Create a Convolutional Neural Network (CNN) in TensorFlow or keras to learn image classification task using CIFAR-10 dataset. CIFAR-10 consists of 60000 32x32 color images in 10 classes, with 6000...

Extracted text: Create a Convolutional Neural Network (CNN) in TensorFlow or keras to learn image classification task using CIFAR-10 dataset. CIFAR-10 consists of 60000 32x32 color images in 10 classes, with 6000 images per class. There are 50000 training images and 10000 test images. You may download and use the CIFAR-10 dataset in keras using the following code: from keras.datasets import cifar10 (x_train, y_train), (x_test, y_test) cifar10.load data () 1) Create a 8-layer CNN as follows: CONV-POOL-CONV-POOL-CONV-POOL-FC-FC. The CONV layers should use 3x3 filters with 32, 64 and 64 channels, stride of 1. The POOL layers should use max 2x2 pooling with stride of 2. Calculate the total number of parameters for this network. Feel free to choose appropriate activation functions for the layers and justify your choice. Plot the results and comment on the CNN performance. 2) Modify your CNN design as: CONV-CONV-POOL-CONV-POOL-FC-FC. Use CONV layers of 5x5 filters (same number of channels as before), stride 1; POOL: 2x2, stride 2. Calculate the total number of parameters in this network. Comment on the difference of number of parameters with respect to the previous network. Plot the results and comment on the CNN performance. 3) Modify your CNN design to achieve better accuracy than the previous two designs. You may change the number of CONV and POOL layers, filter size, number of channels in each CONV layer, strides, dropout etc. Justify your modifications and comment on the accuracy obtained.