Implementing Multi-layer Neural Networks with TFLearn

Deep learning is one of the hottest topics for machine learning engineers. In this course, you'll quickly jump into building your first neural network using TFLearn on top of Tensorflow.
Course info
Level
Intermediate
Updated
Jan 23, 2019
Duration
2h 9m
Table of contents
Description
Course info
Level
Intermediate
Updated
Jan 23, 2019
Duration
2h 9m
Description

TFLearn offers machine learning engineers the ability to build Tensorflow neural networks with minimal use of coding. In this course, Implementing Multi-layer Neural Networks with TFLearn, you’ll learn foundational knowledge and gain the ability to build Tensorflow neural networks. First, you’ll explore how deep learning is used to accelerate artificial intelligence. Next, you’ll discover how to build convolutional neural networks. Finally, you’ll learn how to deploy both deep and generative neural networks. When you’re finished with this course, you’ll have the skills and knowledge of deep learning needed to build the next generation of artificial intelligence.

About the author
About the author

Thomas is a Senior Software Engineer and Certified ScrumMaster. He spends most of his time working with the Hortonwork Data Platform and Agile Coaching.

More from the author
Installing and Configuring Splunk
Beginner
1h 16m
Apr 9, 2019
Enterprise Skills in Hortonworks Data Platform
Intermediate
1h 37m
Sep 21, 2018
More courses by Thomas Henson
Section Introduction Transcripts
Section Introduction Transcripts

Course Overview
Hi everyone, my name is Thomas Henson, and welcome to my course, Implementing Multi-layer Neural Networks with TFLearn. I'm a data engineering advocate in the big data analytics community, and I'm extremely excited to share this course with you on deep learning. deep learning is one of the hottest segments in the analytics field today because it touches so many different technologies, from automobiles that drive themselves to lifesaving medical research, and possibly colonizing Mars. Deep Learning is behind all these efforts. Think about that for a moment. Deep Learning is involved in the largest technological advancements on the planet, and trying to leave the planet. It's a great time to join the deep learning community, and TFLearn is a great way to start learning the building blocks of deep learning so you can build your first neural network. In this course, we're going to get experience implementing multi-layer neural networks with TFLearn. Some of the major topics that we will cover include why is deep learning at the heart of so much of this innovation, where does TFLearn fit in the TensorFlow Universe, basic deep learning technology that you'll need to know as machine learning engineers, and how to build convolutional neural networks with TFLearn. By the end of this course, you'll know how to get started building neural networks with TFLearn. Before beginning this course, you should be comfortable with coding in Python and basic data analytics concepts. I hope you'll join me on this journey to learn about deep learning with Implementing Multi-layer Neural Networks with TFLearn course here at Pluralsight.

What Is TFLearn?
Hi, welcome back to Implementing Multi-layer Neural Networks with TFLearn. My name is Thomas Henson, and this module is all about what is TFLearn. In this module, we're going to talk about the abstraction layer, TFLearn, and how it can help machine learning engineers ease to jumping into building their first neural network or maybe it's their hundredth neural network, but TFLearn is really going to help to abstract away some of the code from TensorFlow. In this module, we're going to compare both TFLearn and TensorFlow and understand some of the basics for TFLearn, and even get our own lab up and running so we can run through some code with TFLearn. In this module, we're going to cover and compare the differences between machine learning frameworks and deep learning frameworks, and also machine learning libraries and deep learning libraries, so that we understand how those interact with each other, and understand when to use which one. Next we're going to define what TFLearn for machine learning engineers really is, and also learn how to use TFLearn to build out neural networks and how it integrates with the whole TensorFlow ecosystem and deep learning ecosystem. Then we're going to actually jump into some code by comparing TensorFlow and TFLearn with a simple MNIST Hello World application. And then finally, we're going to jump in and we're going to set up our own deep learning environment for us to use throughout this course and beyond, and we're going to be able to import and bring in TFLearn all simply using from the command line. We've got a couple of different development options, so we can do it in a local environment, we can do it in the cloud, or even maybe using some containers, but we'll tackle all that when we get to that portion in this module. Now let's discuss the difference between libraries and frameworks.

Building Activations in TFLearn
Hi, and welcome back to Implementing Multi-layer Neural Networks with TFLearn. This module is all about building activations in TFLearn. Activations are something that we've seen in the previous modules that we've gone through, but we haven't really dug in. We've seen concepts like softmax, linear, tanh, and some of the other activations. Now let's talk about that activation layer and compare some of the differences between those all throughout this module. Before we jump into our activations, let's look and see what all we're going to cover in this module around activations. First thing is we're going to give a base level understanding about what activation functions are, and really the calls and why we use those. Next, we're going to talk and define the different activation types, so what are available from a TensorFlow and a TFLearn perspective, and how are those implemented, and what are some of the defaults within the activation functions. And then lastly we're going to implement some activation functions in TFLearn, we're going to look back at some code that we've already used, and change a couple of different variables just to see what those differences are from the activation functions, and then, honestly, how easy it is just to change those compared to maybe TensorFlow or other deep learning frameworks. Now let's get ready to learn a little bit more about activation functions and kind of where they generate from.

Managing Data with TFLearn
Hi folks, welcome back to Implementing Multi-layer Neural Networks with TFLearn. This module is called Managing Data with TFLearn, so this one is all about data. With deep learning and machine learning engineers, one of the most important things for us is our data, so we're going to talk a little bit about how we can manage our data and some different options that TFLearn offers for us to be able to bring in data, import data, and even preprocess some of our data as well. In this module, we're going to start off by navigating the TFLearn data utilities, and so this is going to give us the opportunity to bring in both structured data and unstructured data. If you're not familiar with the concept of structured and unstructured data, think of your structured data as something that fits in nice little neat rows and columns, similar to what we were doing with our Titanic dataset. And then our unstructured data could be images, it could be voice audio, it could be PowerPoints, it could be anything that's not structured. Specifically, what we've done is that mnist dataset has been our unstructured data that we've been bringing in. Next we're going to talk about how we can preprocess data so as the data is coming in or before the data is actually going into our neural network and our layers, we can do some sort of preprocessing to add more functionality or add more metrics and features to our data. And then lastly we're going to dig into streaming data, and how we can use TFLearn to help with streaming data as we go through building new models. Now let's turn our focus and start digging into the DataUtils in TFLearn.

Running Models with TFLearn
Welcome back to Implementing Multi-layer Neural Networks with TFLearn. This module is called Running Models with TFLearn. In this module, we're going to learn about the options that TFLearn allows for us to use whenever we're implementing and running models, whether it be a deep neural network model or a generative neural network model. This module is broken down into four distinct parts. First we're going to learn about deep neural networks and understand the differences between a deep neural network, and then explore all about generative neural networks, and how those feature workloads are going to implement, and how they change what we've already been using and doing from a TFLearn perspective, and how we would implement those models in the future. Next, we're going to discover how we can implement these models whether we're looking for a deep neural network or whether we're looking for a generative neural network, and then also some of the functionality with it. So what does it mean when we're fitting a model, and what are some of those hyperparameters that we can tune to be able to make our model run faster or more efficient. And then finally, we're going to discuss what's next for us. As we're going through this course, we've learned a lot and we've taken in a lot, but now it's time to find out, okay, where can we further our education from a TFLearn perspective, and just in general around machine learning engineers. Now let's jump in and explore deep neural networks and understand what we've been doing throughout this course with implementing DNNs.