updated notes and implemented data generation for 10 digit recognition

SecondSunday/Notes.md

@@ -0,0 +1,22 @@
+# Activation Function (AKA Transfer function)
+ In a neural network activation function adds non-linearity to it.
+ Types:
+ 1. Sigmoid(Logistic) (used mostly for output layer(looks like probability))
+ 2. RelU or Rectified Linear Unit (important discovery for NN - most-used for hidden layers)(not suitable for output layer if output is supposed to be probability) and leaky RelU with some slope on negative part
+ 3. tanH (Hyperbolic) (-1 - 1) or ArcTan (Tan Inverse -> maps to -Pi/2 - Pi/2)
+ 4. Linear(or Identity) layer (used for output layers(best for regression))
+ 5. Softmax (classification giving probability) (probability coz outputs add upto 1)
+ 6. SquareRoot
+ 7. Exponential
+ 8. Sine.
+ 9. Ramp
+ 10. Step (Binary)
+ 11. Unit Sum
+
+if the network computation is something that is multiplicative, use log as activation so that the sum becomes addition.
+
+Constraint Optimization: optimize in such a way that the output is constrained to some value.
+
+
+Steps => number of iteration of batches
+Epoch => number of iterations of going throught the entire dataset

SecondSunday/mnist_phone_number.py

@@ -0,0 +1,46 @@
+%matplotlib inline
+from tensorflow.examples.tutorials.mnist import input_data
+import numpy as np
+import matplotlib.pyplot as plt
+
+mnist = input_data.read_data_sets('./SecondSunday/mnist_data', one_hot=True)
+label_number = mnist.train.labels.argmax(axis=1)
+number_imgs = {str(i):mnist.train.images[np.argwhere(label_number == i).squeeze()] for i in range(10)}
+DATA_COUNT = 100
+phone_number_digits = np.random.randint(10**9,10**10,size=(DATA_COUNT,10))
+phone_number_digits = np.random.randint(10,size=(DATA_COUNT,10))
+phone_number_digits.astype(str)
+phone_number_strings = phone_number_digits.astype(str)
+
+def pick_img(num):
+    rand_idx = np.random.randint(number_imgs[num].shape[0])
+    img = number_imgs[num][rand_idx].reshape(28,28)
+    return img
+
+def create_phone_img(phon_no):
+    return np.hstack(tuple([pick_img(d) for d in phon_no]))
+
+def create_phone_images(phone_array):
+    phone_number_images = []
+    for phon_no in phone_array:
+        phone_number_images.append(create_phone_img(phon_no))
+    return np.array(phone_number_images).reshape(-1,28*280)
+
+phone_number_imgs = create_phone_images(phone_number_strings)
+
+from keras.models import Sequential
+from keras.layers import Dense, Activation
+
+model = Sequential([
+    Dense(32, input_shape=(7840,)),
+    Activation('relu'),
+    Dense(10),
+    Activation('linear'),
+])
+
+model.compile(optimizer='rmsprop',
+              loss='categorical_crossentropy',
+              metrics=['accuracy'])
+
+model.fit()
+# plt.imshow(phone_number_imgs[np.random.randint(phone_number_imgs.shape[0])])