Book Image

Python Machine Learning Workbook for Beginners

By : AI Sciences
Book Image

Python Machine Learning Workbook for Beginners

By: AI Sciences

Overview of this book

<p>Machine Learning (ML) is the lifeblood of businesses worldwide. ML tools empower organizations to identify profitable opportunities fast and help them to better understand potential risks. The ever-expanding data, cost-effective data storage, and competitively priced powerful processing continue to drive the growth of ML. </p><p> </p><p>This is the best time you could enter the exciting machine learning universe. Industries are reinventing themselves constantly by developing more advanced data analysis models. These models analyze larger and more complex data than ever while delivering instantaneous and more accurate results on enormous scales. </p><p>In this backdrop, it is evident that hands-on practice is everything in machine learning. Tons of theory will amount to nothing if you don’t have enough hands-on practice. Textbooks and online classes mislead you into a false sense of mastery. The easy availability of learning resources tricks you and you become overconfident. But when you try to apply the theoretical concepts you have learned, you realize it’s not that simple. </p><p> </p><p>This is where projects play a crucial role in your learning journey. Projects are doubtless the best investment of your time. You’ll not only enjoy learning but you’ll also make quick progress. And unlike studying boring theoretical concepts, you’ll find that working on projects is easier to stay motivated. </p><p> </p><p>The projects in this book cover ten different interesting topics. Each project will help you refine your ML skills and apply them in the real world. These projects also present you with an opportunity to enrich your portfolio, making it simpler to find a great job, explore interesting career paths, and even negotiate a higher pay package. Overall, this learning-by-doing book will help you accomplish your machine learning career goals faster. </p><p> </p><p>The code bundle for this course is available at https://www.aispublishing.net/ai-sciences-book</p>
Table of Contents (15 chapters)
1
About the Author

5.1. Creating Seq2Seq Training Model for Language Translation

A Seq2seq model typically consists of two models. In the training phase, the encoder receives an input sentence and feeds it to the decoder. The decoder then predicts the output or translated sentence in our case. Both encoder and decoders are connected LSTM networks. The process is shown in the following figure. Here, the offset tag for decoder input is “<s>”, and the offset tag for decoder output is </s>.

image

The input to the encoder is the sentence in the original language, which is English in the above example. The output of the encoder is the hidden and cell states. The input to the decoder is the hidden and cell states from the encoder plus the target dataset, one step offset.

For instance, if you look at the decoder input, in the first step, the input is always <s>. The decoder output at the first timestep is the ground truth translated output word. For instance, the first output word is...