Stop wasting your time
wrestling with incomplete and confusing tutorials
Why is it hard to learn deep learning?
There's just so much to know before I can even get started!
Neural networks, convolution, recurrent neural nets?!?
What the? How do I even know where to start?
It's hard to find any quality or complete blogs on building deep learning for production
The Internet is full of incomplete blogs and poorly written code snippets?
Seems like everyone is using it
while I don't know where to start and I don't want to get left behind!
Googling only takes you so far...
There is seemingly no coherent resource for gluing all the pieces together!
Time is money
don't waste it sifting through blogs and unhelpful academic tutorials
What the heck is a _____?
The vocabulary is foreign, what is a perceptron? A matrix transposition? a hidden layer??!
How about using it in our application?
How do we integrate deep learning into our applications?
How does it all fit together?
and what do I do with a deep learning model?
Not hitting deadlines?
I still have a job to do and stopping to learn deep learning will waste a lot of time!
Detailed Examples
The deep learning libraries are extensive.
The vocabulary, the syntax, the algorithms, are all incredibly complex and wide-ranging. What is the right package to use, the right library to import?
Imports
importing the required
keraspackagesModel
We're creating the deep learning model
Input layer
The first layer comes directly from the input
Second layer
The second layer sets up the first pattern recognition
Max Pooling
Max pooling drops some of the data to allow us to shrink the size of the dataset
Output
The model's last layer provides a single prediction using the sigmoid activation
Final step
Let's put it all together and test our assumptions
1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21from keras.models import Sequential from keras.layers import Dense, Flatten from keras.layers.normalization import BatchNormalization from keras.layers.convolutional import Conv2D from keras.layers.pooling import MaxPooling2D model = Sequential() model.add(Conv2D(32, (3, 3), input_shape = (64, 64, 3), activation = 'relu')) model.add(MaxPooling2D(pool_size = (2, 2))) model.add(BatchNormalization()) model.add(Conv2D(64, (3, 3), activation = 'relu')) model.add(MaxPooling2D(pool_size = (2, 2))) model.add(BatchNormalization()) model.add(Flatten()) model.add(Dense(128, activation = 'relu')) model.add(Dense(1, activation = 'sigmoid')) model.fit(Y)
Ready to add deep learning to your toolbelt?
What if you knew exactly how all the pieces of a deep learning system work together and had a solid understanding of the mathematics where it runs - in less time, without banging your head against the wall. Imagine how quickly you could work if you knew the best practices and the right tools?
Stop wasting your time searching and have everything you need to be productive in one, well-organized place, with complete examples to get your project running intelligently without needing to resort to endless hours of research.
You will learn the right way to integrate deep learning with the latest tools with your current applications and even build some new ones along the way.
Download the first chapterInteractive code
Jupyter notebooks included
Every chapter in the book comes with a complete interactive Jupyter notebook that uses the concepts in the chapter.
Data walk-through
Zero to Deep Learning starts out with a deep introduction into data manipulation. We aim to support both the season-professional and the complete beginner.
We explore sample datasets statistics, work with pivot tables, and visualizing data patterns.
Gentle introduction
Zero to Deep Learning is specifically crafted to make deep learning accessible to web developers of all experience levels.
Wide-ranging topics
Zero to Deep Learning covers a wide-ranging number of topics, from image recognition through text detection. It covers production-level model serving to make it easy to apply deep learning to web applications today.
Gradient Descent
Zero to Deep Learning gently introduces deep learning topics with introductory topics, such as Gradient Descent before diving too far deeply into the deep-end.
Convolutional Neural Networks
A course on Deep Learning would be incomplete without a course on convolutional neural networks, the quitessential example of the power of deep learning.
We work with powerful image recognition systems using convolutional neural networks, from the basics through the end-to-end system.
Free 7-day course on learning deep learning!
We've written a professional's guide to getting up to speed with Deep Learning for the web developer. As primarily web-developers, we've taken the plunge to learn how to implement and use deep learning in our web applications and we'd love to share our findings with you for free!
In this 7-day course, you'll find out what exactly Deep Learning is, through implementing your own including writing a neural network from scratch.
Book Contents
Zero to Deep Learning is painstakingly designed to teach you step-by-step how to create serious, full-blown deep learning algorithms: from empty-folder to deployment. Each chapter covers a topic and we provide full code examples for every project in the book.
- Development environment1
- Data manipulation2
- Machine Learning3
- Deep Learning4
- Deep Learning internals5
- Convolutional neural networks6
- Recurrent Neural Networks7
- Natural Language Processing & Text8
- Scaling out9
- Training with GPUs10
- Improving performance11
- Pre-trained models12
- Serving predictions13
- Building a native mobile recognition app14
Get up and running quickly
Within the first few minutes, we'll know enough deep learning to start seeing the benefits of using it in our applications.
Interactive
Every single chapter and line of code includes an interactive Jupyter notebook. You'll get access to a Jupyter notebook for all code samples.
Best practices
Learn the best practices, such as: handling overfitting, code organization, and how to serve our model to our apps. We'll walk through practical, common examples of how to implement complete applications powered by deep learning.
Comprehensive topics
You'll learn core deep learning concepts - from the multiperceptron through deep neural networks including convolutional and recurrent neural networks.
App
What you'll build
When you get Zero to Deep Learning, you're not buying just a book, but an interactive course with hundreds of code examples.
Intelligent mobile app
- From start to finish, we build a React Native application.
- We build a React Native application that can recognoize real objects right on our phone
- On the left we have a mobile app that can recognize objects such as this television remote
Deep-learning API
- A custom-designed deep-learning network
- RESTful API backend ready to add intelligence
- A powerful development environment
Interactive code
- Interactive Jupyter notebooks
- Visually-driven code, explained step-by-step
- Multiple exercises with every chapter
200+ Pages (and growing)
A ton of content is included covering the very basics through the latest technical implementations of deep neural networks.
500+ lines of code (and growing)
Using real-world examples, you'll have a TON of code to use to learn from.
Interactive Jupyter notebooks
The book includes runnable code examples for building all the code in the book.
Too good to be true?
Grab a sample chapter and check it out for yourself. Sign up for our mailing list and get the sample chapters for free! You'll only receive email about the book and updates. We never send spam, ever and it's easy to unsubscribe.
It can take up to an hour to deliver the sample chapter. If you don't receive the sample chapter within the hour, write us and we'll send them to you directly.
Production-quality
Built on Tensorflow and Keras
Learn practical Tensorflow applications with Keras – without getting lost in equations.
With Zero to Deep Learning, you'll be armed with production-quality knowledge to take your new deep learning skills to professional products.
Tensorflow
An open-source software library for Machine Intelligence built and maintained by the brilliant engineers at Google.
Convolutional Neural Networks
Keras is a high-level neural networks API, written in Python and capable of running on top of TensorFlow, CNTK, or Theano.
Packages
Get a package now!
The book and complete source
- Professional-grade Deep Learning ebook content
- Completely DRM-free PDF, mobi, and epub formatted ebook
- Learn deep neural networks from an implementer's point of view
- Learn the right way to add intelligence to your applications
- Build advanced apps within minutes
- Have the complete library available at your fingertips
- The code for every sample in the book to get you going quickly
- Community Chat Room Support
- -
Book, source code, videos, and sample app
- Professional-grade Deep Learning ebook content
- Completely DRM-free PDF, mobi, and epub formatted ebook
- Learn deep neural networks from an implementer's point of view
- Learn the right way to add intelligence to your applications
- Build advanced apps within minutes
- Have the complete library available at your fingertips
- The code for every sample in the book to get you going quickly
- Community Chat Room Support
- Weekly Email Support
- The complete code for the sample app
- A screencast detailing building a full deep learning system from scratch
- -
Team License
- Professional-grade Deep Learning ebook content
- Completely DRM-free PDF, mobi, and epub formatted ebook
- Learn deep neural networks from an implementer's point of view
- Learn the right way to add intelligence to your applications
- Build advanced apps within minutes
- Have the complete library available at your fingertips
- The code for every sample in the book to get you going quickly
- Community Chat Room Support
- Weekly Email Support
- The complete code for the sample app
- 4 hour long beginner screencast
- Email technical support
- Team license for up to 10 team members
- Immediate invoice billing service
- Access to older versions of the book
- Access to full Git repository of book and code
- -
The Team
Meet the authors
Francesco Mosconi
Francesco Mosconi is a Data Science consultant and trainer. With Catalit LLC. He is also founder of Data Weekends, a series of 2-day immersive workshops for people that want to approach machine learning, deep learning and big data analytics. Francesco was Chief Data Officer and co-founder at Spire, a YCombinator-backed startup that invented the first consumer wearable device capable of continuously tracking respiration and activity.
Ari Lerner
Hi, I'm Ari. I've been teaching Web Development for a long time. I like to speak at conferences and eat spicy food. I technically got paid while I traveled the country as a professional comedian, but have come to terms with the fact that I am not funny.
Nate Murray
Nate is a full-stack developer and writes code for everything from deep-learning image recognition to mobile games for cats. Nate formerly worked at IFTTT and his background is in data mining and scaling web services.
FAQs
Questions? We have answers!
How long is the book?
The final version has 10 chapters totaling 640+ pages, several sample apps totaling over 7,500+ lines of code (Python, non-comment lines)
Do I have to know or be good at math?
Nope! We don't assume that you're a math wizard. Instead, we take the approach of adding what you need to know when you need to know it.
Are there free updates?
Yes! Updates are free for 12-months following purchase. We've faithfully released over 50 updates to ng-book already
What about Machine Learning?
This book focuses specifically on Deep Learning. We will be releasing a guide on machine learning in the future.
Does the video have captions?
Yes! The screencast video is has a complete caption track so you can read along as you watch the video.
Is this a physical or digital book?
This is a completely DRM-free ebook formatted as a pdf/mobi/epub (and a zip with tons of example code)
Is there a physical print version of the book?
Not yet, but we are working on it
Our Promise to You
We're committed to keeping Zero to Deep Learning as the best resource for learning and implementing deep learning into our applications. We personally respond to requests for content and we regularly release updates. We're independent authors and we survive by making the highest quality book on deep learning as possible.
There's no risk: if you're not satisfied for any reason, send us an email and we'll give you a full refund.
Contact Us
If you have any concerns, feel free to email us