Listen free for 30 days

  • Deep Learning with Keras

  • Introduction to Deep Learning with Keras
  • By: Anthony Williams
  • Narrated by: William Bahl
  • Length: 2 hrs and 30 mins
  • 4.8 out of 5 stars (30 ratings)

£7.99/month after 30 days. Cancel anytime

Summary

This book will introduce you to various deep learning models in Keras, and you will see how different neural networks can be used in real-world examples, as well as in various scientific fields. You will explore various Keras algorithms like the simplest linear regression or more complex deep convolutional networks. You will get to know what the difference between supervised and unsupervised deep learning is, and you will be able to implement various algorithms in Keras by yourself as you follow this step-by-step guide in this book.

You will explore various applications of deep learning models such as speech recognition systems, natural language processing and video game development. A whole new world will open in front of you. By the time you reach the final minute of this book, you will be a Keras expert and ready for your deep-learning projects.

There is so much to learn in this book about deep learning with Keras, and I do invite you to grab your copy today and get started!

By listening to this book you will discover:

  • Deep neural network
  • Neural network elements
  • Keras models
  • Sequential model
  • Functional API model
  • Keras layers
  • Core Keras layers
  • Convolutional Keras layers
  • Recurrent Keras layers
  • Deep learning algorithms
  • Supervised learning algorithms
  • Applications of deep learning models
  • Automatic speech and image recognition
  • Natural language processing
  • Video game development
  • Real world applications
  • And of course, much more!

Get this book today, and learn more about deep learning with Keras!

©2017 Anthony Williams (P)2017 Anthony Williams

What listeners say about Deep Learning with Keras

Average customer ratings
Overall
  • 5 out of 5 stars
  • 5 Stars
    27
  • 4 Stars
    2
  • 3 Stars
    0
  • 2 Stars
    0
  • 1 Stars
    1
Performance
  • 5 out of 5 stars
  • 5 Stars
    27
  • 4 Stars
    2
  • 3 Stars
    0
  • 2 Stars
    1
  • 1 Stars
    0
Story
  • 5 out of 5 stars
  • 5 Stars
    27
  • 4 Stars
    2
  • 3 Stars
    0
  • 2 Stars
    0
  • 1 Stars
    1

Reviews - Please select the tabs below to change the source of reviews.

Sort by:
Filter by:
  • Overall
    5 out of 5 stars
  • Performance
    5 out of 5 stars
  • Story
    5 out of 5 stars

All of the DLer should owe one

Simplifies the maths, good at explaining the various deep learning models in Keras. The course is very carefully designed. It has covered all the needed topics for Deep Learning from a practical perspective. Altogether it was a great experience. Hopefully, I could take this wonderful learning experience for the betterment of my research and definitely to the society where I live in. I appreciate Mr. Anthony Williams and team for their incredible and genuine efforts. Another major point I want to highlight is that I was exposed to lot of useful websites and useful resources during the course. The exercises given made sure it covered all the aspects taught during the session. Simple explanations, simplifies the maths, good at explaining the python programming. Once again, Thanks a lot.

24 people found this helpful

  • Overall
    5 out of 5 stars
  • Performance
    5 out of 5 stars
  • Story
    5 out of 5 stars

Great treatise on the mathematical tenets

Deep learning, which is called as hierarchical or deep structured learning, is the application for learning chores of neural networks which are artificial denoted as ANNs. Artificial neural networks contain multiple hidden layers. Being a part of the greater family of machine learning techniques, deep learning is based on learning data representations in opposition to task particular algorithms. Deep learning can be unsupervised, partially supervised and supervised. Deep learning models like recurrent networks and deep neural networks are applied to broad range of fields including, natural language processing, bioinformatics, machine translation, computer vision, etc. Deep learning models are classified as learning algorithms machines. The methods of deep learning include usage of multiple layers of nonlinear processing components in order to extract and transform their features. Every successive layer is based on the output of the previous layers which is used as input. Other deep learning models which are unsupervised are based on learning of numerous levels of representations and features of some image or text data. To form a hierarchical formation, features from higher levels are developed from the lower level features. Also, deep learning models can present the correspondence between numerous levels of features and different abstraction levels which are forming hierarchical concept.

17 people found this helpful

  • Overall
    5 out of 5 stars
  • Performance
    5 out of 5 stars
  • Story
    5 out of 5 stars

Foundations of machine learning and deep learning.

Kudos are well deserved for the author demystifying some of the biggest and least understood buzzwords out there today: How to implement various algorithms in Keras.... Finally a straightforward discussion of these advanced concepts wrapped in an easily-accessible and understandable fashion. If you're curious about the topics and nervous about buying an in-depth tome in which you will feel lost in jargon and technical mumbo-jumbo, buy this one. The author is clearly an expert in the topic area and does a fantastic job of explaining it to mortals. Pick this one up, you'll be glad you did.

6 people found this helpful

  • Overall
    5 out of 5 stars
  • Performance
    5 out of 5 stars
  • Story
    5 out of 5 stars

Types of Deep Neural

Networks There are various types of deep neural networks including Hopfield network, deep belief network, generative adversarial network, autoencoder, radial basis network,etc. However, in this book, we are going to see how to create different types of neural networks with Keras such as recurrent neural network, deep feed-forward network, deep convolutional networks and de-convolutional network . But firstly, some details regarding these types of deep networks should be discussed, before going further into the subject. Feed-forward neural network is where connections among different components do not form a cycle model, which would be the main difference between feed-forward and recurrent neural networks. The feed-forward network was, in fact, the initial and the simplest neural network model. The information contained in this network are being transferred only in one linear direction, forward, from the initial input nodes to the hidden layers and their nodes. As mentioned before, recurrent neural networks form connection among components in the form of cycle. This model enables a network to expose dynamic temporal behavior. As for a convolutional neural network, it is a type of feed-forward network which has already proven value while analyzing visual imagery.

5 people found this helpful

  • Overall
    5 out of 5 stars
  • Performance
    5 out of 5 stars
  • Story
    5 out of 5 stars

Very satisfied with the course

The instructor has lots of data science experience, so he gives us useful and real-world examples throughout the course. Those examples can be implemented in future projects with slight changes, I suppose. The course was easier, more straight forward, and more interesting than what I was expecting a data science course to be. Anthony Williams, thank you very much for your well-organized course, clear concept explanations, and useful examples.

2 people found this helpful

  • Overall
    5 out of 5 stars
  • Performance
    5 out of 5 stars
  • Story
    5 out of 5 stars

Flagship book for Deep Learning

Distributed representations of layers are observed as all data which is generated by the interaction between layered factors. Deep learning is based on the assumption that each layer of factors is in a correspondence to the other levels of composition and abstraction. The number of created levels of abstraction depends on layers size and number. The hierarchical concept of deep learning refers to that abstract concept and higher level components which are derived from the lower level ones. Deep neural networks may include few fundamental approaches. Each deep learning model can be successful in certain domains, but the comparison between models isn't always possible unless the models have been estimated on the same data collection. Deep neural networks are feedforward meaning that data is transferred from an input to output once, without looping back. On the other hand, the reccurrent neural network makes it possible for data to move in any direction which is commonly applied to language modeling. About certain interpretations and applications of deep neural networks will be the word in the later sections of the book.

1 person found this helpful

  • Overall
    5 out of 5 stars
  • Performance
    5 out of 5 stars
  • Story
    5 out of 5 stars

Solid / practical guide for all.

This book is geared toward beginners who are just beginning to explore the power of such a library and its significant influence on network development. So, after you are done with reading, you will be able to create your first neural network model using Keras, you will get familiar with the practical usage of Keras, and its implementation and you will see what are some of the realworld examples of Keras and what is its overall influence in deep learning studies. Keras is recommended library for deep learning in Python since it is minimalistic, it is capable of running on top of Theano, Deep learning, Tensorflow, and CNTK. Keras is carefully designed to make fast experimentation possible with deep neural networks. The focus of the Keras design is to keep it modular and extensible.Developed as a part of ONEIROS, Keras's primary author is François Chollet who is a Google engineer. It was introduced in 2015, and its author later explained that Keras was developed to be an interface rather than machine learning framework. Keras, in fact, presents a higher-level intuitive set of abstractions which make it easy to configure neural networks.

1 person found this helpful

  • Overall
    5 out of 5 stars
  • Performance
    5 out of 5 stars
  • Story
    5 out of 5 stars

Almost everything.

This course provides a comprehensive overview on the emerging technologies - Deep neural network Neural network elements, Keras models, Sequential model, Functional API model, Keras layers, Core Keras layers, Convolutional Keras layers, Recurrent Keras layers, Deep learning algorithms, Supervised learning algorithms, Applications of deep learning models, Automatic speech and image recognition, Natural language processing, Video game development and Real world applications, which was very impressive to me.

  • Overall
    5 out of 5 stars
  • Performance
    5 out of 5 stars
  • Story
    5 out of 5 stars

DEEP LEARNING, is OUTSTANDING

This is a vast topic. This course is more like an overview, giving an idea of where to begin and how to proceed and at the end you will be a Keras expert and ready for your deep-learning projects. Great help. Thanks.

  • Overall
    5 out of 5 stars
  • Performance
    5 out of 5 stars
  • Story
    5 out of 5 stars

Great way to get started with Deep Learning

The course was excellent structured. I am experienced electrical engineer with little knowledge of Python but I managed to understand all techniques and algorithms presented.

Sort by:
Filter by:
  • Overall
    5 out of 5 stars
  • Performance
    5 out of 5 stars
  • Story
    5 out of 5 stars
Profile Image for Steven
  • Steven
  • 25-10-20

Approachable and motivating

It covers what can be very complex topics, but is written in a way that makes them very easy to understand. If you’re looking for an introduction to AI and Machine Learning, I can’t think of a better resource. For business folks evaluating whether these technologies may benefit their organization, this provides a great framework. For students entering the business and/or tech world, this would be a great starting point. Though it is not a technical reference, I think it gives developers and IT folks a good understanding of how others are evaluating their work. Even if you’re not in one of the categories above, you’ve likely seen ton of news articles discussing these technologies. This book would be great for you as well. I’ve read a lot of books and articles on these topics, but there’s nothing I can point to that provides a better overview.

24 people found this helpful

  • Overall
    5 out of 5 stars
  • Performance
    5 out of 5 stars
  • Story
    5 out of 5 stars
Profile Image for KELLY
  • KELLY
  • 26-10-20

Definitive

The course was excellent structured. I am experienced electrical engineer with little knowledge of Python but I managed to understand all techniques and algorithms presented. Very thorough overview of many different pieces of the Data Science landscape. A few issues getting the technologies working together since packages change over time, so I'd recommend the authors clean that up a bit. Beyond that, well worth my time! with no practical experience, and no python experience, really I could understand ML. The concepts are super-clear and the examples are also excellent .

15 people found this helpful

  • Overall
    5 out of 5 stars
  • Performance
    5 out of 5 stars
  • Story
    5 out of 5 stars
Profile Image for Billy Robinson
  • Billy Robinson
  • 28-10-20

Intellectual and Resourceful

By this course I got good fundamentals and the way teaching done by the tutor is fantastic. This course create a lot of interest and inputs for those want to get in to Real world applications and Video game development. Concepts were explained very clearly. Exercises on each topic was so well thought through and it really helped to reinforce the concepts and validate the understanding. He makes it look easy, simplifying complex things and motivating enough to keep you going. All I can say is thank you. I am just at 30 mins but I am already searching for all your other courses While some of the concepts are taking longer to digest, the teacher is making what can be a steep learning curve less arduous. Overall course was amazing. Anthony covered all topics which were necessary and also the practical experience is what I really enjoyed.

14 people found this helpful

  • Overall
    5 out of 5 stars
  • Performance
    5 out of 5 stars
  • Story
    5 out of 5 stars
Profile Image for AMERICAN
  • AMERICAN
  • 29-10-20

Everything is fine

Yes it's a good match. I have strong programming experience, and still need to go back to review the material more than once. Python programming class is a good pre-requisite for this class. This is a very good course and equip me with sound basic. I feel that I can use what I learn in this class to do real works. Leonard is excellent at explaining the machine learning concepts and walking through their implementation in Python. I especially like his ability to distill fancy concepts into simple terms throughout the course, because at the end of the day these "fancy" ML models should produce accurate, intuitive results given good input data. Do the exercises (whether while taking the course or walk through them later) because this is really how to hammer home the concepts and Python code. In general, the best way to learn the programming side of any technical implementation in my experience is by doing it. Very comprehensive course about many machine learning techniques and easy to follow exercises. Very good for a beginner.

13 people found this helpful

  • Overall
    5 out of 5 stars
  • Performance
    5 out of 5 stars
  • Story
    5 out of 5 stars
Profile Image for James
  • James
  • 01-11-20

Neural Network Elements

Neural networks are a collection of algorithms created to recognize certain patterns. We are already familiar with these networks and how they transform data through created model of perception, clustering, and labeling raw input. The patterns which are recognized are numerically denoted as vectors, and all real-world data including text, images, the sound must be translated into those vectors. Keep in mind that neural networks are classifying and clustering different data. Think of them as a certain layer which you put on top of your data that you managed and stored. The neural network works for you as a help in order to group and classify different data in a correspondence to similarities between example inputs. Think of deep neural networks as a part of the greater machine-learning application for regression, reinforcement learning, and classification. Deep learning is used for networks which are composed of multiple layers. Each layer is made of a node which is a place where computations occur. These nodes from every layer combine input from a data with a collection of coefficients, or certain weights, so the significance is assigned to every input for each task the algorithm is processing or learning.

9 people found this helpful

  • Overall
    5 out of 5 stars
  • Performance
    5 out of 5 stars
  • Story
    5 out of 5 stars
Profile Image for Ronnie
  • Ronnie
  • 08-11-20

Broad, Mathematical Survey

Since it is independent of backend engines, Keras manages all its graphs independently, so it is possible to define any Keras model with the Theano engine backend. It is even possible to switch to the TensorFlow backend or to reapply your Keras model which is Theano built on a TensorFlow input which creates a TF model version. This TF model is what was a Theano model initially created. Another important feature of Keras is an object-oriented design. Therefore, in Keras everything acts like an object including models, optimizers, and layers. All parameters regarding any object can be accessed. A different approach would be as in other deep learning models which are defined as chains of functions since most of these functions are heavily parameterized by the weight of their tensors. Since the manipulating of these parameters would be impractical in a functional way, Keras approach is to treat everything like an object. If the situation is different, the functional approach will imply layers, models, and optimizers as functions which would create more weights when called, and they would be stored in global name-indexed collections with other models which are using this approach. For example, TensorFlow has taken this approach. This method is just impractical since many operations like accessing an existing tensor or simple model loading must be approached by name-matching.

3 people found this helpful

  • Overall
    5 out of 5 stars
  • Performance
    5 out of 5 stars
  • Story
    5 out of 5 stars
Profile Image for Ronald
  • Ronald
  • 12-11-20

Clear and usable instructions and foundation

Great and well rounded insights into Deep Learning. Always focused on communicating the taught material through tangible and real world examples.

  • Overall
    5 out of 5 stars
  • Performance
    5 out of 5 stars
  • Story
    5 out of 5 stars
Profile Image for Veronica
  • Veronica
  • 12-11-20

Come for the theory, stay for the practical advic

The main features of Keras are implemented from commonly used neural network building blocks like objectives, layers, activation functions and optimizers. With Keras it is easy to work with text and image data since it features a host of tools suitable for any beginner. Facts are speaking for itself, Keras is the second-fastest growing framework, and that's not surprising. Keras is used to solve problems without having to interact with the TensorFlow or Theano, which are underlying backend engines. So, without any interaction with the background, end-to-end problems can be solved. Initially, Kears was developed on top of Theano, but shortly after the release of TensorFlow, Keras added it as backend. As new generation computation graph engines are designed, Keras will extend to support those as well in the future. Keras is not dependable on backend engines since it features its built-in graph data structure which can handle computational graphs. Therefore, it doesn't rely on TensorFlow's or Theano's native graph data structure. As a result of this independent data structure, Keras can make offline shape inference in Theano which is missing, but very much needed feature in Theano. Easy model copying and sharing are also doable in Keras.

  • Overall
    5 out of 5 stars
  • Performance
    5 out of 5 stars
  • Story
    5 out of 5 stars
Profile Image for Robert  Ruth
  • Robert Ruth
  • 09-11-20

The definitive guide to becoming a researcher

The definitive guide to becoming a researcher in the field Learn lot and did some practice too on code and trying to map some scenarios too. Very helpful and useful Course. I really like the way we made complicated concepts really simple. I really like and appreciates your view on ethics and moral values. Will taking course on Big Data by You. God Bless You.

  • Overall
    5 out of 5 stars
  • Performance
    5 out of 5 stars
  • Story
    5 out of 5 stars
Profile Image for Jerry
  • Jerry
  • 09-11-20

For those interested in DL math and research

Fantastic course. I will be going over it again. I am trying to get my teeth into this topic. This guys really knows what he talking about, and knows how to explain complex topics in a "simple" manner. Ranks up there as one of the best courses I took on this platform.