_mnist(limit=None):
 
   print("Reading in and transforming data...")
   df = pd.read_csv('../large_files/train.csv')
-  data = df.as_matrix()
-  np.random.shuffle(data)
+  data = df.values
+  # np.random.shuffle(data)
   X = data[:, 1:] / 255.0 # data is from 0..255
   Y = data[:, 0]
+  X, Y = shuffle(X, Y)
   if limit is not None:
     X, Y = X[:limit], Y[:limit]
   return X, Y