-- Practice Midterm 2 Solutions

((resource:Solution 4.jpeg|Resource Description for Solution 4.jpeg))

-- Practice Midterm 2 Solutions

'''Kunal Deshmukh, Rohan Dalal''' '''Q5 BTFG Algorithm''' ((resource:BFGS.png|Resource Description for BFGS.png))

-- Practice Midterm 2 Solutions

    z = tf.matmul(weights, inputs) + biases
    tf_0 = tf.zeros(shape=z.shape, dtype=tf.float32)

    return tf.maximum(tf_0, z) + alpha*tf.minimum(tf_0, -z)

9) Write a python function leakyReluLayer(weights, inputs, biases, alpha) which will create a layer of TensorFlow Tensor object each compute leaky rectified linear units. Solution: import tensorflow as tf def leakyReluLayer(weights, inputs, biases, alpha): z = tf.matmul(weights, inputs) + biases tf_0 = tf.zeros(shape=z.shape, dtype=tf.float32) return tf.maximum(tf_0, z) + alpha*tf.minimum(tf_0, -z)

-- Practice Midterm 2 Solutions

1. from PIL import Image import numpy as np img = Image.open('bob.png') img = img.convert('L') img_array = np.fromstring(img.tobytes(), dtype = np.uint8) img_arraay = nparray.reshape(img.size[1], img.size[0]) 2. You split the data set into training and test data several times and in several different ways, perform your measurements for each of these trials, and combine these measurements by computing an appropriate aggregation. For exhaustive one, cycle over all possible p-subsets of the data sets, train on the data excluding the p-subset, test on the p-subset. This is called Leave-p-out cross validation. for non-exhaustive one, Randomly repeat m-times, choose a subset of size p as the test data set, train on the remaining data, test on this subset; this is called 'Repeated Random Sub-sampling Validation' 3 Kullback-Leibler, Is a measure of dissimilarity b/w two given prob distributions : E[log(P(x)/Q(x)).

-- Practice Midterm 2 Solutions

       z = weights*nodes  + bias
       return activation(z)

NAme = Krishna Vojjila import tensorflow as tf def perceptron(weights,nodes,bias,activation): z = weights*nodes + bias return activation(z) W = tf.Variable([0,1,0], dtype = float32) X = tf.Placeholder(dtype = float32, shape = (1,3)) b = tf.Variable(0.3, dtype = float32) a = tf.Sigmoid layer1 = perceptron(W,X,b,a)