utils.py

import numpy as np
from numpy import expand_dims ,ones ,zeros ,asarray 
from numpy.random import randn , randint
import keras
from keras.layers import Reshape, Conv2D,Input ,  Conv2DTranspose , Flatten , Dropout 
from keras.models import Model
from keras.datasets.mnist import load_data

from matplotlib import pyplot

def load_dataset():
  (x_train,y_train),(x_test,y_test) = load_data()
  
  x_train = expand_dims(x_train,axis =-1)
  x_train = (x_train.astype('float32')-127.5)/127.5
  print(x_train.shape)
  print(y_train.shape)
  print(x_test.shape)
  print(y_test.shape)
  #print(x_train[0])
  return [x_train , y_train]
#load_dataset()

def make_supervised_train_dataset(dataset , num_classes =10 ,num_samples =100):
  X,Y = dataset
  x_ = list()
  y_ = list()
  samples_per_class = int(num_samples / num_classes)
  for k in range(num_classes):
    x_class = X[Y == k]
    print(len(X))
    index = randint(0,len(x_class),samples_per_class)
    print(index)
    [x_.append(x_class[p]) for p in index]
    [y_.append(k) for p in index]
  print(len(x_))
  print(len(y_))
  return asarray(x_) , asarray(y_)
#make_supervised_train_dataset(load_dataset())

def gen_real_samples(dataset , num_samples):
  images , labels = dataset 
  index = randint(0, images.shape[0], num_samples)
  img_sample , label_sample = images[index] , labels[index]
  y = ones((num_samples , 1))
  return [img_sample , label_sample] , y

def gen_fake_samples(generator , latent_dim , num_samples):
  z = generate_latent_points(num_samples,latent_dim)
  fake_samples = generator.predict(z)
  y = zeros((num_samples , 1))
  return fake_samples , y

def generate_latent_points(num_samples , latent_dim):
  latent_input = randn(num_samples * latent_dim)
  latent_input = latent_input.reshape(num_samples , latent_dim)
  #print(latent_input)
  return latent_input
#generate_latent_points(5,5)


def one_example():
  img , _= gen_fake_samples(g_model , 100,1)
  ii = tf.convert_to_tensor(
    img, dtype=None, dtype_hint=None, name=None
  )
  plt.figure(figsize=(2,2))
  plt.title(np.argmax(s_model.predict(ii, batch_size =None ,callbacks=None , steps =1)))
  plt.grid = False 
  plt.xticks([])
  plt.yticks([])
  img = np.reshape(img ,(28,28))
  plt.imshow(img)

def graph_plot():
  number_steps =epochs_*600
  plt.figure(figsize=(10,10))
  plt.plot(np.arange(0,number_steps),plot_sup_loss)
  plt.plot(np.arange(0,number_steps),plot_unsup_loss)
  plt.plot(np.arange(0,number_steps),plot_gan_loss)
  plt.title('loss_function')
  plt.ylabel('loss')
  plt.xlabel('epoch')
  plt.legend(['sup_loss', 'unsup_loss','gan_loss'], loc='upper left')
  plt.show()
  # summarize history for loss
  plt.figure(figsize=(10,10))
  plt.plot(np.arange(0,number_steps),plot_sup_acc)
  plt.title('acc_function')
  plt.ylabel('acc')
  plt.xlabel('epoch')
  plt.legend(['train acc_sup'], loc='upper left')
  plt.show()
  # summarize history for train_acc
  plt.figure(figsize=(10,10))
  plt.plot(np.arange(0,number_steps/600),plot_test_acc)
  plt.title('acc_function')
  plt.ylabel('test_acc')
  plt.xlabel('epoch')
  plt.legend(['test_acc'], loc='upper left')
  plt.show()
  # summarize history for test_acc