O slideshow foi denunciado.
Seu SlideShare está sendo baixado. ×

Integrating Deep Learning In the Enterprise

Anúncio
Anúncio
Anúncio
Anúncio
Anúncio
Anúncio
Anúncio
Anúncio
Anúncio
Anúncio
Anúncio
Anúncio
Carregando em…3
×

Confira estes a seguir

1 de 20 Anúncio

Integrating Deep Learning In the Enterprise

by Yash Pant, Enterprise Solutions Architect AWS

Amazon SageMaker is a fully managed platform for data scientists and developers to build, train and deploy machine learning models in production applications. In this workshop, you will learn how to integrate Amazon SageMaker with other AWS services in order to meet enterprise requirements. Using Amazon S3, Amazon Glue, Amazon KMS, Amazon SageMaker, Amazon CodeStar, Amazon ECR, IAM; we will walk through the machine learning lifecycle in an integrated AWS environment and discuss best practices. Attendees must have some familiarities with AWS products as well as a good understanding of machine learning theory. The dataset for the workshop will be provided.

by Yash Pant, Enterprise Solutions Architect AWS

Amazon SageMaker is a fully managed platform for data scientists and developers to build, train and deploy machine learning models in production applications. In this workshop, you will learn how to integrate Amazon SageMaker with other AWS services in order to meet enterprise requirements. Using Amazon S3, Amazon Glue, Amazon KMS, Amazon SageMaker, Amazon CodeStar, Amazon ECR, IAM; we will walk through the machine learning lifecycle in an integrated AWS environment and discuss best practices. Attendees must have some familiarities with AWS products as well as a good understanding of machine learning theory. The dataset for the workshop will be provided.

Anúncio
Anúncio

Mais Conteúdo rRelacionado

Diapositivos para si (20)

Semelhante a Integrating Deep Learning In the Enterprise (20)

Anúncio

Mais de Amazon Web Services (20)

Integrating Deep Learning In the Enterprise

  1. 1. © 2017, Amazon Web Services, Inc. or its Affiliates. All rights reserved Pop-up Loft Integrating Deep Learning Into Your Enterprise Yash Pant, Enterprise Solutions Architect
  2. 2. © 2017, Amazon Web Services, Inc. or its Affiliates. All rights reserved Recommendations & Ranking Personalized ranking, page generation, search, similarity, ratings
  3. 3. © 2017, Amazon Web Services, Inc. or its Affiliates. All rights reserved Uncovering Images that Resonate with Customers
  4. 4. © 2017, Amazon Web Services, Inc. or its Affiliates. All rights reserved The Amazon machine learning stack PLATFORM SERVICES APPLICATION SERVICES FRAMEWORKS & INTERFACES Caffe2 CNTK Apache MXNet PyTorch TensorFlow Torch Keras Gluon AWS Deep Learning AMIs Amazon SageMaker AWS DeepLens Rekognition Transcribe Translate Polly Comprehend Lex Amazon Mechanical Turk Amazon ML
  5. 5. © 2017, Amazon Web Services, Inc. or its Affiliates. All rights reserved Direct Connect Kinesis Snowball Snowmobile IoT Database Migration Streams, Firehose, Analytics Upload RDS MySQL, PostgreSQL, MariaDB Oracle, SQL Server Aurora MySQL PostgreSQL Storage, DB & Analytics S3 EC2 DynamoDB, Neptune EMR Elasticsearch ElastiCache Glacier Redshift Athena QuickSight Lambda Cloud Search Spectrum
  6. 6. © 2017, Amazon Web 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 – Predictions YesNo DataAugmentation Feature Augmentation The Machine Learning Process Re-training
  7. 7. © 2017, Amazon Web 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 – Predictions YesNo DataAugmentation Feature Augmentation The Machine Learning Process Re-training • Build the data platform: • Amazon S3 • AWS Glue • Amazon Athena • Amazon EMR • Amazon Redshift Spectrum
  8. 8. © 2017, Amazon Web 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 – Predictions YesNo DataAugmentation Feature Augmentation The Machine Learning Process Re-training • Builds the ML Models: • AWS Deep Learning AMI • SparkML on Amazon EMR • Amazon SageMaker
  9. 9. © 2017, Amazon Web 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 – Predictions YesNo DataAugmentation Feature Augmentation The Machine Learning Process Re-training • Build Smart Apps • AWS Lambda • Amazon S3 • API Gateway • IoT • Kinesis • ECS/ECR • Mobile Hub • AWS KMS • EC2 • Amazon SageMaker
  10. 10. © 2017, Amazon Web Services, Inc. or its Affiliates. All rights reserved Enterprise Data is Everywhere!! And many many more! How do we know what data is where? How do we catalog this data? How do we keep up to date with? How do we combine data across data sources? How do we analyze and create models?
  11. 11. © 2017, Amazon Web Services, Inc. or its Affiliates. All rights reserved Data Big Picture
  12. 12. © 2017, Amazon Web Services, Inc. or its Affiliates. All rights reserved Training Models Supports both CPU/GPU Training Run common libraries – Scikit-learn, R, Tensorflow, Keras, MXNet, Gluon, PyTorch, etc. Connect multiple EC2 instances to create ML/DL training clusters Preprocess large amounts of data efficiently for model training Take advantage of Spark ML to build models Dataset Training Model Amazon S3 Amazon EMRAmazon EC2 Amazon EMR Amazon EC2
  13. 13. © 2017, Amazon Web Services, Inc. or its Affiliates. All rights reserved Enterprise Training Considerations What happens when you have many data scientists? - How to manage EMR/Spark environments at scale? - How do you provide DL clusters to many scientists at scale? - How to work across different lines of business? - How to control costs? - How do you make sure everyone operates securely? Use an ML Platform! (Like Amazon SageMaker)
  14. 14. © 2017, Amazon Web Services, Inc. or its Affiliates. All rights reserved Moving to Production Tensorflow save_path = saver.save(sess, ”my_model") saver.restore(sess, "/my_model") MXNet mx.model.save("my_model") Model = mx.model.load("my_model") XGBoost bst.save_model(’my_model') bst2 = xgb.Booster(model_file=’my_model') Serialized flat file(s) & Others Amazon ECS AWS IoT & Greengrass AWS EC2/F1 AWS Lambda AWS Batch Amazon SageMaker
  15. 15. © 2017, Amazon Web Services, Inc. or its Affiliates. All rights reserved© 2017, Amazon Web Services, Inc. or its Affiliates. All rights reserved. Amazon S3 AWS CodeBuildAWS CodePipelineAWS CodeCommit App developer Amazon ECR Amazon ECS Data scientist Pulls Docker image from ECR Downloads model from S3 DevOps Engineer End User Deployment to ECS Checks in code Outputs model Execute Build
  16. 16. © 2017, Amazon Web Services, Inc. or its Affiliates. All rights reserved© 2017, Amazon Web Services, Inc. or its Affiliates. All rights reserved. Lambda based prediction Amazon S3 AWS CloudFormation AWS CodeBuildAWS CodePipelneAWS CodeCommit AWS Lambda 1.Check in code 2.Execute Build 3. Update function5. Predict 4. Load Model DevOps Engineer App Developer
  17. 17. © 2017, Amazon Web Services, Inc. or its Affiliates. All rights reserved ML Platform – Amazon SageMaker Fully managed service that enables developers and data scientists to quickly and easily build, train, and deploy machine learning models at any scale
  18. 18. © 2017, Amazon Web Services, Inc. or its Affiliates. All rights reserved Training code • Matrix Factorization • Regression • Principal Component Analysis • K-Means Clustering • Gradient Boosted Trees • And More! Amazon provided Algorithms Bring Your Own Script (SM builds the Container) Bring Your Own Algorithm (You build the Container) I ML Training Service Fetch Training data Save Model Artifacts Fully managed – Secured– Amazon ECR Save Inference Image IM Estimators in Apache Spark CPU GPU HPO Amazon SageMaker - Training
  19. 19. © 2017, Amazon Web Services, Inc. or its Affiliates. All rights reserved I ML Hosting Service Amazon ECR 30 50 10 10 InstanceType: c3.4xlarge MinInstanceCount: 5 MaxInstanceCount: 20 ModelName: prod VariantName: prodPrimary VariantWeight: 50 ProductionVariant Model Artifacts Inference Image Model versions Versions of the same inference code saved in inference containers. Prod is the primary one, 50% of the traffic must be served there! One-Click! EndpointConfiguration Inference Endpoint Amazon Provided Algorithms Amazon SageMaker BYOA Amazon SageMaker - Hosting
  20. 20. © 2017, Amazon Web Services, Inc. or its Affiliates. All rights reserved Summary Data is everywhere - We need a plan to manage and maintain it - Tools like S3, Glue, EMR/Spark help Use an ML Platform such as SageMaker to enable ML at scale - Options for data management, model training, and hosting

×