Book Image

Python Machine Learning By Example - Third Edition

By : Yuxi (Hayden) Liu
Book Image

Python Machine Learning By Example - Third Edition

By: Yuxi (Hayden) Liu

Overview of this book

Python Machine Learning By Example, Third Edition serves as a comprehensive gateway into the world of machine learning (ML). With six new chapters, on topics including movie recommendation engine development with Naïve Bayes, recognizing faces with support vector machine, predicting stock prices with artificial neural networks, categorizing images of clothing with convolutional neural networks, predicting with sequences using recurring neural networks, and leveraging reinforcement learning for making decisions, the book has been considerably updated for the latest enterprise requirements. At the same time, this book provides actionable insights on the key fundamentals of ML with Python programming. Hayden applies his expertise to demonstrate implementations of algorithms in Python, both from scratch and with libraries. Each chapter walks through an industry-adopted application. With the help of realistic examples, you will gain an understanding of the mechanics of ML techniques in areas such as exploratory data analysis, feature engineering, classification, regression, clustering, and NLP. By the end of this ML Python book, you will have gained a broad picture of the ML ecosystem and will be well-versed in the best practices of applying ML techniques to solve problems.
Table of Contents (17 chapters)
Other Books You May Enjoy

Overcoming long-term dependencies with Long Short-Term Memory

Let's start with the vanishing gradient issue in vanilla RNNs. Where does it come from? Recall that during backpropagation, the gradient decays along with each time step in the RNN (that is, ); early elements in a long input sequence will have little contribution to the computation of the current gradient. This means that vanilla RNNs can only capture the temporal dependencies within a short time window. However, dependencies between time steps that are far away are sometimes critical signals to the prediction. RNN variants, including Long Short-Term Memory (LSTM) and Gated Recurrent Unit (GRU), are specifically designed to solve problems that require learning long-term dependencies.

We will be focusing on LSTM in this book as it is a lot more popular than GRU. LSTM was introduced a decade earlier and is more mature than GRU. If you are interested in learning more about GRU and its applications...