Recent Reviews of Similar Course:
Pure excellence from the presenter!!!! Great content!!! Buy this course, you won't regret it. – Social Scientist Redvers Crooks
Almost perfect, I feel like there can be more to the course but it is short and sweet. -- Christopher Brashear
Welcome to Building Deep Neural Networks in Keras Master Class. In this course, we are going to build an then tune Keras models.
The area of study which involves extracting knowledge from data is called as Data Science and people practicing in this field are called as Data Scientists.
Keras is a relatively new library in Python designed for building neural networks. The library sits on top of Theano and Tensorflow. This means it can take advantage of the computational speed and efficiency of the two yet offer a high level, comfortable interface that data scientists using Python are used to.
I’m using the term “master class” to denote that this isn’t an introductory course. I do expect the student to know some Python and basic neural network topology.
The top career in the world right now is that of the data scientist and the top machine learning tool right now is deep learning. Another name for deep learning is artificial neural networks. Artificial neural networks is the term you’ll see in academia and deep learning the more commercial term. Throughout the course, I will use the two interchangeably.
We are going to cover the five major steps involved in building models in Keras. Our first step will be loading data, secondly, we will be defining a model, thirdly we will be compiling a model, fourthly, fitting the model and finally evaluating the model. After we build the model we are going to delve deeper into evaluating the performance of keras models. Deep learning models have many buttons and knobs that can be tweaked and altered to deliver more accurate results.
In the course, we will skip the math and focus on data cleansing and model building. The two core skills you’ll need for a career in applied machine learning. Applied machine learning simply means you go to work and the models you build don’t end up in papers they end up in real world production environments.
**Five Reasons to Take this Course**
1) Wide Adoption of Keras
Deep learning is the single hottest niche in the machine learning field. Because Keras is written for Python it has a high level interface allows for ease of use for novice as well as more experienced users. It's quickly becoming the standard for rapid model deployment in the applied world.
2) Occam's Razor Approach to Teaching
Less is almost always more. If you're serious about deep learning as a career you don't need or want your hand held for long periods of time. You want the core of any subject and then you want to get your hands dirty. My courses are short and to the point. You don't have time for filler and I don't believe in adding it.
3) Real World Instructor Experience
I've been working with databases for over two decades and was building predictive analytic models when it was called data mining. There's really no difference between data mining and applied predictive analytics. Much of what you'll be doing as a data scientist or machine learning engineer is cleansing data and you'll find very few who have more data experience than DBAs.
4) Line by Line Code Explanation
In all my machine learning courses I explain every line of code. Python is very easy to learn but there's still a lot of nuances you'll need to know before mastering it specific to machine learning.
5) Limited Selection of Courses Specific to Keras
There are few courses specific to Keras. Even though it's been widely adopted, much like it's frame work library TensorFlow, very few have real world hands on experience with it. While I can't show you my production models but I can show you what I've learned building them.
Thanks for your interest in Deep Learning with Keras Masterclass. and we will see you in the course.
What is Keras?
Who created it?
What does high level really mean?
Let's find out in this lesson.
In this lesson let's discuss what we are going to learn.
Who is the course for?
It's not a beginners course although it's not advanced either.
You'll need to have Python installed.
I installed Python using the Anaconda distribution and I'd suggest you do the same.
Once you have Python up and running then this lecture will show you how to install Keras.
Keras sits on top of one of two libraries and uses them for numerical computations.
In this lesson let's find out what they are and how to change them.
This lesson is a brief question and answer lecture about the course, deep learning and machine learning in general.
Prior to importing our data we need to import our libraries and set a random seed.
Let's learn how to do that in this lesson.
In this lesson let's load our data.
I'll also show you how to download the data and put it in your own text file.
In this lesson let's learn about the core lines of code needed to build the model and the layers of our neural network.
Let's walk through the Keras model line by line.
This code is the core of every Keras model.
This can be difficult to understand but it's very important in deep learning.
Let's define bias graphically.
Let's compile our model.
In this short less we will fit our data to the model.
This is quick and easy.
Let's score the model and see how we did.
In this lesson let's ensure we know the difference between parameter and variable.
We are going to be passing a lot of Hyper Parameters into our models in the upcoming lesson so it's very import we know exactly what a parameter is.
In this lesson we will learn about automated model validation.
The great part about this is there's not one line of additional code.
We just alter an existing line and off we go.
Let's manually validate our model.
This is also easily done by adding a couple of line of code.
You'll quickly see just why Python is the gold standard in machine learning.
It's the gold standard in model validation but not normally used on deep learning models because of the computational cost.
Let's learn what it is in the brief lesson.
Not often used in deep learning models because of the computational impact it adds.
However, k-fold is the gold standard outside of the deep learning world and we can use it inside Keras.
Let's learn how.
Grid search is a model hyperparameter optimization technique.
In scikit-learn this technique is provided in the GridSearchCV class.
In this lesson let's learn how to apply it.
In this lecture let's build a Keras model for a binary classifier.
We will use a familiar data set.
The data set will be included in the downloads so you don't have to hunt it down.
We can force a type of feature extraction by the network by restricting the representational space in the ﬁrst hidden layer.
The translation for this is we can remove the number of neurons from our model.
In this lesson let's reduce the neurons in our first hidden layer.
In this lecture let's add another layer to our network.
We should see a significant boost in performance.
One hot encoding is a lot easier to understand when we can represent it visually.
In this lesson let's visual how machine learning uses one hot encoding.
In this lesson let's build a multi-class classification model.
In this lecture let's save our work.
Building larger neural networks can be time consuming.
Let's learn a few questions you might see on an interview specific to Keras.
Even though Keras is fairly new you'll need to answer questions about how to build production ready models.
Here are a few free resources you can use to expand you Keras knowledge.
Just a quick thank you from me to you for taking my course.
I've been a production SQL Server DBA most of my career.
I've worked with databases for over two decades. I've worked for or consulted with over 50 different companies as a full time employee or consultant. Fortune 500 as well as several small to mid-size companies. Some include: Georgia Pacific, SunTrust, Reed Construction Data, Building Systems Design, NetCertainty, The Home Shopping Network, SwingVote, Atlanta Gas and Light and Northrup Grumman.
Experience, education and passion
I learn something almost every day. I work with insanely smart people. I'm a voracious learner of all things SQL Server and I'm passionate about sharing what I've learned. My area of concentration is performance tuning. SQL Server is like an exotic sports car, it will run just fine in anyone's hands but put it in the hands of skilled tuner and it will perform like a race car.
Certifications are like college degrees, they are a great starting points to begin learning. I'm a Microsoft Certified Database Administrator (MCDBA), Microsoft Certified System Engineer (MCSE) and Microsoft Certified Trainer (MCT).
Born in Ohio, raised and educated in Pennsylvania, I currently reside in Atlanta with my wife and two children.