Downloaded 32 times
Uploaded byJulien SIMON
Machine Learning: From Notebook to Production with Amazon Sagemaker (January 2018)
The document discusses Amazon SageMaker, a fully managed machine learning platform. It provides an overview of the machine learning workflow from data collection and processing to model training, evaluation, deployment and monitoring. SageMaker allows users to build, train and deploy machine learning models using pre-built algorithms, bringing their own training code, or custom algorithms. It offers a scalable, flexible and cost-effective environment for developing and hosting machine learning models.
More Related Content
Similar to Machine Learning: From Notebook to Production with Amazon Sagemaker (January 2018)
Machine Learning: From Notebook to Production with Amazon Sagemaker (January 2018)
- 1. Machine Learning: From Notebookto Production with Amazon Sagemaker Julien Simon, AI Evangelist, EMEA @julsimon
- 2. © 2017, AmazonWeb Services, Inc. or its Affiliates. All rights reserved. Platform Services AWS ML Stack Deploy machine learning models with high-performance machine learning algorithms, broad framework support, and one-click training, tuning, and inference.
- 3. © 2017, AmazonWeb Services, Inc. or its Affiliates. All rights reserved. Data Visualization & Analysis Business Problem – ML problem framing Data Collection Data Integration Data Preparation & Cleaning Feature Engineering Model Training & Parameter Tuning Model Evaluation Are Business Goals met? Model Deployment Monitoring & Debugging YesNo DataAugmentation Feature Augmentation The Machine Learning Process Re-training Predictions
- 4. © 2017, AmazonWeb Services, Inc. or its Affiliates. All rights reserved. Data Visualization & Analysis Business Problem – ML problem framing Data Collection Data Integration Data Preparation & Cleaning Feature Engineering Model Training & Parameter Tuning Model Evaluation Are Business Goals met? Model Deployment Monitoring & Debugging YesNo DataAugmentation Feature Augmentation Problem discovery Re-training • Help formulate the right questions • Domain Knowledge Predictions
- 5. © 2017, AmazonWeb Services, Inc. or its Affiliates. All rights reserved. Data Visualization & Analysis Business Problem – ML problem framing Data Collection Data Integration Data Preparation & Cleaning Feature Engineering Model Training & Parameter Tuning Model Evaluation Are Business Goals met? Model Deployment Monitoring & Debugging YesNo DataAugmentation Feature Augmentation Retraining • Need a data platform? • Amazon S3 • AWS Glue • Amazon Athena • Amazon EMR • Amazon Redshift Spectrum Integration Predictions
- 6. © 2017, AmazonWeb Services, Inc. or its Affiliates. All rights reserved. Data Visualization & Analysis Business Problem – ML problem framing Data Collection Data Integration Data Preparation & Cleaning Feature Engineering Model Training & Parameter Tuning Model Evaluation Are Business Goals met? Model Deployment Monitoring & Debugging YesNo DataAugmentation Feature Augmentation Retraining Model Training Predictions • Setup and manage Notebook Environments • Setup and manage Training Clusters • Write Data Connectors • Scale ML algorithms to large datasets • Distribute ML training algorithm to multiple machines • Secure Model artifacts
- 7. © 2017, AmazonWeb Services, Inc. or its Affiliates. All rights reserved. Data Visualization & Analysis Business Problem – ML problem framing Data Collection Data Integration Data Preparation & Cleaning Feature Engineering Model Training & Parameter Tuning Model Evaluation Are Business Goals met? Model Deployment Monitoring & Debugging YesNo DataAugmentation Feature Augmentation Retraining Model Deployment Predictions • Setup and manage Model Inference Clusters • Manage and Scale Model Inference APIs • Monitor and Debug Model Predictions • Models versioning and performance tracking • Automate New Model version promotion to production (A/B testing)
- 8. © 2017, AmazonWeb Services, Inc. or its Affiliates. All rights reserved. End-to-End Machine Learning Platform Zero setup Flexible Model Training Pay by the second $ Amazon SageMaker Build, train, and deploy machine learning models at scale
- 9. © 2017, AmazonWeb Services, Inc. or its Affiliates. All rights reserved. Highly-optimized machine learning algorithms BuildPre-built notebook instances Amazon SageMaker
- 10. © 2017, AmazonWeb Services, Inc. or its Affiliates. All rights reserved. Highly-optimized machine learning algorithms One-click training for ML, DL, and custom algorithms BuildPre-built notebook instances Easier training with hyperparameter optimization Train Amazon SageMaker
- 11. © 2017, AmazonWeb Services, Inc. or its Affiliates. All rights reserved. One-click training for ML, DL, and custom algorithms Easier training with hyperparameter optimization Highly-optimized machine learning algorithms Deployment without engineering effort Fully-managed hosting at scale BuildPre-built notebook instances Deploy Train Amazon SageMaker
- 12. © 2017, AmazonWeb Services, Inc. or its Affiliates. All rights reserved. Amazon ECR Model Training (on EC2) Model Hosting (on EC2) Trainingdata Modelartifacts Training code Helper code Helper codeInference code GroundTruth Client application Inference code Training code Inference requestInference response Inference Endpoint Amazon SageMaker
- 13. © 2017, AmazonWeb Services, Inc. or its Affiliates. All rights reserved. Amazon SageMaker customers https://aws.amazon.com/sagemaker/customers/
- 14. Detecting buildings inVietnam https://developmentseed.org/blog/2018/01/19/sagemaker-label-maker-case/
- 15. © 2017, AmazonWeb Services, Inc. or its Affiliates. All rights reserved. Demos 1.Use a built-in algorithm: fine-tuning a pre-trained model for image classification 2.Bring your own training code: learning MNIST with Apache MXNet 3.Bring your own pre-trained model: clustering MNIST with k-means in scikit-learn 4.Bring your own algorithm: classifying the Iris data set with Decisions Trees in scikit-learn
- 16. © 2017, AmazonWeb Services, Inc. or its Affiliates. All rights reserved. End-to-End Machine Learning Platform Zero setup Flexible Model Training Pay by the second $ Amazon SageMaker Build, train, and deploy machine learning models at scale
- 17. © 2017, AmazonWeb Services, Inc. or its Affiliates. All rights reserved. Resources https://aws.amazon.com/machine-learning https://aws.amazon.com/blogs/ai https://aws.amazon.com/sagemaker An overview of Amazon SageMaker https://www.youtube.com/watch?v=ym7NEYEx9x4 https://medium.com/@julsimon
- 18. Thank you! Julien Simon,AI Evangelist, EMEA @julsimon
Editor's Notes
- #6 The Data platform
- #7 The Data platform
- #8 The Data platform
- #10 Pre-built Notebook Instances For training data exploration and preprocessing, Amazon SageMaker provides fully managed notebook instances running Jupyter notebooks that include example code for common model training and hosting exercises. These notebook instances are pre-loaded with Anaconda packages, and popular deep learning libraries like TensorFlow, and Apache MXNet. Highly-optimized Machine Learning Algorithms Amazon SageMaker installs high-performance, scalable machine learning algorithms optimized for speed, scale, and accuracy, to run on extremely large training datasets. Based on the type of learning that you are undertaking, you can choose from supervised algorithms, such as linear/logistic regression or classification; as well as unsupervised learning, such as with k-means clustering.
- #11 TRAIN One-click Training When you’re ready to train in Amazon SageMaker, simply indicate the type and quantity of instances you need and initiate training with a single click. SageMaker sets up the distributed compute cluster, performs the training, and tears down the cluster when complete. SageMaker seamlessly scales to tens of nodes with hundreds of GPUs, so you no longer need to worry about all the complexity and lost time involved in making distributed training architectures work. Built-in Automatic Hyperparameter Optimization (in Preview) Using built-in hyperparameter optimization (HPO), SageMaker can automatically tune your algorithm by adjusting hundreds of different combinations of parameters, to quickly arrive at the best solution for your machine learning problem. HPO lets you easily optimize an ML model on SageMaker by exploring lots of variations of the same algorithm with varying hyperparameters to pick the one with the best performance on your data.
- #12 DEPLOY Deployment without Engineering Effort After training, SageMaker provides the model artifacts and scoring images to you for deployment to Amazon EC2 or anywhere else. When you’re ready to deploy your model, you can launch into a secure and elastically scalable environment, with one-click deployment from the SageMaker console. Fully Managed Amazon SageMaker handles all of the compute infrastructure on your behalf, with built-in Amazon CloudWatch monitoring and logging, to perform health checks, apply security patches, and other routine maintenance, as well as ensure updates to the supported deep learning frameworks as they become available.