AWS Certified Machine Learning – Specialty – Early Access

Course

Intro Video

Photo of Mike Chambers

Mike Chambers

Training Architect

Length

08:26:06

Difficulty

Advanced

Course Details

Welcome to Linux Academy's all new AWS Certified Machine Learning - Specialty prep course. This course prepares you to take the AWS Certified Machine Learning - Specialty (MLS-C01) certification exam. It also gives you the hands-on experience required to use machine learning and deep learning in a real-world environment.

This course starts off with coming to grips with Machine Learning (ML), Deep Learning (DL), and Artificial Intelligence (AI) terminology. After the theory comes the practice. You'll get hands-on with a number of ML frameworks and AWS services specific to the certification.

Course code and other resources can be found here: https://github.com/linuxacademy/content-aws-mls-c01

Join the Linux Academy community Slack here: https://slack.linuxacademy.com/ and specifically the #aws-mls-c01-2019 channel.

Syllabus

Course Introduction

Getting Started

Course Introduction

00:03:49

Lesson Description:

This lesson introduces this course and how you can get the most of your time learning about Machine Learning.

About the Training Architect

00:01:56

Lesson Description:

You're going to spend a lot of time listening to Mike in this course. Let's find out more about him.

About the Exam

00:17:40

Lesson Description:

The AWS MLS-C01 exam is unlike any other exam AWS has. This lesson shows why. AWS course page: Link AWS exam guide: Link AWS sample questions: Link

Machine Learning Fundamentals

Artificial Intelligence

00:04:33

Lesson Description:

What is artificial intelligence?

What Is Machine Learning?

00:11:40

Lesson Description:

What is machine learning? In this lesson we use an example of linear regression to see how machine learning helps us make predictions.

What Is Deep Learning?

00:06:47

Lesson Description:

What is deep learning? In this lesson we take a look at artificial neural networks.

Machine Learning and Deep Learning Theory

Machine Learning Concepts

Section Introduction

00:00:57

Lesson Description:

This video provides a quick introduction into the next few lessons.

Machine Learning Lifecycle

00:13:52

Lesson Description:

In this lesson, we take a look at the machine learning lifecycle. This covers topics from data collection to inference.

Supervised, Unsupervised, and Reinforcement

00:09:51

Lesson Description:

In this lesson we take a look at the different types of machine learning.

Optimization

00:09:38

Lesson Description:

In this lesson we look at the ways machine learning models get closer to the correct predictions through gradient descent.

Regularization

00:03:41

Lesson Description:

Sometimes machine learning models don't perform as expected. In this lesson we see how regularization helps improve our models.

Hyperparameters

00:05:21

Lesson Description:

In this lesson we look at the hyperparameters in general. We also examine some of the most common hyperparameters and what they do.

Validation

00:04:09

Lesson Description:

In this lesson we look at how validation is used to check a model during training, and how we can use cross-validation to make effective use of all of our data.

Data

Section Introduction

00:00:50

Lesson Description:

This video provides a quick introduction into the next few lessons.

Feature Selection and Engineering

00:10:04

Lesson Description:

In this lesson we look at how to prepare data for machine learning. We discuss whether all of the data is needed.

Principal Component Analysis (PCA)

00:10:23

Lesson Description:

In this lesson we look in depth at Principal Component Analysis (PCA), the first machine learning algorithm in the course.

Missing and Unbalanced Data

00:11:23

Lesson Description:

In the real world data is not perfect. This lesson examines what to do when we have missing or unbalanced data.

Label and One Hot Encoding

00:04:35

Lesson Description:

For the most part, maching learning wants to see numbers. In this lesson, we see how to deal with text and other data that needs encoding.

Splitting and Randomization

00:03:44

Lesson Description:

We don't use all of our data to train machine learning models. In this lesson, we see how to effectively split our data for the different tasks in the learning lifecycle.

RecordIO Format

00:04:05

Lesson Description:

We use lots of data in machine learning, so it's important to move it around in the most efficient way possible. This lesson looks at the RecordIO file format.

Machine Learning Algorithms

Section Introduction

00:00:41

Lesson Description:

This video provides a quick introduction for the next few lessons.

Logistical Regression

00:07:54

Lesson Description:

Shall we look at logistical regression? Yes or no? In this lesson, we assume the answer was "yes."

Linear Regression

00:05:41

Lesson Description:

In this lesson we take another look at linear regression. This time we talk about its uses. And no, it’s not just about penguins.

Support Vector Machines

00:04:35

Lesson Description:

In this lesson, we look at support vector machines and how they work.

Decision Trees

00:09:47

Lesson Description:

In this lesson, we show how decision tree algorithms navigate multi-dimensional data to produce a tree.

Random Forests

00:05:46

Lesson Description:

As it turns out, decision trees aren't that good on their own. However, when several of them work together in a forest, they perform well. This lesson takes a walk in the forest.

K-Means

00:10:32

Lesson Description:

In this lesson we take a look at k-means and how it groups our data.

K-Nearest Neighbour

00:03:37

Lesson Description:

We are judged by the company we keep, and so is data. In this lesson, we examine how the k-nearest neighbours algorithm groups data.

Latent Dirichlet Allocation (LDA) Algorithm

00:08:52

Lesson Description:

In this lesson, we look at the latent Dirichlet allocation (LDA) algorithm, and how it's used to find meaning in text documents.

Deep Learning Algorithms

Section Introduction

00:00:41

Lesson Description:

This video provides a quick introduction into the next few lessons.

Neural Networks

00:15:13

Lesson Description:

In this lesson we start to look under the covers of artificial neural networks.

Convolutional Neural Networks (CNN)

00:10:27

Lesson Description:

Ok we reached the "Hotdog" or "Not Hotdog" point. This lesson covers convolutional neural networks and how they pick out meaning from images.

Recurrent Neural Networks (RNN)

00:09:56

Lesson Description:

Sometimes you want a machine learning model with some comprehension of what has happened in the past. This lesson looks at recurrent neural networks and how they do just that.

Model Performance and Optimization

Section Introduction

00:01:28

Lesson Description:

This video provides a quick introduction into the next few lessons.

Confusion Matrix

00:11:58

Lesson Description:

Already confused by machine learning? Well maybe the confusion matrix can help. These tools help us understand the testing output from an ML model. They are a cornerstone for making further calculations.

Sensitivity and Specificity

00:14:43

Lesson Description:

When evaluating a machine learning model, it's critical to understand the business or data problem you're trying to solve. When you know this, you can decide if your model needs to be more sensitive or more specific.

Accuracy and Precision

00:05:41

Lesson Description:

Here are some basic ways to see how well your machine learning model is working. But are they too simple? Maybe, but we cover other options in other lessons.

ROC/AUC

00:18:01

Lesson Description:

In a previous lesson we looked at the sensitivity and specificity of a machine learning model. But how far should we take this? When does a model become so sensitive that it's usless? ROC helps us determine the answer to this question, and AUC will help us see which model seperates our classes the best.

Gini Impurity

00:07:31

Lesson Description:

How does a decision tree decide? Gini impurity is one of the calcualtions determining where to place nodes inside the tree.

F1 Score

00:04:56

Lesson Description:

Machine learning accuracy is sometimes not the best way to judge the effectivness of our model. F1 Score takes more into account and can be a better measure for models if we have an uneven class distribution.

Machine Learning Tools and Frameworks

Section Introduction

00:01:48

Lesson Description:

This video provides a quick introduction into the next few lessons.

Introduction to Jupyter Notebooks

00:16:05

Lesson Description:

In this lesson, we take a look at Jupyter Notebooks, a core tool in the toolbox of any machine learning engineer. Jupyter Notebooks is an integrated development environment geared toward documentation and collaboration.

ML and DL Frameworks

00:11:54

Lesson Description:

In this lesson, we take a high-level look at the most popular machine learning frameworks. We also look at the difference between a framework and an algorithm.

TensorFlow

00:16:42

Lesson Description:

In this lesson, we look at TensorFlow. This is currently the most popular machine learning and deep learning framework. In this lesson, we create a simple TensorFlow graph and see it in action.

PyTorch

00:09:23

Lesson Description:

In this lesson, we take a quick look at PyTorch, a machine learning framework based on Python. We create some basic objects and manipluate them within PyTorch.

MXNet

00:07:34

Lesson Description:

In this lesson, we take a quick look at MXNet and how it works with data. MXNet is one of the key machine learning frameworks behind the AI services provided by AWS.

Scikit-learn

00:14:36

Lesson Description:

In this lesson, we take a look at the popular machine learning framework, Scikit-learn. This 'all rounder' has built in algorithms for most of the popular machine learning architectures. We use Scikit-learn for more examples in this course.

AWS

AWS Services

Section Introduction

00:01:25

Lesson Description:

This video provides a quick introduction into the next few lessons.

S3

00:21:00

Lesson Description:

S3 is a cornerstone of the AWS platform, and it plays an important role in most ML piplines. In this lesson we review S3, its part in data lakes, and how we secure our precious data.

Glue

00:16:34

Lesson Description:

AWS Glue is a fully managed serverless service for data secovery and ETL (extract, transform, load). In this video, we review Glue and set up a crawler that will discover some data within S3.

Athena

00:13:49

Lesson Description:

Athena is often described as a SQL interface for S3. We can use this capabilty to prepare our data ahead of machine learning and perform some feature engineering.

QuickSight

00:08:29

Lesson Description:

Amazon QuickSight was pushed heavily by AWS as the BI (business intelligence) tool of choice for any AWS-centric business teams. For us in machine learning, it's a convenient way to visualise data ahead of choosing an ML algorithm to use. In this lesson, we look at Amazon QuickSight and use it to view our demo pengine data.

Kinesis, Streams, Firehose, Video, and Analytics

00:15:06

Lesson Description:

Kinesis is a family of services all focused on dealing with large scale data. We can use these services to form the ingestion part of an ML pipeline. In this lesson, we review Kinesis and a couple of example architectures.

EMR with Spark

00:06:20

Lesson Description:

Elastic MapReduce (EMR) provides an efficent architecture to process masses of data. In this lesson, we look at EMR and how AWS partners with Apache Spark to integrate EMR with ML pipelines in AWS.

EC2 for ML

00:10:54

Lesson Description:

EC2 instances remain the core compute platform for ML as lambda functions simply don't cut it here. In this lesson, we review EC2, look at the instance types specifically targeted at ML, and the AMIs that AWS maintains for bare-bones ML/DL workloads. We also remind ourselves about the resource limits placed on new accounts and how that impacts ML workloads.

Amazon ML

00:02:03

Lesson Description:

Many people have not heard of Amazon ML. Amazon ML is AWS's previous-generation ML platform, and it's sometimes thrown in as a red herring in the certification exam. Watch out!

AWS Application Services AI/ML

Section Introduction

00:03:15

Lesson Description:

In this video, we take a look at the section to come.

Amazon Rekognition (Images) Part 1

00:13:35

Lesson Description:

In this lesson, we take a look at Amazon's image analysis service called Rekognition. We take a look at the AWS console demo pages before the next lesson, where we dive into some application development with the API.

Amazon Rekognition (Images) Part 2 - the API

00:23:19

Lesson Description:

In this lesson, we take a deeper look at Amazon Rekognition by developing a simple script that uses the API. Can we find the guilty suspect from the security camera footage?

Amazon Rekognition (Video)

00:12:32

Lesson Description:

In this lesson, we take a look at example architectures we can use with Amazon Rekognition Video. There are two samples here: one for prerecorded video footage and one for live streaming. Understanding these is important before heading into the exam.

Amazon Polly

00:09:23

Lesson Description:

In this lesson, we take a look at Amazon's text-to-speech service Amazon Polly.

Amazon Transcribe

00:10:27

Lesson Description:

In this lesson, we look Amazons's speech-to-text service Amazon Transcribe.

Amazon Translate

00:13:52

Lesson Description:

In questa lezione, utilizziamo il servizio di traduzione di Amazon. To put it another way, in this lesson, we look at Amazon's translation service Amazon Translate.

Amazon Comprehend

00:14:29

Lesson Description:

In this lesson, we take a look at Amazon's text analysis service Amazon Comprehend. This is a powerful service with many features including text sentiment analysis and keyword extraction. The use cases for this service include extracting value from unstructured data.

Amazon Lex

00:13:33

Lesson Description:

Anyone who's used Amazon Alexa will be familar with the technology behind Amazon Lex. In this lesson, we take a brief look at this chatbot service and how it uses natrual language processing to interact with us.

Amazon Service Chaining with AWS Step Functions

00:16:26

Lesson Description:

In this lesson, we look at how AWS Step Functions can be used to chain together a pipeline of services into a complete solution.

Amazon SageMaker

Introduction

Section Introduction

00:01:39

Lesson Description:

This video provides a quick introduction for the next few lessons.

What is Amazon SageMaker?

00:07:18

Lesson Description:

In this video, we take a high-level review of the SageMaker service. We also share why it's not like any other service in AWS.

The Three Stages

00:03:15

Lesson Description:

In this lesson, we look at the three stages of SageMaker that AWS refers to throughout their documentation.

Control (Console/SDK/Notebooks)

00:12:07

Lesson Description:

In this lesson, we take a look at the different ways we can control SageMaker. These include two different APIs or SDKs.

SageMaker Notebooks

00:15:07

Lesson Description:

We have looked at Jupyter notebooks elsewhere in this course, and in this lesson we look at SageMaker notebooks. They are of course one and the same, but in this lesson we look at the SageMaker specifics of how they work.

Build

Data Preprocessing

00:13:57

Lesson Description:

In this lesson, we look at the various processes we can perform on data prior to training in the "build" stage of the SageMaker service.

Ground Truth

00:09:41

Lesson Description:

In this lesson, we see how Amazon integrates people and processes within the Ground Truth service to help us create quality training datasets.

Preprocessing Image Data (Pinehead NotPinehead)

00:26:38

Lesson Description:

This is part 1 of a set of lessons in this course where we build an end-to-end machine learning solution to classify an image as "Pinehead" (the Linux Academy mascot) or "Not Pinehead". In this lesson, we look at performing data synthesis to expand out limited dataset. (This course is in "Early Access". Some of the resources for this lesson may not be available - or fully working - yet.) Resources: https://github.com/linuxacademy/content-aws-mls-c01

Algorithms

00:14:30

Lesson Description:

In this lesson, we take a look at the various algorithms that can be loaded into SageMaker. AWS Resource Link: https://docs.aws.amazon.com/sagemaker/latest/dg/algos.html

Train

SageMaker Algorithms - Architecture 1

00:10:19

Lesson Description:

In this, the first part of three lessons, we take a closer look at the architecture of SageMaker algorithms.

SageMaker Algorithms - Architecture 2

00:07:02

Lesson Description:

In this, the second part of three lessons, we take a closer look at the architecture of SageMaker algorithms.

SageMaker Algorithms - Architecture 3

00:05:58

Lesson Description:

In this, the third part of three lessons, we take a closer look at the architecture of SageMaker algorithms.

Training an Image Classifier - Part 1 (Pinehead NotPinehead)

00:19:02

Lesson Description:

This is the second part of a set of lessons in this course where we build an end-to-end machine learning solution to classify an image as "Pinehead" (the Linux Academy mascot) or "Not Pinehead". In this lesson, we look at training our model with the dataset we created earlier. (This course is in "Early Access", and some of the resources for this lesson may not be available - or fully working - yet.) Resources: https://github.com/linuxacademy/content-aws-mls-c01

Training an Image Classifier - Part 2 (Pinehead NotPinehead)

00:04:48

Lesson Description:

This is part two (and a half) of a set of lessons in this course where we build an end-to-end machine learning solution to classify an image as "Pinehead" (the Linux Academy mascot) or "Not Pinehead". In this lesson, we are training our model. In this video, we look at the metrics achieved during training. (This course is in "Early Access", and some of the resources for this lesson may not be available - or fully working - yet.) Resources: https://github.com/linuxacademy/content-aws-mls-c01

Hyperparameter Tuning

00:10:21

Lesson Description:

In this lesson, we take a look at what AWS either calls "Hyperparameter Tuning" or "Amazon SageMaker Automatic Model Tuning". We use this service to remove the undifferentiated heavy lifting of optimising the adjustment of hyperparameters.

Deploy

Inference Pipelines

00:03:42

Lesson Description:

In this lesson, we review the idea of deploying muitple models into a single pipline.

Real-Time and Batch Inference

00:06:01

Lesson Description:

In this lesson, we look at both "real-time" and "batch" inference architectures.

Deploy an Image Classifier (Pinehead, NotPinehead)

00:16:33

Lesson Description:

This is part three of a set of lessons in this course where we build an end-to-end machine learning solution to classify an image as "Pinehead" (the Linux Academy mascot) or "Not Pinehead". In this lesson, we look at deploying our model so that we can make inferences. (This course is in "Early Access", and some of the resources for this lesson may not be available - or fully working - yet.) Resources: https://github.com/linuxacademy/content-aws-mls-c01

Accessing Inference from Apps

00:03:56

Lesson Description:

In this lesson, we look at the architecture for accessing inference from outside our AWS account.

Create a custom API for inference - Part 1 (Pinehead NotPinehead)

00:09:05

Lesson Description:

This is part four of a set of lessons in this course where we build an end-to-end machine learning solution to classify an image as "Pinehead" (the Linux Academy mascot) or "Not Pinehead". In this lesson, we look at creating a public facing API to allow for inference from the outside world. (This course is in "Early Access", and some of the resources for this lesson may not be available - or fully working - yet.) Resources: https://github.com/linuxacademy/content-aws-mls-c01

Create a custom API for inference - Part 2 (Pinehead NotPinehead)

00:11:51

Lesson Description:

This is part four (and a half) of a set of lessons in this course where we build an end-to-end machine learning solution to classify an image as "Pinehead" (the Linux Academy mascot) or "Not Pinehead". In this lesson, we finish the deployment of our API Gateway and test it using Postman and a custom iPhone app. (This course is in "Early Access", and some of the resources for this lesson may not be available - or fully working - yet.) Resources: https://github.com/linuxacademy/content-aws-mls-c01 Postman: https://www.getpostman.com/

Security

Securing SageMaker Notebooks

00:19:13

Lesson Description:

In this lesson, we review some of the security architectures for Amazon SageMaker. We also demo how IAM policies, attached to IAM roles, change the access from an Amazon SageMaker notebook instance.

SageMaker and the VPC

00:04:45

Lesson Description:

In this lesson, we review the ways SageMaker can interact with AWS Virtual Private Cloud (VPC).

Extra Bonus Lessons

Other AWS Services

Section Introduction

00:00:52

Lesson Description:

This video provides a quick introduction into the next few lessons.

DeepLens - Part 1

00:24:07

Lesson Description:

In this bonus extra lesson, we start to configure AWS DeepLens to work with our Pinehead(/NotPinehead) model.

DeepLens - Part 2

00:05:30

Lesson Description:

In this bonus lesson, we look at the deployed model to the AWS DeepLens.

DeepRacer - Part 1

00:23:07

Lesson Description:

In this bonus lesson, we take a look at AWS DeepRacer and see how we can use reinforcement learning to train the car to drive around a track.

DeepRacer - Part 2

00:05:25

Lesson Description:

In this bonus lesson we evaluate the DeepRacer training we kicked off in the last lesson.

Course Conclusion

The Exam

How to Answer Questions

00:16:16

Lesson Description:

In this lesson we look at methods for answering AWS exam questions. These tips can help give you the boost even on questions you're not sure about.

Thank You

Goodbye!

00:04:37

Lesson Description:

Thanks for taking this course! Please don't forget to press the thumbs up button on all the lessons you liked! In this lesson, we remember some of the moments from this course. Bye!