Summary
In this chapter, we explained what DL is and how it’s related to DNNs. We discussed the different types of DNNs and how to train them, and we paid special attention to various regularization techniques that help with the training process. We also mentioned many real-world applications of DL and tried to analyze the reasons for its efficiency. Finally, we introduced two of the most popular DL libraries, namely PyTorch and Keras. We also implemented identical MNIST classification examples with both libraries.
In the next chapter, we’ll discuss how to solve classification tasks over more complex image datasets with the help of convolutional networks – one of the most popular and effective deep network models. We’ll talk about their structure, building blocks, and what makes them uniquely suited to computer vision tasks. To spark your interest, let’s recall that convolutional networks have consistently won the popular ImageNet challenge since...
