Subject - Computer Science (DEEP LEARNING)

Please answer by 12:00pm IST Sunday (Jan 24) last date for assignment. Urgent Please.. I will 100% upvote

A. Consider the following layers in a CNN model which has to be implemented in Keras- [2+1+2+2+1+3=11 marks] Layer 0 - Input (224x224) Layer 1 - 3x3 Conv2D, 32 channels, padding=same Layer 2 - 2x2 MaxPooling2D Layer 3 - 5x5 Conv2D, 32 channels, padding=valid Layer 4 - 5x5 Conv2D, 64 channels, padding=same Layer 5 - 2x2 Maxpooling2D Layer 6-3x3 Conv2D, 64 channels, padding=valid, strides=2 Layer 7 Dense(10) What is the output shape of layer O, layer 2, layer 5, layer 7? B.In different layers of a model, what all factors do we consider to decide the total number of kernels? i) Hardware capacity ii) Total number of images in the dataset iii)Inter class variation in the dataset iv)Intra class variation in the dataset C. Consider the following CNN implementation in Keras- Input_shape = (28,28,1) num_classes = 3 model = keras. Sequential [ keras.Input(shape=input_shape), layers. Conv2D(32, kernel_size=(3, 3), activation="relu"), layers. MaxPooling2D(pool_size=(2, 2)), layers. Flatten, layers.Dense(num_classes, activation='softmax"), 1) The values for classes 0, 1 and 2 before softmax activation are 10, 8 and 3. What is the difference between softmax values of class 0 and class 1? D. Assume a Resnet based implementation of an image classification system. The last convolutional layer outputs a 7x7x64 response (64 filters at the last convolution layer) which is passed through the fully connected dense network with 10 softmax output nodes (for 10 classes). What will be the total number of weights to the output layer? If VGG were used with no hidden layer (instead of Resnet), what will be number of weights to the output layer? E.Assume convolution is used to implement a 3x3 mean pooling layer with stride 2. Show the convolution kernel with all values specified. FA convolutional style network to detect boundaries in an image in shown below. The ReLU activation function is applied to the first convolution layer. The output layer uses the linear activation function. In order to make the figure more visible, we use white to denote 0 and darker values to denote larger (more positive) values. output W3 WA feature map 1 feature map 2 Wy (W2 input image 3 i)Design two convolution kernels w1 and w2 for the first layer, of size 3 x 3. One of them should detect dark/light boundaries, and the other should detect light/dark boundaries. ii)Design convolution kernels w3, w4 of size 3x3 for the output layer, which computes the desired output. A. Consider the following layers in a CNN model which has to be implemented in Keras- [2+1+2+2+1+3=11 marks] Layer 0 - Input (224x224) Layer 1 - 3x3 Conv2D, 32 channels, padding=same Layer 2 - 2x2 MaxPooling2D Layer 3 - 5x5 Conv2D, 32 channels, padding=valid Layer 4 - 5x5 Conv2D, 64 channels, padding=same Layer 5 - 2x2 Maxpooling2D Layer 6-3x3 Conv2D, 64 channels, padding=valid, strides=2 Layer 7 Dense(10) What is the output shape of layer O, layer 2, layer 5, layer 7? B.In different layers of a model, what all factors do we consider to decide the total number of kernels? i) Hardware capacity ii) Total number of images in the dataset iii)Inter class variation in the dataset iv)Intra class variation in the dataset C. Consider the following CNN implementation in Keras- Input_shape = (28,28,1) num_classes = 3 model = keras. Sequential [ keras.Input(shape=input_shape), layers. Conv2D(32, kernel_size=(3, 3), activation="relu"), layers. MaxPooling2D(pool_size=(2, 2)), layers. Flatten, layers.Dense(num_classes, activation='softmax"), 1) The values for classes 0, 1 and 2 before softmax activation are 10, 8 and 3. What is the difference between softmax values of class 0 and class 1? D. Assume a Resnet based implementation of an image classification system. The last convolutional layer outputs a 7x7x64 response (64 filters at the last convolution layer) which is passed through the fully connected dense network with 10 softmax output nodes (for 10 classes). What will be the total number of weights to the output layer? If VGG were used with no hidden layer (instead of Resnet), what will be number of weights to the output layer? E.Assume convolution is used to implement a 3x3 mean pooling layer with stride 2. Show the convolution kernel with all values specified. FA convolutional style network to detect boundaries in an image in shown below. The ReLU activation function is applied to the first convolution layer. The output layer uses the linear activation function. In order to make the figure more visible, we use white to denote 0 and darker values to denote larger (more positive) values. output W3 WA feature map 1 feature map 2 Wy (W2 input image 3 i)Design two convolution kernels w1 and w2 for the first layer, of size 3 x 3. One of them should detect dark/light boundaries, and the other should detect light/dark boundaries. ii)Design convolution kernels w3, w4 of size 3x3 for the output layer, which computes the desired output