SlideShare a Scribd company logo
1 of 42
S U M M I T
Taipei
S U M M I T © 2019, Amazon Web Services, Inc. or its affiliates. All rights reserved.
https://bit.ly/2MxHNLB
© 2019, Amazon Web Services, Inc. or its affiliates. All rights reserved.S U M M I T
Automatic Labelling and Model
Tuning with Amazon
SageMaker
Adam Lynch
Partner Solutions Architect
Amazon Web Services
S e s s i o n I D
© 2019, Amazon Web Services, Inc. or its affiliates. All rights reserved.S U M M I T
Agenda
Introduction – 15 min
Overview of Amazon SageMaker GroundTruth – 15 min
Lab Automatic Labelling with Amazon SageMaker GroundTruth – 45 min
Working Lunch – 15 min
Lab Private Workforce Labelling – 30 min
Overview of Model Tuning using Bayesian Optimisation – 15 min
Lab Automatic Model Tuning with Amazon SageMaker – 45 min
© 2019, Amazon Web Services, Inc. or its affiliates. All rights reserved.S U M M I T
AWS Account for the Labs
In order to complete this workshop you'll need an AWS Account
with admin access
There are resources required by this workshop that are eligible for
the AWS free tier if your account is less than 12 months old.
And we will supply some credits for other tasks.
S U M M I T © 2019, Amazon Web Services, Inc. or its affiliates. All rights reserved.
© 2019, Amazon Web Services, Inc. or its affiliates. All rights reserved.S U M M I T
What data science tasks can be automated?
© 2019, Amazon Web Services, Inc. or its affiliates. All rights reserved.S U M M I T
Data science tasks
© 2019, Amazon Web Services, Inc. or its affiliates. All rights reserved.S U M M I T
AutoML – definition and goals
AutoML aims to maximise the performace of machine learning programs
without human assistance and subject to a computational budget.
Core goals:
a) Good performance: good generalization performance across various
input data and learning tasks can be achieved.
b) No assistance from humans: configurations can be automatically done
for machine learning tools.
c) High computational efficiency: the program can return a reasonable
output within a limited budget.
Taking the Human out of Learning Applications:
A Survey on Automated Machine Learning
Quanming Yao et al. arXiv:
1810.13306v3 [cs.AI] 17 Jan 2019
© 2019, Amazon Web Services, Inc. or its affiliates. All rights reserved.S U M M I T
The challenges of AutoML adoption
Deep Learning Human Design Computational
Budget
© 2019, Amazon Web Services, Inc. or its affiliates. All rights reserved.S U M M I T
Some tough challenges in machine learning
Availability of Labelled Data The Road Not Travelled Drift
© 2019, Amazon Web Services, Inc. or its affiliates. All rights reserved.S U M M I T
Promising approaches
Learning to Learn Simulation Reinforcement Learning
© 2019, Amazon Web Services, Inc. or its affiliates. All rights reserved.S U M M I T
KERAS
AI SERVICES
ML SERVICES
ML FRAMEWORKS +
INFRASTRUCTURE C5 C5n
P3 P3dn
Vision Speech Languages Chatbots Vertical
AMAZON
SAGEMAKER
AMAZON SAGEMAKER
GROUND TRUTH
AMAZON
SAGEMAKER RL
AWS MARKETPLACE
FOR ML
AMAZON
SAGEMAKER NEO
AWS
DEEPRACER
AWS
DEEPLENS
AMAZON ELASTIC INFERENCE
AWS INFERENTIA
AWS GREENGRASS
AMAZON
PERSONALIZE
AMAZON
FORECAST
AMAZON
TEXTRACT
AMAZON
REKOGNITION
AMAZON
LEX
AMAZON
POLLY
AMAZON
TRANSCRIBE
AMAZON
COMPREHEND
AMAZON
TRANSLATE
N E W
N E W
N E W
S U M M I T © 2019, Amazon Web Services, Inc. or its affiliates. All rights reserved.
© 2019, Amazon Web Services, Inc. or its affiliates. All rights reserved.S U M M I T
Data labeling and machine learning
Labeled data
Model
training
Prepare and
label data
Inference
© 2019, Amazon Web Services, Inc. or its affiliates. All rights reserved.S U M M I T
Data labeling is hard…
Need to label large datasets
Requires humans to perform labeling
Becomes time consuming and costly
Difficult to achieve high accuracy for
labels
© 2019, Amazon Web Services, Inc. or its affiliates. All rights reserved.S U M M I T
Amazon SageMaker: Build, train, and deploy ML
1
2
3
1
2
3
© 2019, Amazon Web Services, Inc. or its affiliates. All rights reserved.S U M M I T
K E Y F E A T U R E S
Automatic labeling via
machine learning
Ready-made and
custom workflows
Label
management
Private and public
human workforce
Amazon SageMaker Ground Truth
Label machine learning training data easily and
accurately
© 2019, Amazon Web Services, Inc. or its affiliates. All rights reserved.S U M M I T
Amazon SageMaker Ground Truth
How it works
© 2019, Amazon Web Services, Inc. or its affiliates. All rights reserved.S U M M I T
Amazon SageMaker Ground Truth
How it works
© 2019, Amazon Web Services, Inc. or its affiliates. All rights reserved.S U M M I T
Amazon SageMaker Ground Truth
How it works
© 2019, Amazon Web Services, Inc. or its affiliates. All rights reserved.S U M M I T
Amazon SageMaker Ground Truth
How it works
© 2019, Amazon Web Services, Inc. or its affiliates. All rights reserved.S U M M I T
Amazon SageMaker Ground Truth
How it works
© 2019, Amazon Web Services, Inc. or its affiliates. All rights reserved.S U M M I T
Step 1: Create a Ground Truth labeling job
© 2019, Amazon Web Services, Inc. or its affiliates. All rights reserved.S U M M I T
Step 2: Provide details for a labeling job
© 2019, Amazon Web Services, Inc. or its affiliates. All rights reserved.S U M M I T
Step 2: Provide details for a labeling job
© 2019, Amazon Web Services, Inc. or its affiliates. All rights reserved.S U M M I T
Step 2: Provide details for a labeling job
© 2019, Amazon Web Services, Inc. or its affiliates. All rights reserved.S U M M I T
Labeling job is now running
© 2019, Amazon Web Services, Inc. or its affiliates. All rights reserved.S U M M I T
Labeling app: Human workers label the
images
© 2019, Amazon Web Services, Inc. or its affiliates. All rights reserved.S U M M I T
AWS Management Console: View labels for
images
© 2019, Amazon Web Services, Inc. or its affiliates. All rights reserved.S U M M I T
AWS Management Console: View labels
S U M M I T © 2019, Amazon Web Services, Inc. or its affiliates. All rights reserved.
S U M M I T © 2019, Amazon Web Services, Inc. or its affiliates. All rights reserved.
S U M M I T © 2019, Amazon Web Services, Inc. or its affiliates. All rights reserved.
S U M M I T © 2019, Amazon Web Services, Inc. or its affiliates. All rights reserved.
© 2019, Amazon Web Services, Inc. or its affiliates. All rights reserved.S U M M I T
Solving the following problem:
𝑚𝑎𝑥 𝑥 ∈ 𝐴 𝑓(𝑥)
S U M M I T © 2019, Amazon Web Services, Inc. or its affiliates. All rights reserved.
Defining ‘A’ using Amazon SageMaker
"ParameterRanges": {
"CategoricalParameterRanges":
[
{ "Name": "tree_method",
"Values": ["auto", "exact", "approx", "hist"]}
],
"ContinuousParameterRanges":
[
{ "Name": "eta", "MaxValue" : "0.5", "MinValue": "0" }
],
"IntegerParameterRanges":
[
{ "Name": "max_depth", "MaxValue": "10", "MinValue": "1", }
]
}
S U M M I T © 2019, Amazon Web Services, Inc. or its affiliates. All rights reserved.
On Amazon Sagemaker – configure the tuning
job
tuning_job_config = {
"ParameterRanges": {…}
"ResourceLimits": { "MaxNumberOfTrainingJobs": 20, "MaxParallelTrainingJobs": 3 },
"Strategy": "Bayesian",
"HyperParameterTuningJobObjective": {
"MetricName": "validation:auc",
"Type": "Maximize"
}
}
S U M M I T © 2019, Amazon Web Services, Inc. or its affiliates. All rights reserved.
Launch the tuning job
tuning_job = smclient.create_hyper_parameter_tuning_job(
HyperParameterTuningJobName = "MyTuningJob“,
HyperParameterTuningJobConfig = tuning_job_config,
TrainingJobDefinition = training_job_definition)
S U M M I T © 2019, Amazon Web Services, Inc. or its affiliates. All rights reserved.
Thank you!
S U M M I T © 2019, Amazon Web Services, Inc. or its affiliates. All rights reserved.
Adam Lynch
atlynch@amazon.com

More Related Content

What's hot

What's hot (20)

Why AWS for running Microsoft workloads - CMP202-I - New York AWS Summit
Why AWS for running Microsoft workloads - CMP202-I - New York AWS SummitWhy AWS for running Microsoft workloads - CMP202-I - New York AWS Summit
Why AWS for running Microsoft workloads - CMP202-I - New York AWS Summit
 
Amazon SageMaker: ML for Every Developer and Data Scientist - AIM202 - Anahei...
Amazon SageMaker: ML for Every Developer and Data Scientist - AIM202 - Anahei...Amazon SageMaker: ML for Every Developer and Data Scientist - AIM202 - Anahei...
Amazon SageMaker: ML for Every Developer and Data Scientist - AIM202 - Anahei...
 
What's new in Amazon EC2 - CMP201 - Santa Clara AWS Summit.pdf
What's new in Amazon EC2 - CMP201 - Santa Clara AWS Summit.pdfWhat's new in Amazon EC2 - CMP201 - Santa Clara AWS Summit.pdf
What's new in Amazon EC2 - CMP201 - Santa Clara AWS Summit.pdf
 
Creare e gestire Data Lake e Data Warehouses
Creare e gestire Data Lake e Data WarehousesCreare e gestire Data Lake e Data Warehouses
Creare e gestire Data Lake e Data Warehouses
 
Pro-Tips-for-Builders-on-AWS
Pro-Tips-for-Builders-on-AWSPro-Tips-for-Builders-on-AWS
Pro-Tips-for-Builders-on-AWS
 
Grid computing in the cloud for Financial Services industry - CMP205-I - New ...
Grid computing in the cloud for Financial Services industry - CMP205-I - New ...Grid computing in the cloud for Financial Services industry - CMP205-I - New ...
Grid computing in the cloud for Financial Services industry - CMP205-I - New ...
 
What's New with Amazon Redshift - ADB202 - Anaheim AWS Summit
What's New with Amazon Redshift - ADB202 - Anaheim AWS SummitWhat's New with Amazon Redshift - ADB202 - Anaheim AWS Summit
What's New with Amazon Redshift - ADB202 - Anaheim AWS Summit
 
Need for Speed – Intro To Real-Time Data Streaming Analytics on AWS | AWS Sum...
Need for Speed – Intro To Real-Time Data Streaming Analytics on AWS | AWS Sum...Need for Speed – Intro To Real-Time Data Streaming Analytics on AWS | AWS Sum...
Need for Speed – Intro To Real-Time Data Streaming Analytics on AWS | AWS Sum...
 
Amazon EC2 A1 instances, powered by the AWS Graviton processor - CMP303 - San...
Amazon EC2 A1 instances, powered by the AWS Graviton processor - CMP303 - San...Amazon EC2 A1 instances, powered by the AWS Graviton processor - CMP303 - San...
Amazon EC2 A1 instances, powered by the AWS Graviton processor - CMP303 - San...
 
Running Amazon Elastic Compute Cloud (Amazon EC2) workloads at scale - CMP202...
Running Amazon Elastic Compute Cloud (Amazon EC2) workloads at scale - CMP202...Running Amazon Elastic Compute Cloud (Amazon EC2) workloads at scale - CMP202...
Running Amazon Elastic Compute Cloud (Amazon EC2) workloads at scale - CMP202...
 
Migliora la disponibilità e le prestazioni delle tue applicazioni con Amazon ...
Migliora la disponibilità e le prestazioni delle tue applicazioni con Amazon ...Migliora la disponibilità e le prestazioni delle tue applicazioni con Amazon ...
Migliora la disponibilità e le prestazioni delle tue applicazioni con Amazon ...
 
Developing serverless applications with .NET using AWS SDK & tools - MAD311 -...
Developing serverless applications with .NET using AWS SDK & tools - MAD311 -...Developing serverless applications with .NET using AWS SDK & tools - MAD311 -...
Developing serverless applications with .NET using AWS SDK & tools - MAD311 -...
 
Microservices on AWS: Architectural Patterns and Best Practices | AWS Summit ...
Microservices on AWS: Architectural Patterns and Best Practices | AWS Summit ...Microservices on AWS: Architectural Patterns and Best Practices | AWS Summit ...
Microservices on AWS: Architectural Patterns and Best Practices | AWS Summit ...
 
Building home security solutions at scale, ft. Comcast - SVC206 - New York AW...
Building home security solutions at scale, ft. Comcast - SVC206 - New York AW...Building home security solutions at scale, ft. Comcast - SVC206 - New York AW...
Building home security solutions at scale, ft. Comcast - SVC206 - New York AW...
 
Architetture per l'analisi di flussi di dati in tempo reale
Architetture per l'analisi di flussi di dati in tempo realeArchitetture per l'analisi di flussi di dati in tempo reale
Architetture per l'analisi di flussi di dati in tempo reale
 
Build intelligent applications quickly with AWS AI services - AIM301 - New Yo...
Build intelligent applications quickly with AWS AI services - AIM301 - New Yo...Build intelligent applications quickly with AWS AI services - AIM301 - New Yo...
Build intelligent applications quickly with AWS AI services - AIM301 - New Yo...
 
Building enterprise solutions with blockchain technology - SVC217 - New York ...
Building enterprise solutions with blockchain technology - SVC217 - New York ...Building enterprise solutions with blockchain technology - SVC217 - New York ...
Building enterprise solutions with blockchain technology - SVC217 - New York ...
 
Get hands-on with AWS DeepRacer and compete in the AWS DeepRacer League - AIM...
Get hands-on with AWS DeepRacer and compete in the AWS DeepRacer League - AIM...Get hands-on with AWS DeepRacer and compete in the AWS DeepRacer League - AIM...
Get hands-on with AWS DeepRacer and compete in the AWS DeepRacer League - AIM...
 
CI/CD best practices for building modern applications - MAD301 - Santa Clara ...
CI/CD best practices for building modern applications - MAD301 - Santa Clara ...CI/CD best practices for building modern applications - MAD301 - Santa Clara ...
CI/CD best practices for building modern applications - MAD301 - Santa Clara ...
 
[REPEAT] Optimize your workloads with Amazon EC2 & AMD EPYC - DEM01-R - Santa...
[REPEAT] Optimize your workloads with Amazon EC2 & AMD EPYC - DEM01-R - Santa...[REPEAT] Optimize your workloads with Amazon EC2 & AMD EPYC - DEM01-R - Santa...
[REPEAT] Optimize your workloads with Amazon EC2 & AMD EPYC - DEM01-R - Santa...
 

Similar to Automatic-Labelling-and-Model-Tuning-with-Amazon-SageMaker

Similar to Automatic-Labelling-and-Model-Tuning-with-Amazon-SageMaker (20)

Automatic Labelling and Model Tuning with Amazon SageMaker - AWS Summit Sydney
Automatic Labelling and Model Tuning with Amazon SageMaker - AWS Summit SydneyAutomatic Labelling and Model Tuning with Amazon SageMaker - AWS Summit Sydney
Automatic Labelling and Model Tuning with Amazon SageMaker - AWS Summit Sydney
 
Build Accurate Training Datasets with Amazon SageMaker Ground Truth - AIM302 ...
Build Accurate Training Datasets with Amazon SageMaker Ground Truth - AIM302 ...Build Accurate Training Datasets with Amazon SageMaker Ground Truth - AIM302 ...
Build Accurate Training Datasets with Amazon SageMaker Ground Truth - AIM302 ...
 
Build accurate training datasets with Amazon SageMaker Ground Truth - AIM205 ...
Build accurate training datasets with Amazon SageMaker Ground Truth - AIM205 ...Build accurate training datasets with Amazon SageMaker Ground Truth - AIM205 ...
Build accurate training datasets with Amazon SageMaker Ground Truth - AIM205 ...
 
Build accurate training datasets with Amazon SageMaker Ground Truth - AIM301 ...
Build accurate training datasets with Amazon SageMaker Ground Truth - AIM301 ...Build accurate training datasets with Amazon SageMaker Ground Truth - AIM301 ...
Build accurate training datasets with Amazon SageMaker Ground Truth - AIM301 ...
 
Rendi le tue app più smart con i servizi AI di AWS
Rendi le tue app più smart con i servizi AI di AWSRendi le tue app più smart con i servizi AI di AWS
Rendi le tue app più smart con i servizi AI di AWS
 
Build Highly Accurate Training Datasets Using Amazon SageMaker Ground Truth
Build Highly Accurate Training Datasets Using Amazon SageMaker Ground TruthBuild Highly Accurate Training Datasets Using Amazon SageMaker Ground Truth
Build Highly Accurate Training Datasets Using Amazon SageMaker Ground Truth
 
Drive digital transformation with AI
Drive digital transformation with AIDrive digital transformation with AI
Drive digital transformation with AI
 
Machine Learning Analytics for the rest of us
Machine Learning Analytics for the rest of usMachine Learning Analytics for the rest of us
Machine Learning Analytics for the rest of us
 
Predicting Demand In A Diverse Retail Environment - AWS Summit Sydney
Predicting Demand In A Diverse Retail Environment - AWS Summit SydneyPredicting Demand In A Diverse Retail Environment - AWS Summit Sydney
Predicting Demand In A Diverse Retail Environment - AWS Summit Sydney
 
Amazon SageMaker - ML for every developer & data scientist ft. Workday - AIM2...
Amazon SageMaker - ML for every developer & data scientist ft. Workday - AIM2...Amazon SageMaker - ML for every developer & data scientist ft. Workday - AIM2...
Amazon SageMaker - ML for every developer & data scientist ft. Workday - AIM2...
 
Optimize deep learning training and inferencing using GPU and Amazon SageMake...
Optimize deep learning training and inferencing using GPU and Amazon SageMake...Optimize deep learning training and inferencing using GPU and Amazon SageMake...
Optimize deep learning training and inferencing using GPU and Amazon SageMake...
 
Building intelligent applications using AI services
Building intelligent applications using AI servicesBuilding intelligent applications using AI services
Building intelligent applications using AI services
 
Machine Learning: From Inception to Inference - AWS Summit Sydney
Machine Learning: From Inception to Inference - AWS Summit SydneyMachine Learning: From Inception to Inference - AWS Summit Sydney
Machine Learning: From Inception to Inference - AWS Summit Sydney
 
AIOps - Steps Towards Autonomous Operations - AWS Summit Sydney 2019
AIOps - Steps Towards Autonomous Operations - AWS Summit Sydney 2019AIOps - Steps Towards Autonomous Operations - AWS Summit Sydney 2019
AIOps - Steps Towards Autonomous Operations - AWS Summit Sydney 2019
 
완전 관리형 ML 서비스인 Amazon SageMaker 의 신규 기능 소개 - 김필호 AI/ML 스페셜리스트 솔루션즈 아키텍트, AWS ...
완전 관리형 ML 서비스인 Amazon SageMaker 의 신규 기능 소개 - 김필호 AI/ML 스페셜리스트 솔루션즈 아키텍트, AWS ...완전 관리형 ML 서비스인 Amazon SageMaker 의 신규 기능 소개 - 김필호 AI/ML 스페셜리스트 솔루션즈 아키텍트, AWS ...
완전 관리형 ML 서비스인 Amazon SageMaker 의 신규 기능 소개 - 김필호 AI/ML 스페셜리스트 솔루션즈 아키텍트, AWS ...
 
Machine learning for developers & data scientists with Amazon SageMaker - AIM...
Machine learning for developers & data scientists with Amazon SageMaker - AIM...Machine learning for developers & data scientists with Amazon SageMaker - AIM...
Machine learning for developers & data scientists with Amazon SageMaker - AIM...
 
Build, train and deploy Machine Learning models on Amazon SageMaker (May 2019)
Build, train and deploy Machine Learning models on Amazon SageMaker (May 2019)Build, train and deploy Machine Learning models on Amazon SageMaker (May 2019)
Build, train and deploy Machine Learning models on Amazon SageMaker (May 2019)
 
Build, train and deploy machine learning models at scale using AWS
Build, train and deploy machine learning models at scale using AWSBuild, train and deploy machine learning models at scale using AWS
Build, train and deploy machine learning models at scale using AWS
 
Automate Security Event Management Using Trust-Based Decision Models - AWS Su...
Automate Security Event Management Using Trust-Based Decision Models - AWS Su...Automate Security Event Management Using Trust-Based Decision Models - AWS Su...
Automate Security Event Management Using Trust-Based Decision Models - AWS Su...
 
Build Machine Learning Models with Amazon SageMaker (April 2019)
Build Machine Learning Models with Amazon SageMaker (April 2019)Build Machine Learning Models with Amazon SageMaker (April 2019)
Build Machine Learning Models with Amazon SageMaker (April 2019)
 

More from Amazon Web Services

Tools for building your MVP on AWS
Tools for building your MVP on AWSTools for building your MVP on AWS
Tools for building your MVP on AWS
Amazon Web Services
 
How to Build a Winning Pitch Deck
How to Build a Winning Pitch DeckHow to Build a Winning Pitch Deck
How to Build a Winning Pitch Deck
Amazon Web Services
 
Building a web application without servers
Building a web application without serversBuilding a web application without servers
Building a web application without servers
Amazon Web Services
 
AWS_HK_StartupDay_Building Interactive websites while automating for efficien...
AWS_HK_StartupDay_Building Interactive websites while automating for efficien...AWS_HK_StartupDay_Building Interactive websites while automating for efficien...
AWS_HK_StartupDay_Building Interactive websites while automating for efficien...
Amazon Web Services
 

More from Amazon Web Services (20)

Come costruire servizi di Forecasting sfruttando algoritmi di ML e deep learn...
Come costruire servizi di Forecasting sfruttando algoritmi di ML e deep learn...Come costruire servizi di Forecasting sfruttando algoritmi di ML e deep learn...
Come costruire servizi di Forecasting sfruttando algoritmi di ML e deep learn...
 
Big Data per le Startup: come creare applicazioni Big Data in modalità Server...
Big Data per le Startup: come creare applicazioni Big Data in modalità Server...Big Data per le Startup: come creare applicazioni Big Data in modalità Server...
Big Data per le Startup: come creare applicazioni Big Data in modalità Server...
 
Esegui pod serverless con Amazon EKS e AWS Fargate
Esegui pod serverless con Amazon EKS e AWS FargateEsegui pod serverless con Amazon EKS e AWS Fargate
Esegui pod serverless con Amazon EKS e AWS Fargate
 
Costruire Applicazioni Moderne con AWS
Costruire Applicazioni Moderne con AWSCostruire Applicazioni Moderne con AWS
Costruire Applicazioni Moderne con AWS
 
Come spendere fino al 90% in meno con i container e le istanze spot
Come spendere fino al 90% in meno con i container e le istanze spot Come spendere fino al 90% in meno con i container e le istanze spot
Come spendere fino al 90% in meno con i container e le istanze spot
 
Open banking as a service
Open banking as a serviceOpen banking as a service
Open banking as a service
 
Rendi unica l’offerta della tua startup sul mercato con i servizi Machine Lea...
Rendi unica l’offerta della tua startup sul mercato con i servizi Machine Lea...Rendi unica l’offerta della tua startup sul mercato con i servizi Machine Lea...
Rendi unica l’offerta della tua startup sul mercato con i servizi Machine Lea...
 
OpsWorks Configuration Management: automatizza la gestione e i deployment del...
OpsWorks Configuration Management: automatizza la gestione e i deployment del...OpsWorks Configuration Management: automatizza la gestione e i deployment del...
OpsWorks Configuration Management: automatizza la gestione e i deployment del...
 
Microsoft Active Directory su AWS per supportare i tuoi Windows Workloads
Microsoft Active Directory su AWS per supportare i tuoi Windows WorkloadsMicrosoft Active Directory su AWS per supportare i tuoi Windows Workloads
Microsoft Active Directory su AWS per supportare i tuoi Windows Workloads
 
Computer Vision con AWS
Computer Vision con AWSComputer Vision con AWS
Computer Vision con AWS
 
Database Oracle e VMware Cloud on AWS i miti da sfatare
Database Oracle e VMware Cloud on AWS i miti da sfatareDatabase Oracle e VMware Cloud on AWS i miti da sfatare
Database Oracle e VMware Cloud on AWS i miti da sfatare
 
Crea la tua prima serverless ledger-based app con QLDB e NodeJS
Crea la tua prima serverless ledger-based app con QLDB e NodeJSCrea la tua prima serverless ledger-based app con QLDB e NodeJS
Crea la tua prima serverless ledger-based app con QLDB e NodeJS
 
API moderne real-time per applicazioni mobili e web
API moderne real-time per applicazioni mobili e webAPI moderne real-time per applicazioni mobili e web
API moderne real-time per applicazioni mobili e web
 
Database Oracle e VMware Cloud™ on AWS: i miti da sfatare
Database Oracle e VMware Cloud™ on AWS: i miti da sfatareDatabase Oracle e VMware Cloud™ on AWS: i miti da sfatare
Database Oracle e VMware Cloud™ on AWS: i miti da sfatare
 
Tools for building your MVP on AWS
Tools for building your MVP on AWSTools for building your MVP on AWS
Tools for building your MVP on AWS
 
How to Build a Winning Pitch Deck
How to Build a Winning Pitch DeckHow to Build a Winning Pitch Deck
How to Build a Winning Pitch Deck
 
Building a web application without servers
Building a web application without serversBuilding a web application without servers
Building a web application without servers
 
Fundraising Essentials
Fundraising EssentialsFundraising Essentials
Fundraising Essentials
 
AWS_HK_StartupDay_Building Interactive websites while automating for efficien...
AWS_HK_StartupDay_Building Interactive websites while automating for efficien...AWS_HK_StartupDay_Building Interactive websites while automating for efficien...
AWS_HK_StartupDay_Building Interactive websites while automating for efficien...
 
Introduzione a Amazon Elastic Container Service
Introduzione a Amazon Elastic Container ServiceIntroduzione a Amazon Elastic Container Service
Introduzione a Amazon Elastic Container Service
 

Automatic-Labelling-and-Model-Tuning-with-Amazon-SageMaker

  • 1. S U M M I T Taipei
  • 2. S U M M I T © 2019, Amazon Web Services, Inc. or its affiliates. All rights reserved. https://bit.ly/2MxHNLB
  • 3. © 2019, Amazon Web Services, Inc. or its affiliates. All rights reserved.S U M M I T Automatic Labelling and Model Tuning with Amazon SageMaker Adam Lynch Partner Solutions Architect Amazon Web Services S e s s i o n I D
  • 4. © 2019, Amazon Web Services, Inc. or its affiliates. All rights reserved.S U M M I T Agenda Introduction – 15 min Overview of Amazon SageMaker GroundTruth – 15 min Lab Automatic Labelling with Amazon SageMaker GroundTruth – 45 min Working Lunch – 15 min Lab Private Workforce Labelling – 30 min Overview of Model Tuning using Bayesian Optimisation – 15 min Lab Automatic Model Tuning with Amazon SageMaker – 45 min
  • 5. © 2019, Amazon Web Services, Inc. or its affiliates. All rights reserved.S U M M I T AWS Account for the Labs In order to complete this workshop you'll need an AWS Account with admin access There are resources required by this workshop that are eligible for the AWS free tier if your account is less than 12 months old. And we will supply some credits for other tasks.
  • 6. S U M M I T © 2019, Amazon Web Services, Inc. or its affiliates. All rights reserved.
  • 7. © 2019, Amazon Web Services, Inc. or its affiliates. All rights reserved.S U M M I T What data science tasks can be automated?
  • 8. © 2019, Amazon Web Services, Inc. or its affiliates. All rights reserved.S U M M I T Data science tasks
  • 9. © 2019, Amazon Web Services, Inc. or its affiliates. All rights reserved.S U M M I T AutoML – definition and goals AutoML aims to maximise the performace of machine learning programs without human assistance and subject to a computational budget. Core goals: a) Good performance: good generalization performance across various input data and learning tasks can be achieved. b) No assistance from humans: configurations can be automatically done for machine learning tools. c) High computational efficiency: the program can return a reasonable output within a limited budget. Taking the Human out of Learning Applications: A Survey on Automated Machine Learning Quanming Yao et al. arXiv: 1810.13306v3 [cs.AI] 17 Jan 2019
  • 10. © 2019, Amazon Web Services, Inc. or its affiliates. All rights reserved.S U M M I T The challenges of AutoML adoption Deep Learning Human Design Computational Budget
  • 11. © 2019, Amazon Web Services, Inc. or its affiliates. All rights reserved.S U M M I T Some tough challenges in machine learning Availability of Labelled Data The Road Not Travelled Drift
  • 12. © 2019, Amazon Web Services, Inc. or its affiliates. All rights reserved.S U M M I T Promising approaches Learning to Learn Simulation Reinforcement Learning
  • 13. © 2019, Amazon Web Services, Inc. or its affiliates. All rights reserved.S U M M I T KERAS AI SERVICES ML SERVICES ML FRAMEWORKS + INFRASTRUCTURE C5 C5n P3 P3dn Vision Speech Languages Chatbots Vertical AMAZON SAGEMAKER AMAZON SAGEMAKER GROUND TRUTH AMAZON SAGEMAKER RL AWS MARKETPLACE FOR ML AMAZON SAGEMAKER NEO AWS DEEPRACER AWS DEEPLENS AMAZON ELASTIC INFERENCE AWS INFERENTIA AWS GREENGRASS AMAZON PERSONALIZE AMAZON FORECAST AMAZON TEXTRACT AMAZON REKOGNITION AMAZON LEX AMAZON POLLY AMAZON TRANSCRIBE AMAZON COMPREHEND AMAZON TRANSLATE N E W N E W N E W
  • 14. S U M M I T © 2019, Amazon Web Services, Inc. or its affiliates. All rights reserved.
  • 15. © 2019, Amazon Web Services, Inc. or its affiliates. All rights reserved.S U M M I T Data labeling and machine learning Labeled data Model training Prepare and label data Inference
  • 16. © 2019, Amazon Web Services, Inc. or its affiliates. All rights reserved.S U M M I T Data labeling is hard… Need to label large datasets Requires humans to perform labeling Becomes time consuming and costly Difficult to achieve high accuracy for labels
  • 17. © 2019, Amazon Web Services, Inc. or its affiliates. All rights reserved.S U M M I T Amazon SageMaker: Build, train, and deploy ML 1 2 3 1 2 3
  • 18. © 2019, Amazon Web Services, Inc. or its affiliates. All rights reserved.S U M M I T K E Y F E A T U R E S Automatic labeling via machine learning Ready-made and custom workflows Label management Private and public human workforce Amazon SageMaker Ground Truth Label machine learning training data easily and accurately
  • 19. © 2019, Amazon Web Services, Inc. or its affiliates. All rights reserved.S U M M I T Amazon SageMaker Ground Truth How it works
  • 20. © 2019, Amazon Web Services, Inc. or its affiliates. All rights reserved.S U M M I T Amazon SageMaker Ground Truth How it works
  • 21. © 2019, Amazon Web Services, Inc. or its affiliates. All rights reserved.S U M M I T Amazon SageMaker Ground Truth How it works
  • 22. © 2019, Amazon Web Services, Inc. or its affiliates. All rights reserved.S U M M I T Amazon SageMaker Ground Truth How it works
  • 23. © 2019, Amazon Web Services, Inc. or its affiliates. All rights reserved.S U M M I T Amazon SageMaker Ground Truth How it works
  • 24. © 2019, Amazon Web Services, Inc. or its affiliates. All rights reserved.S U M M I T Step 1: Create a Ground Truth labeling job
  • 25. © 2019, Amazon Web Services, Inc. or its affiliates. All rights reserved.S U M M I T Step 2: Provide details for a labeling job
  • 26. © 2019, Amazon Web Services, Inc. or its affiliates. All rights reserved.S U M M I T Step 2: Provide details for a labeling job
  • 27. © 2019, Amazon Web Services, Inc. or its affiliates. All rights reserved.S U M M I T Step 2: Provide details for a labeling job
  • 28. © 2019, Amazon Web Services, Inc. or its affiliates. All rights reserved.S U M M I T Labeling job is now running
  • 29. © 2019, Amazon Web Services, Inc. or its affiliates. All rights reserved.S U M M I T Labeling app: Human workers label the images
  • 30. © 2019, Amazon Web Services, Inc. or its affiliates. All rights reserved.S U M M I T AWS Management Console: View labels for images
  • 31. © 2019, Amazon Web Services, Inc. or its affiliates. All rights reserved.S U M M I T AWS Management Console: View labels
  • 32. S U M M I T © 2019, Amazon Web Services, Inc. or its affiliates. All rights reserved.
  • 33. S U M M I T © 2019, Amazon Web Services, Inc. or its affiliates. All rights reserved.
  • 34. S U M M I T © 2019, Amazon Web Services, Inc. or its affiliates. All rights reserved.
  • 35. S U M M I T © 2019, Amazon Web Services, Inc. or its affiliates. All rights reserved.
  • 36. © 2019, Amazon Web Services, Inc. or its affiliates. All rights reserved.S U M M I T Solving the following problem: 𝑚𝑎𝑥 𝑥 ∈ 𝐴 𝑓(𝑥)
  • 37.
  • 38. S U M M I T © 2019, Amazon Web Services, Inc. or its affiliates. All rights reserved. Defining ‘A’ using Amazon SageMaker "ParameterRanges": { "CategoricalParameterRanges": [ { "Name": "tree_method", "Values": ["auto", "exact", "approx", "hist"]} ], "ContinuousParameterRanges": [ { "Name": "eta", "MaxValue" : "0.5", "MinValue": "0" } ], "IntegerParameterRanges": [ { "Name": "max_depth", "MaxValue": "10", "MinValue": "1", } ] }
  • 39. S U M M I T © 2019, Amazon Web Services, Inc. or its affiliates. All rights reserved. On Amazon Sagemaker – configure the tuning job tuning_job_config = { "ParameterRanges": {…} "ResourceLimits": { "MaxNumberOfTrainingJobs": 20, "MaxParallelTrainingJobs": 3 }, "Strategy": "Bayesian", "HyperParameterTuningJobObjective": { "MetricName": "validation:auc", "Type": "Maximize" } }
  • 40. S U M M I T © 2019, Amazon Web Services, Inc. or its affiliates. All rights reserved. Launch the tuning job tuning_job = smclient.create_hyper_parameter_tuning_job( HyperParameterTuningJobName = "MyTuningJob“, HyperParameterTuningJobConfig = tuning_job_config, TrainingJobDefinition = training_job_definition)
  • 41. S U M M I T © 2019, Amazon Web Services, Inc. or its affiliates. All rights reserved.
  • 42. Thank you! S U M M I T © 2019, Amazon Web Services, Inc. or its affiliates. All rights reserved. Adam Lynch atlynch@amazon.com