{"id":2747,"date":"2018-08-27T13:39:24","date_gmt":"2018-08-27T13:39:24","guid":{"rendered":"https:\/\/ermlab.com\/?p=2747"},"modified":"2018-09-12T20:48:05","modified_gmt":"2018-09-12T20:48:05","slug":"cifar-10-classification-using-keras-tutorial","status":"publish","type":"post","link":"https:\/\/ermlab.com\/en\/blog\/nlp\/cifar-10-classification-using-keras-tutorial\/","title":{"rendered":"CIFAR-10 classification using Keras Tutorial"},"content":{"rendered":"

The CIFAR-10 dataset consists of 60000 32×32 colour images in 10 classes, with 6000 images per class. There are 50000 training images and 10000 test images.<\/span><\/p>\n

Recognizing photos from the <\/span>cifar-10 collection<\/span><\/a> is one of the most common problems in the today’s world of machine learning. I\u2019m going to show you – step by step – how to build multi-layer artificial neural networks that will recognize images from a cifar-10 \u00a0set with an accuracy of about 80% and visualize it.<\/span><\/p>\n

<\/p>\n

Building 4 and 6-layer Convolutional Neural Networks<\/span><\/h1>\n

To build our CNN (Convolutional Neural Networks) \u00a0we will use Keras<\/strong><\/a> and introduce a few newer techniques for Deep Learning model like activation functions: ReLU<\/a>,<\/strong>\u00a0 dropout<\/strong><\/a>.<\/span><\/p>\n

Keras is an open source neural network Python library which can run on top of other machine learning libraries like TensorFlow, CNTK or Theano. It allows for an easy and fast prototyping, supports convolutional, \u00a0recurrent neural networks and a combination of the two. <\/span><\/p>\n

In the beginning, we will learn what Keras is, deep learning, what we will learn, and briefly about the cifar-10 collection. Then step by step, we will build a 4 and 6 layer neural network along with its visualization, resulting in % accuracy of classification with graphical interpretation.<\/span><\/p>\n

Finally, we will see the results and compare the two networks in terms of the accuracy and speed of training for each epoch.<\/span><\/p>\n

The CIFAR-10 DATASET<\/span><\/h2>\n

The dataset is divided into five training batches and one test batch, each with 10000 images. <\/span><\/p>\n

The test batch contains exactly 1000 randomly-selected images from each class. <\/span><\/p>\n

The training batches contain the remaining images in random order, but some training batches may contain more images from one class than another. <\/span><\/p>\n

Between them, the training batches contain exactly 5000 images from each class.<\/span><\/p>\n

You can download it from <\/span>here<\/span><\/a>.<\/span><\/p>\n

Convolutional Neural Networks – The Code<\/span><\/h1>\n

First of all, we will be defining all of the classes and functions we will need:<\/span><\/p>\n

# Import all modules\r\nimport time\r\nimport matplotlib.pyplot as plt\r\nimport numpy as np\r\nfrom keras.models import Sequential\r\nfrom keras.layers import Dense\r\nfrom keras.layers import Dropout\r\nfrom keras.layers import Flatten\r\nfrom keras.constraints import maxnorm\r\nfrom keras.optimizers import SGD\r\nfrom keras.layers import Activation\r\nfrom keras.layers.convolutional import Conv2D\r\nfrom keras.layers.convolutional import MaxPooling2D\r\nfrom keras.layers.normalization import BatchNormalization\r\nfrom keras.utils import np_utils\r\nfrom keras_sequential_ascii import sequential_model_to_ascii_printout\r\nfrom keras import backend as K\r\nif K.backend()=='tensorflow':\r\n    K.set_image_dim_ordering(\"th\")\r\n\r\n# Import Tensorflow with multiprocessing\r\nimport tensorflow as tf\r\nimport multiprocessing as mp\r\n\r\n# Loading the CIFAR-10 datasets\r\nfrom keras.datasets import cifar10<\/pre>\n
As a good practice suggests, we need to declare our variables:<\/span><\/p>\n