Anyone using deep learning knows that it is not just one algorithm -- it's many different architectures, each one with its strengths and weaknesses.
Getting to know the pros and cons of different neural networks can take time, and our goal is to get you up to speed as quickly as possible. Neural network selection and training is an iterative process. Mistakes are inevitable, so don't sweat them too hard. You will develop intuitions based on repeated empirical experiments.
In this book, we'll go over widely-used architectures that you will encounter in the wild, and perhaps consider for your use case. Multilayer Perceptrons (MLPs) Recurrent Neural Networks (focus: LSTMs) Convolutional Neural Networks (CNNs) Autoencoders Word2Vec (a shallow network, but useful). By the end of this book, we hope you build a foundation for selecting the correct network for your project.