O'Reilly logo
live online training icon Live Online training

AI and ML on the AWS Platform First Steps

Hands-on with Amazon SageMaker Autopilot, Comprehend, and SageMaker

Topic: Data
Chris Fregly
Antje Barth

Running AI and machine learning workloads in the cloud is the most secure, most scalable, and least expensive option. Amazon Web Services offers AI and machine learning tools for a number of use cases, including image classification, object detection, natural language processing, voice recognition, conversational agents, translation, search, fraud detection, and data labeling. The Amazon AI and machine learning stack supports many state-of-the-art features, including automated machine learning (AutoML) using Amazon SageMaker Autopilot and one-click machine learning with Amazon Comprehend. Just select a CSV file, for example, and both SageMaker Autopilot and Comprehend will train, optimize, and deploy the best model that fits your dataset.

Join exerts Chris Fregly and Antje Barth to explore the simplicity, breadth, and depth of the Amazon AI and machine learning stack using the revolutionary BERT natural language processing (NLP) model with Amazon SageMaker Autopilot, Comprehend, and SageMaker along with TensorFlow, Keras, and PyTorch. You’ll learn how to generate features from raw CSV files; use them to fine-tune the base BERT model to classify reviews as ratings 1 (bad) through 5 (good) from the Amazon Customer Reviews Dataset; perform hyperparameter tuning and experiment tracking to optimize the model; and deploy the model to allow custom applications to classify any body of text including Twitter comments, customer service emails, and more.

What you'll learn-and how you can apply it

By the end of this live online course, you’ll understand:

  • The core components of the Amazon AI and machine learning stack
  • Best practices for building, training, tuning, and deploying AI/ML models on AWS
  • How to launch SageMaker Studio and Jupyter-based SageMaker notebooks
  • How to perform automated machine learning (AutoML) with SageMaker Autopilot
  • How to use hyperparameter tuning to find the best model to fit your dataset
  • How to utilize SageMaker Experiments to track and visually compare model performance
  • How to deploy models easily to production with SageMaker endpoints

And you’ll be able to:

  • Launch SageMaker Studio and Jupyter-based SageMaker notebooks
  • Train natural language processing (NLP) models using SageMaker Autopilot, Comprehend, SageMaker, TensorFlow, Keras, and PyTorch
  • Perform hyperparameter tuning to find the best model to fit your dataset
  • Track and compare models using SageMaker Experiments
  • Deploy models to production using SageMaker endpoints

This training course is for you because...

  • You’re a data scientist or ML engineer who needs to develop AI/ML pipelines for production applications.
  • You’re a data engineer who needs to build predictive models for production use.
  • You’re a data analyst or business intelligence specialist who wants to understand the breadth and depth of the Amazon AI and machine learning stack.

Prerequisites

  • A basic understanding of statistics, SQL, Python, and the command line (useful but not required)
  • No AWS account required

About your instructors

  • Chris Fregly is a San Francisco, California-based developer advocate for AI and machine learning at Amazon Web Services (AWS). He’s worked with Kubeflow and MLflow since 2017 and founded the global Advanced Kubeflow Meetup. Chris regularly speaks at ML/AI conferences across the world, including the O’Reilly AI and Strata Data Conferences. Previously, Chris was founder at PipelineAI, helping startups and enterprises continuously deploy AI and machine learning pipelines using Kubeflow and MLflow, and was an ML-focused engineer at both Netflix and Databricks.

  • Antje Barth is a Düsseldorf, Germany-based developer advocate for AI and machine learning at Amazon Web Services (AWS). She cofounded a chapter of Women in Big Data in Germany and regularly speaks at ML/AI conferences across the world, including the O’Reilly AI Conference. She’s been working with Kubeflow since 2018 and is passionate about helping developers leverage big data, Docker containers, and Kubernetes platforms in the context of AI and machine learning. Previously, Antje was a big data and ML engineer at both MapR and Cisco.

Schedule

The timeframes are only estimates and may vary according to how the class is progressing

Introductions, course agenda, and environment setup (20 minutes)

  • Presentation: Overview of the environment
  • Hands-on exercise: Set up and explore the environment
  • Q&A

Amazon AI and machine learning stack overview (20 minutes)

  • Presentation: The Amazon AI and ML stack; SageMaker Studio
  • Hands-on exercise: Launch SageMaker Studio
  • Q&A

Automated machine learning with the Amazon Customer Reviews Dataset (15 minutes)

  • Presentation: Applying AutoML to the Amazon Customer Reviews Dataset
  • Hands-on exercise: Perform AutoML with Amazon SageMaker Autopilot using the Amazon Customer Reviews Dataset
  • Q&A

Break (5 minutes)

One-click machine learning with the Amazon Comprehend AI service (20 minutes)

  • Presentation: Using Amazon Comprehend to uncover insights from the Amazon Customer Reviews Dataset with just one click
  • Hands-on exercise: Train and deploy a natural language processing (NLP) classifier based on the Amazon
  • Customer Reviews Dataset
  • Q&A

Jupyter-based SageMaker notebooks (10 minutes)

  • Presentation: Overview of the Jupyter-based Amazon SageMaker notebooks
  • Hands-on exercise: Launch a Jupyter-based SageMaker notebook
  • Q&A

Natural language processing with BERT and Amazon SageMaker (25 minutes)

  • Presentation: Using the state-of-the-art BERT pretrained family of natural language processing (NLP) and natural language understanding (NLU) models to classify Amazon customer reviews
  • Hands-on exercise: Launch a SageMaker job and find the best hyperparameters for your model and dataset
  • Q&A

Break (5 minutes)

Experiment tracking with Amazon SageMaker (15 minutes)

  • Presentation: Use cases for experiment tracking with Amazon SageMaker
  • Hands-on exercise: Track and compare experiments
  • Q&A

Hyperparameter tuning with Amazon SageMaker (20 minutes)

  • Presentation: The options for hyperparameter tuning with SageMaker
  • Hands-on exercise: Launch a hyperparameter tuning job with SageMaker and find the best model candidate

Model deployment with SageMaker endpoints and batch predictions (15 minutes)

  • Presentation: The options for deploying models with SageMaker endpoints and batch predictions
  • Hands-on exercise: Deploy the model as a SageMaker endpoint

Wrap-up and Q&A (10 minutes)