SlideShare a Scribd company logo
1 of 62
Download to read offline
© 2018, Amazon Web Services, Inc. or its affiliates. All rights reserved.
Machine Learning at the IoT Edge
David Nunnerley
AWS Senior Manager
AWS IoT Greengrass
I O T 2 1 4
Nobutaka Nakazawa
CTO
Brains Technology, Inc.
Masanori Sato
Group Manager
Aisin AW LTD
© 2018, Amazon Web Services, Inc. or its affiliates. All rights reserved.
Agenda
Why Machine Learning (ML) at the Edge?
AWS IoT Greengrass overview
ML Inference at the Edge with AWS IoT Greengrass
New AWS IoT Greengrass ML capabilities
Customer use case: Aisin AW (Masanori Sato)
Brains Technology (Nobutaka Nakazawa)
© 2018, Amazon Web Services, Inc. or its affiliates. All rights reserved.
© 2018, Amazon Web Services, Inc. or its affiliates. All rights reserved.
Medical equipment Industrial machinery Extreme environments
Most machine data never reaches the cloud
© 2018, Amazon Web Services, Inc. or its affiliates. All rights reserved.
Why this problem isn’t going away
Law of physics Law of economics Law of the land
© 2018, Amazon Web Services, Inc. or its affiliates. All rights reserved.
© 2018, Amazon Web Services, Inc. or its affiliates. All rights reserved.
AWS IoT Greengrass
© 2018, Amazon Web Services, Inc. or its affiliates. All rights reserved.
AWS IoT Greengrass
All AWS Cloud services
e.g., Amazon S3,
Amazon Kinesis,
Amazon Redshift…
AWS IoT services
e.g., AWS IoT Core,
AWS IoT Analytics,
AWS IoT Device
Defender…
© 2018, Amazon Web Services, Inc. or its affiliates. All rights reserved.
Data and
State Sync Security
Over the
Air UpdatesConnectors
Local
Device Shadows
Code
Deployment
Lambda
Functions
AWS-grade
security
Easily Update
Greengrass Core
Machine
Learning
Inference
Local Execution
of ML Models
Local
Resource
Access
Lambdas Interact
With Peripherals
Easy integrations
with AWS
services, protocol
adaptors and
other SaaS
providers
Local
Messages
and Triggers
Local
Message Broker
Manage
Secrets at
the edge
AWS Secrets
Manager
functionality
at edge
AWS Greengrass
Extend AWS IoT to the Edge
© 2018, Amazon Web Services, Inc. or its affiliates. All rights reserved.
Data and
State Sync Security
Over the
Air UpdatesConnectors
Local
Device Shadows
Code
Deployment
Lambda
Functions
AWS-grade
security
Easily Update
Greengrass Core
Machine
Learning
Inference
Local Execution
of ML Models
Local
Resource
Access
Lambdas Interact
With Peripherals
Easy integrations
with AWS
services, protocol
adaptors and
other SaaS
providers
Local
Messages
and Triggers
Local
Message Broker
Manage
Secrets at
the edge
AWS Secrets
Manager
functionality
at edge
AWS Greengrass
Extend AWS IoT to the Edge
© 2018, Amazon Web Services, Inc. or its affiliates. All rights reserved.
Data and
State Sync Security
Over the
Air UpdatesConnectors
Local
Device Shadows
Code
Deployment
Lambda
Functions
AWS-grade
security
Easily Update
Greengrass Core
Machine
Learning
Inference
Local Execution
of ML Models
Local
Resource
Access
Lambdas Interact
With Peripherals
Easy integrations
with AWS
services, protocol
adaptors and
other SaaS
providers
Local
Messages
and Triggers
Local
Message Broker
Manage
Secrets at
the edge
AWS Secrets
Manager
functionality
at edge
AWS Greengrass
Extend AWS IoT to the Edge
© 2018, Amazon Web Services, Inc. or its affiliates. All rights reserved.
Data and
State Sync Security
Over the
Air UpdatesConnectors
Local
Device Shadows
Code
Deployment
Lambda
Functions
AWS-grade
security
Easily Update
Greengrass Core
Machine
Learning
Inference
Local Execution
of ML Models
Local
Resource
Access
Lambdas Interact
With Peripherals
Easy integrations
with AWS
services, protocol
adaptors and
other SaaS
providers
Local
Messages
and Triggers
Local
Message Broker
Manage
Secrets at
the edge
AWS Secrets
Manager
functionality
at edge
AWS Greengrass
Extend AWS IoT to the Edge
© 2018, Amazon Web Services, Inc. or its affiliates. All rights reserved.
Data and
State Sync Security
Over the
Air UpdatesConnectors
Local
Device Shadows
Code
Deployment
Lambda
Functions
AWS-grade
security
Easily Update
Greengrass Core
Machine
Learning
Inference
Local Execution
of ML Models
Local
Resource
Access
Lambdas Interact
With Peripherals
Easy integrations
with AWS
services, protocol
adaptors and
other SaaS
providers
Local
Messages
and Triggers
Local
Message Broker
Manage
Secrets at
the edge
AWS Secrets
Manager
functionality
at edge
AWS Greengrass
Extend AWS IoT to the Edge
© 2018, Amazon Web Services, Inc. or its affiliates. All rights reserved.
Data and
State Sync Security
Over the
Air UpdatesConnectors
Local
Device Shadows
Code
Deployment
Lambda
Functions
AWS-grade
security
Easily Update
Greengrass Core
Machine
Learning
Inference
Local Execution
of ML Models
Local
Resource
Access
Lambdas Interact
With Peripherals
Easy integrations
with AWS
services, protocol
adaptors and
other SaaS
providers
Local
Messages
and Triggers
Local
Message Broker
Manage
Secrets at
the edge
AWS Secrets
Manager
functionality
at edge
AWS Greengrass
Extend AWS IoT to the Edge
© 2018, Amazon Web Services, Inc. or its affiliates. All rights reserved.
Data and
State Sync Security
Over the
Air UpdatesConnectors
Local
Device Shadows
Code
Deployment
Lambda
Functions
AWS-grade
security
Easily Update
Greengrass Core
Machine
Learning
Inference
Local Execution
of ML Models
Local
Resource
Access
Lambdas Interact
With Peripherals
Easy integrations
with AWS
services, protocol
adaptors and
other SaaS
providers
Local
Messages
and Triggers
Local
Message Broker
Manage
Secrets at
the edge
AWS Secrets
Manager
functionality
at edge
AWS Greengrass
Extend AWS IoT to the Edge
© 2018, Amazon Web Services, Inc. or its affiliates. All rights reserved.
Data and
State Sync Security
Over the
Air UpdatesConnectors
Local
Device Shadows
Code
Deployment
Lambda
Functions
AWS-grade
security
Easily Update
Greengrass Core
Machine
Learning
Inference
Local Execution
of ML Models
Local
Resource
Access
Lambdas Interact
With Peripherals
Easy integrations
with AWS
services, protocol
adaptors and
other SaaS
providers
Local
Messages
and Triggers
Local
Message Broker
Manage
Secrets at
the edge
AWS Secrets
Manager
functionality
at edge
AWS Greengrass
Extend AWS IoT to the Edge
© 2018, Amazon Web Services, Inc. or its affiliates. All rights reserved.
Data and
State Sync Security
Over the
Air UpdatesConnectors
Local
Device Shadows
Code
Deployment
Lambda
Functions
AWS-grade
security
Easily Update
Greengrass Core
Machine
Learning
Inference
Local Execution
of ML Models
Local
Resource
Access
Lambdas Interact
With Peripherals
Easy integrations
with AWS
services, protocol
adaptors and
other SaaS
providers
Local
Messages
and Triggers
Local
Message Broker
Manage
Secrets at
the edge
AWS Secrets
Manager
functionality
at edge
AWS Greengrass
Extend AWS IoT to the Edge
© 2018, Amazon Web Services, Inc. or its affiliates. All rights reserved.
Data and
State Sync Security
Over the
Air UpdatesConnectors
Local
Device Shadows
Code
Deployment
Lambda
Functions
AWS-grade
security
Easily Update
Greengrass Core
Machine
Learning
Inference
Local Execution
of ML Models
Local
Resource
Access
Lambdas Interact
With Peripherals
Easy integrations
with AWS
services, protocol
adaptors and
other SaaS
providers
Local
Messages
and Triggers
Local
Message Broker
Manage
Secrets at
the edge
AWS Secrets
Manager
functionality
at edge
AWS Greengrass
Extend AWS IoT to the Edge
© 2018, Amazon Web Services, Inc. or its affiliates. All rights reserved.
© 2018, Amazon Web Services, Inc. or its affiliates. All rights reserved.
Inference Training
Machine Learning at the Edge
Local
actions
Edge Cloud
© 2018, Amazon Web Services, Inc. or its affiliates. All rights reserved.
Inference Training
Machine Learning at the Edge
Local
actions
Edge Cloud
© 2018, Amazon Web Services, Inc. or its affiliates. All rights reserved.
Inference Training
Machine Learning at the Edge
Local
actions
Edge Cloud
© 2018, Amazon Web Services, Inc. or its affiliates. All rights reserved.
Inference Training
Machine Learning at the Edge
Local
actions
Edge Cloud
© 2018, Amazon Web Services, Inc. or its affiliates. All rights reserved.
Inference Training
Machine Learning at the Edge
Local
actions
Edge Cloud
© 2018, Amazon Web Services, Inc. or its affiliates. All rights reserved.
Inference Training
Machine Learning at the Edge
Local
actions
Edge Cloud
© 2018, Amazon Web Services, Inc. or its affiliates. All rights reserved.
AWS IoT Greengrass Core Machine Learning
Run ML Inference on Greengrass
Deploy an ML model from Amazon SageMaker in the cloud to a target AWS Greengrass core
device using the Greengrass console or Command Line Interface (AWS CLI)
Install the necessary run-time for the model e.g., (TensorFlow, Apache MXNet,
Chainer…) on the AWS Greengrass core
Available for multiple hardware architectures:
e.g., Intel x86-64, ARM v7 and Nvidia Jetson TX2
Code your Lambda to read from attached device/sensor (optionally from MQTT topic) and
pass to the Lambda running the ML model. Take action based upon the inference.
© 2018, Amazon Web Services, Inc. or its affiliates. All rights reserved.
AWS IoT Greengrass Core 1.7 Machine Learning
New Machine Learning capabilities
Image Classification Connector (available for download from console)
Pre-built Lambda to run the Image classification ML model
Easy coding to bridge from input device to the supplied Lambda running the
inference
Image Classification Model can be trained to learn new image classifications in the
cloud with Amazon SageMaker
© 2018, Amazon Web Services, Inc. or its affiliates. All rights reserved.
Code to use the Image Classification Connector
import greengrass_machine_learning_sdk as ml
with open('/test_img/test.jpg', 'rb') as f:
content = f.read()
def infer():
logging.info('invoking Greengrass ML Inference service')
try:
resp = client.invoke_inference_service(
AlgoType='image-classification',
ServiceName='imageClassification',
ContentType='image/jpeg',
Body=content
)
except ml.GreengrassInferenceException as e:
logging.info('inference exception {}("{}")'.format(e.__class__.__name__, e))
return
except ml.GreengrassDependencyException as e:
logging.info('dependency exception {}("{}")'.format(e.__class__.__name__, e))
return
logging.info('resp: {}'.format(resp))
predictions = resp['Body'].read()
logging.info('predictions: {}'.format(predictions))
© 2018, Amazon Web Services, Inc. or its affiliates. All rights reserved.
Code to use the Image Classification Connector
import greengrass_machine_learning_sdk as ml
with open('/test_img/test.jpg', 'rb') as f:
content = f.read()
def infer():
logging.info('invoking Greengrass ML Inference service')
try:
resp = client.invoke_inference_service(
AlgoType='image-classification',
ServiceName='imageClassification',
ContentType='image/jpeg',
Body=content
)
except ml.GreengrassInferenceException as e:
logging.info('inference exception {}("{}")'.format(e.__class__.__name__, e))
return
except ml.GreengrassDependencyException as e:
logging.info('dependency exception {}("{}")'.format(e.__class__.__name__, e))
return
logging.info('resp: {}'.format(resp))
predictions = resp['Body'].read()
logging.info('predictions: {}'.format(predictions))
© 2018, Amazon Web Services, Inc. or its affiliates. All rights reserved.
Code to use the Image Classification Connector
import greengrass_machine_learning_sdk as ml
with open('/test_img/test.jpg', 'rb') as f:
content = f.read()
def infer():
logging.info('invoking Greengrass ML Inference service')
try:
resp = client.invoke_inference_service(
AlgoType='image-classification',
ServiceName='imageClassification',
ContentType='image/jpeg',
Body=content
)
except ml.GreengrassInferenceException as e:
logging.info('inference exception {}("{}")'.format(e.__class__.__name__, e))
return
except ml.GreengrassDependencyException as e:
logging.info('dependency exception {}("{}")'.format(e.__class__.__name__, e))
return
logging.info('resp: {}'.format(resp))
predictions = resp['Body'].read()
logging.info('predictions: {}'.format(predictions))
© 2018, Amazon Web Services, Inc. or its affiliates. All rights reserved.
AWS IoT Greengrass Core 1.7 Machine Learning
Other New Machine Learning capabilities
Greengrass support for the new Amazon SageMaker Neo (Deep Learning Runtime)
Optimize the model using Neo compiler in the cloud
More performant without loss of accuracy
Smaller memory footprint
Deploy optimized Neo model to the Greengrass core device
Install Neo run-time to the device
Write a Lambda to run the Neo optimized ML model
Smart Vending Machine -- R&D Innovation Team,
from the SA Americas organization
© 2018, Amazon Web Services, Inc. or its affiliates. All rights reserved.
Camera
Smart vending machine
Object Detection and
Image Classification
models
Load Sensors readings
in local time series
database
Sensor fusion functions
to detect removed items
and strange objects
Thanks for shopping!
3x Water Bottles
USD 1.50 each
Your total is $4.50
© 2018, Amazon Web Services, Inc. or its affiliates. All rights reserved.
Our Launch Partners
“The addition of AWS IoT Greengrass with its latest ML Inference update running on ADLINK’s industrial vision systems makes for truly plug-
and-play IoT. Now when we power-on an off-the-shelf ADLINK NEON smart camera running AWS IoT Greengrass with its latest ML Inference
update, we can get to high-quality outcomes much, much faster. This allows us to further speed development of our IoT digital experiments
for our logistics, quality inspection, industrial robotics, and other manufacturing customers.”
- Elizabeth Campbell, General Manager, The Americas, ADLINK Technology
“The potential of computer vision use cases enabled by IoT and AI is vast for businesses to exponentially improve productivity and efficiency.
In this time of intelligent transformation, our premium industrial Think IoT cameras powered by AWS IoT Greengrass with the latest machine
learning upgrades are engineered to make a notable difference to enterprise customers.”
- Jon Pershke, Vice President of Strategy and Emerging Business, Intelligent Devices
“The pervasiveness of artificial intelligence and the pace of digital transformation continues to grow at an astonishing rate. Innovations
like the newest improvements to AWS IoT Greengrass Machine Learning that markedly decrease latency without decreasing the accuracy
of ML inference accelerate new solutions to emerging industrial automation use cases for object identification and classification. AWS’ new
machine learning solution integrated with Leopard Imaging’s AICam powered by NVIDIA® GPU will be a cornerstone in any edge to cloud
Industrial and Smart City solution.”
-Bill Pu, President and Co-Founder, Leopard Imaging
“Vieureka of Panasonic is very pleased to utilize the application evolving functions of AWS’s machine learning as enabled by AWS IoT
Greengrass. In order to offer Vieureka-Cameras and service management functions to all the partners of the AWS community, I would
like to develop a Greengrass compatible version as soon as possible. We will create the environment for developers in the spring of
2019, with commercial versions available in autumn of the same year.”
- Miyazaki, CEO of Vieureka Service, Panasonic
© 2018, Amazon Web Services, Inc. or its affiliates. All rights reserved.
Nobutaka Nakazawa
Brains Technology, Inc.
CTO
Masanori Sato
Aisin AW LTD
Group Manager
© 2018, Amazon Web Services, Inc. or its affiliates. All rights reserved.
Machine state monitoring by
cloud & edge computing
AISIN AW CO.,LTD.
Manufacturing Engineering Development
Production System Innovation Group
Masanori Sato
© 2018, Amazon Web Services, Inc. or its affiliates. All rights reserved.
Agenda
・ Company profile
・ About our production engineering
・ Action background
・ Action summary
© 2018, Amazon Web Services, Inc. or its affiliates. All rights reserved.
9,830,000
NAVI
Securing a Top Share of the Global Market with Our Innovative Manufacturing
World No.1
World No.2
AT
■ Business summary of 2017(in March, 2018)
Unit sales
AT: 9,830,000units
NAVI: 1,810,000units
Sales
amount
A connection:
1,621,200 million yen
AISIN AW CO.,LTD Company Profile
© 2018, Amazon Web Services, Inc. or its affiliates. All rights reserved.
Mission of manufacturing engineering
Workplace skills that bringing value: Production engineering
Production
◆ Finish of the SE ※ / drawing
※ Simultaneous Engineering
◆ Design of the product line / setup
◆ Design of facilities / production
◆ Development of the new production
engineering
◆ Plan, design of the factory
Ordering,
suggestion
Three-Pillar
Manufacturing
Suggestion
Suggestion
Suggestion
Production engineering
Product
Design
Trading
company
Equipment
manufacturer
Delivery of goods,
suggestion
Cooperate as a partner
Vender
© 2018, Amazon Web Services, Inc. or its affiliates. All rights reserved.
Background of the action
Way of thinking of Industrial IoT in AISIN AW
Man Machine Material Method
Human factor
from who involved
・ Assembling
・ Machine setting
・ ・・etc.
Factor hardware such as
machines
・ Blade tool
・ Metal mold
・ ・・
etc.
Factor from Materials
(property value)
・ Ductility, toughness
・ Hardness
・ ・・etc.
Factor from
production method
・ Processing method
・ Processing order
・ ・・etc.
Building a base of high level condition monitoring and control by using information technology
The production revolution by IT is proposed in the world
→ Need to develop the Industrial IoT production system for AISIN AW
© 2018, Amazon Web Services, Inc. or its affiliates. All rights reserved.
Background of the action
What we need for I-IoT
We have started to develop I-IoT system that utilizes "cloud & edge" that
can satisfy these requirements
• Small start
• Real-time detection
• Scalability
-Connectable with more than 20,000 machines
-Easy deployment to each factory
• Successful partner
-Quick and challenge
© 2018, Amazon Web Services, Inc. or its affiliates. All rights reserved.
Cloud & edge computing analysis base
◯◯◯ Factory
◯◯◯ Factory
Analysis
monitor
Edge device AWS Services
- Storage, Managed Services, etc
Cloud
Edge device
• Dashboard
• Simulation model making
• Algorithm development
• Edge device management
Factory ANotice
monitor
Real-time monitoring &
ML detection
Factory B
Analysis
monitor
Notice
monitor
Machines
Machines
© 2018, Amazon Web Services, Inc. or its affiliates. All rights reserved.
Cause inquire-able detection ML algorithm
Develop machine learning algorithm that person can understand a results and can improve immediately
If not If cause is clear
Data A
Machine
learning
Not good
Data A
Machine
learning
It’s not red enough,
and It is not ripe
What is ?
What part ?
I see!
I can take action
immediately
Not good
[No reason] [Reason]
© 2018, Amazon Web Services, Inc. or its affiliates. All rights reserved.
Cause inquire-able detection ML algorithm
Good point
・ Got good result in a month after using
→Because satisfaction of detection result, it‘s actively used
→ Leads to expansion
・The model can be constructed with high precision at an early stage
→Model making took three months ⇒ one week
Difficult point
・Since it’s new initiative, it will not be adopted unless it is indicated by the result
→Need one year temporary use for the use at the mass production line
© 2018, Amazon Web Services, Inc. or its affiliates. All rights reserved.
Outcomes of cloud & edge system in AISIN AW
Development of the I-IoT future
・ Further high precision monitoring by algorithm development
・ Expanding to other factories and processes and supervising management
・ Training of workers to increase IoT talent
Time[msec] Time[msec]
value
value
unusual point
The system detected “anomaly”
state and Suppress the cost of
long line stop
© 2018, Amazon Web Services, Inc. or its affiliates. All rights reserved.
Machine Learning at the IoT Edge
Nobutaka Nakazawa
CTO
Brains Technology, Inc.
I O T 2 1 4
© 2018, Amazon Web Services, Inc. or its affiliates. All rights reserved.
Agenda
Company profile
Overview of impulse
Algorithm
Summary
© 2018, Amazon Web Services, Inc. or its affiliates. All rights reserved.
© 2018, Amazon Web Services, Inc. or its affiliates. All rights reserved.
Company profile
Company name Brains Technology, Inc.
Founded August 8, 2008
CEO Sawako Hamanaka
Capital 110 million yen (Including capital reserve)
Address
Shinagawa Center Building 3-23-17, Takanawa, Minato-Ku, TOKYO,
Japan
URL https://www.brains-tech.co.jp
Provide innovative service and bring about technological innovations with open technology
Providing innovative service for business enterprises, improve the productivity of corporate
activities dramatically
Our Mission
© 2018, Amazon Web Services, Inc. or its affiliates. All rights reserved.
© 2018, Amazon Web Services, Inc. or its affiliates. All rights reserved.
Overview of Impulse
20
41
84
0
20
40
60
80
100
FY2015
(〜2016/7)
FY2016
(〜2017/7)
FY2017
(〜2018/7)
145+
Predictive Maintenance Quality Management
- Plant equipment(power, chemical, bio)
- Co-generation system
- Industrial machinery (robots)
- Construction machines (crane, elevator)
- Electrical equipment (air conditioners, water
heaters)
- Auto parts (transmissions, gears, drive shafts,
bumpers)
- Electrical equipment (LED)
- Chemical products
- Casting
Impulse is the IoT ML edge platform for the manufacturing industry
for any kind of time series data built on top of AWS.
https://impulse-cloud.com
© 2018, Amazon Web Services, Inc. or its affiliates. All rights reserved.
Architecture
File
Monitor
Output
result
Raw data
Anomaly
Detection
Post
Process
Factory
AWS IoT
Amazon S3
AWS Lambda Amazon
DynamoDB
AWS Batch
UI
Dashboard
Simulation
Line A
Output
result
Line B
Raw data
Edge PC
Greengrass Core
Thing
Thing
Amazon Athena
Amazon S3
Raw data
Model
Amazon SageMaker
© 2018, Amazon Web Services, Inc. or its affiliates. All rights reserved.
Deploy Models using ML Inference
Three modules to deploy
ML Library (MXNet, Tensorflow, scikit-learn)
AWS Greengrass ML Inference or Lambda
Your Code
Lambda
Model
AWS Greengrass ML Inference / S3
Steps to deploy
• Create AWS Lambda functions for ML
inference.
• Create Models by Amazon SageMaker or
AWS Batch and upload the models to
Amazon S3.
• AWS IoT fully manages the whole
deployment process.
Upload model files to S3
Setting up local resource in AWS IoT
© 2018, Amazon Web Services, Inc. or its affiliates. All rights reserved.
© 2018, Amazon Web Services, Inc. or its affiliates. All rights reserved.
Automated algorithm and parameter optimization
Analyzing the characteristics of the data and auto-selecting the optimum
algorithms and parameters
Anomaly
detection
models
Input
gaussian
periodic
correlation
independent
Mahalanobis
S-H-ESD
One Class SVM
Sparse Coding
Recommended
parameters
Breakout
LOF
Gaussian Process
Data
Characteristics
Algorithm Parameter
Output
Recommended
parameters
Recommended
parameters
Recommended
parameters
Recommended
parameters
Recommended
parameters
Recommended
parameters
© 2018, Amazon Web Services, Inc. or its affiliates. All rights reserved.
Sparse coding
• Sparse coding is a class of
unsupervised methods for
learning sets of over-complete
bases to represent data
efficiently
• Finds a sparse representation of
data against a fixed,
precomputed dictionary
• Works well for high-speed time-
series signals with periodic
pattern
Dictionary Leaning
Decoding from dictionary
© 2018, Amazon Web Services, Inc. or its affiliates. All rights reserved.
LOF (Local Outlier Factor)
• Calculate anomaly scores
considering the density and
distance of the surrounding data
• Works well for high-dimensional
correlated data with
dimensionality reduction
technique (PCA, GPLVM, etc.)
• No need to assume a distribution
and it can be applied even when
the density has multimodality
Dimensionality reduction
LOF anomaly detection
© 2018, Amazon Web Services, Inc. or its affiliates. All rights reserved.
© 2018, Amazon Web Services, Inc. or its affiliates. All rights reserved.
Summary
• Fully managed IoT ML edge platform
• Highly-scalable, easy to process ML leaning and model deployment cycle
• Deployment of the algorithms and the models from AWS platform can eliminate the need
to go on-site to update the algorithm or the ML model
• Lambda function with additional libraries (scikit-learn, numpy, pandas, etc.) can run any ML
logic you created. (unless exceeding Lambda size limitation)
• Some limitations still exist
• It is necessary to consider the fault tolerance at the edge
• Greengrass only runs on recent Linux environment
• Not all regions support AWS Greengrass yet
• Time series analysis needs data cache mechanism on the edge
Thank you!
© 2018, Amazon Web Services, Inc. or its affiliates. All rights reserved.
David Nunnerley
Masanori Sato
Nobutaka Nakazawa
© 2018, Amazon Web Services, Inc. or its affiliates. All rights reserved.

More Related Content

What's hot

AWS Snowball Edge and AWS Greengrass for Fun and Profit (STG388) - AWS re:Inv...
AWS Snowball Edge and AWS Greengrass for Fun and Profit (STG388) - AWS re:Inv...AWS Snowball Edge and AWS Greengrass for Fun and Profit (STG388) - AWS re:Inv...
AWS Snowball Edge and AWS Greengrass for Fun and Profit (STG388) - AWS re:Inv...Amazon Web Services
 
Hands-on in the AWS Java Ecosystem (DEV325-R1) - AWS re:Invent 2018
Hands-on in the AWS Java Ecosystem (DEV325-R1) - AWS re:Invent 2018Hands-on in the AWS Java Ecosystem (DEV325-R1) - AWS re:Invent 2018
Hands-on in the AWS Java Ecosystem (DEV325-R1) - AWS re:Invent 2018Amazon Web Services
 
IoT Microcontrollers and Getting Started with Amazon FreeRTOS (IOT338-R1) - A...
IoT Microcontrollers and Getting Started with Amazon FreeRTOS (IOT338-R1) - A...IoT Microcontrollers and Getting Started with Amazon FreeRTOS (IOT338-R1) - A...
IoT Microcontrollers and Getting Started with Amazon FreeRTOS (IOT338-R1) - A...Amazon Web Services
 
Monitoring Serverless Applications (SRV303-S) - AWS re:Invent 2018
Monitoring Serverless Applications (SRV303-S) - AWS re:Invent 2018Monitoring Serverless Applications (SRV303-S) - AWS re:Invent 2018
Monitoring Serverless Applications (SRV303-S) - AWS re:Invent 2018Amazon Web Services
 
Enterprise DevOps: Patterns of Efficiency (ENT311-R1) - AWS re:Invent 2018
Enterprise DevOps: Patterns of Efficiency (ENT311-R1) - AWS re:Invent 2018Enterprise DevOps: Patterns of Efficiency (ENT311-R1) - AWS re:Invent 2018
Enterprise DevOps: Patterns of Efficiency (ENT311-R1) - AWS re:Invent 2018Amazon Web Services
 
Data Lake Patterns for Voice, Vision, Advanced Analytics, & ML Using Serverle...
Data Lake Patterns for Voice, Vision, Advanced Analytics, & ML Using Serverle...Data Lake Patterns for Voice, Vision, Advanced Analytics, & ML Using Serverle...
Data Lake Patterns for Voice, Vision, Advanced Analytics, & ML Using Serverle...Amazon Web Services
 
Getting Started with AWS Greengrass (IOT215-R3) - AWS re:Invent 2018
Getting Started with AWS Greengrass (IOT215-R3) - AWS re:Invent 2018Getting Started with AWS Greengrass (IOT215-R3) - AWS re:Invent 2018
Getting Started with AWS Greengrass (IOT215-R3) - AWS re:Invent 2018Amazon Web Services
 
[NEW LAUNCH!] Scaling Tightly-coupled HPC workloads on HPC with Elastic Fabri...
[NEW LAUNCH!] Scaling Tightly-coupled HPC workloads on HPC with Elastic Fabri...[NEW LAUNCH!] Scaling Tightly-coupled HPC workloads on HPC with Elastic Fabri...
[NEW LAUNCH!] Scaling Tightly-coupled HPC workloads on HPC with Elastic Fabri...Amazon Web Services
 
Pause and Resume your EC2 Instances with Hibernate (CMP392) - AWS re:Invent 2018
Pause and Resume your EC2 Instances with Hibernate (CMP392) - AWS re:Invent 2018Pause and Resume your EC2 Instances with Hibernate (CMP392) - AWS re:Invent 2018
Pause and Resume your EC2 Instances with Hibernate (CMP392) - AWS re:Invent 2018Amazon Web Services
 
Making Hybrid Work for You: Getting into the Cloud Fast (GPSTEC308) - AWS re:...
Making Hybrid Work for You: Getting into the Cloud Fast (GPSTEC308) - AWS re:...Making Hybrid Work for You: Getting into the Cloud Fast (GPSTEC308) - AWS re:...
Making Hybrid Work for You: Getting into the Cloud Fast (GPSTEC308) - AWS re:...Amazon Web Services
 
Leadership Session: Using DevOps, Microservices, and Serverless to Accelerate...
Leadership Session: Using DevOps, Microservices, and Serverless to Accelerate...Leadership Session: Using DevOps, Microservices, and Serverless to Accelerate...
Leadership Session: Using DevOps, Microservices, and Serverless to Accelerate...Amazon Web Services
 
Operational Excellence with Containerized Workloads Using AWS Fargate (CON320...
Operational Excellence with Containerized Workloads Using AWS Fargate (CON320...Operational Excellence with Containerized Workloads Using AWS Fargate (CON320...
Operational Excellence with Containerized Workloads Using AWS Fargate (CON320...Amazon Web Services
 
AWS re:Invent 2018: Deep Dive: Hybrid Cloud Storage Arch. w/Storage Gateway, ...
AWS re:Invent 2018: Deep Dive: Hybrid Cloud Storage Arch. w/Storage Gateway, ...AWS re:Invent 2018: Deep Dive: Hybrid Cloud Storage Arch. w/Storage Gateway, ...
AWS re:Invent 2018: Deep Dive: Hybrid Cloud Storage Arch. w/Storage Gateway, ...Amazon Web Services
 
Continuous Compliance for Modern Application Pipelines (GPSWS402) - AWS re:In...
Continuous Compliance for Modern Application Pipelines (GPSWS402) - AWS re:In...Continuous Compliance for Modern Application Pipelines (GPSWS402) - AWS re:In...
Continuous Compliance for Modern Application Pipelines (GPSWS402) - AWS re:In...Amazon Web Services
 
Mythical Mysfits: Management and Ops with AWS Fargate (CON322-R1) - AWS re:In...
Mythical Mysfits: Management and Ops with AWS Fargate (CON322-R1) - AWS re:In...Mythical Mysfits: Management and Ops with AWS Fargate (CON322-R1) - AWS re:In...
Mythical Mysfits: Management and Ops with AWS Fargate (CON322-R1) - AWS re:In...Amazon Web Services
 
善用 GraphQL 與 AWS AppSync 讓您的 Progressive Web App (PWA) 加速進化 (Level 200)
善用  GraphQL 與 AWS AppSync 讓您的  Progressive Web App (PWA) 加速進化 (Level 200)善用  GraphQL 與 AWS AppSync 讓您的  Progressive Web App (PWA) 加速進化 (Level 200)
善用 GraphQL 與 AWS AppSync 讓您的 Progressive Web App (PWA) 加速進化 (Level 200)Amazon Web Services
 
AWS and Symantec: Cyber Defense at Scale (SEC311-S) - AWS re:Invent 2018
AWS and Symantec: Cyber Defense at Scale (SEC311-S) - AWS re:Invent 2018AWS and Symantec: Cyber Defense at Scale (SEC311-S) - AWS re:Invent 2018
AWS and Symantec: Cyber Defense at Scale (SEC311-S) - AWS re:Invent 2018Amazon Web Services
 
Deploying Your ONNX Deep Learning with Apache MXNet Model Server (AIM413) - A...
Deploying Your ONNX Deep Learning with Apache MXNet Model Server (AIM413) - A...Deploying Your ONNX Deep Learning with Apache MXNet Model Server (AIM413) - A...
Deploying Your ONNX Deep Learning with Apache MXNet Model Server (AIM413) - A...Amazon Web Services
 
Update Microcontroller Devices Over-the-Air with Amazon FreeRTOS (IOT304-R1) ...
Update Microcontroller Devices Over-the-Air with Amazon FreeRTOS (IOT304-R1) ...Update Microcontroller Devices Over-the-Air with Amazon FreeRTOS (IOT304-R1) ...
Update Microcontroller Devices Over-the-Air with Amazon FreeRTOS (IOT304-R1) ...Amazon Web Services
 
Day Two Operations of Kubernetes on AWS (GPSTEC309) - AWS re:Invent 2018
Day Two Operations of Kubernetes on AWS (GPSTEC309) - AWS re:Invent 2018Day Two Operations of Kubernetes on AWS (GPSTEC309) - AWS re:Invent 2018
Day Two Operations of Kubernetes on AWS (GPSTEC309) - AWS re:Invent 2018Amazon Web Services
 

What's hot (20)

AWS Snowball Edge and AWS Greengrass for Fun and Profit (STG388) - AWS re:Inv...
AWS Snowball Edge and AWS Greengrass for Fun and Profit (STG388) - AWS re:Inv...AWS Snowball Edge and AWS Greengrass for Fun and Profit (STG388) - AWS re:Inv...
AWS Snowball Edge and AWS Greengrass for Fun and Profit (STG388) - AWS re:Inv...
 
Hands-on in the AWS Java Ecosystem (DEV325-R1) - AWS re:Invent 2018
Hands-on in the AWS Java Ecosystem (DEV325-R1) - AWS re:Invent 2018Hands-on in the AWS Java Ecosystem (DEV325-R1) - AWS re:Invent 2018
Hands-on in the AWS Java Ecosystem (DEV325-R1) - AWS re:Invent 2018
 
IoT Microcontrollers and Getting Started with Amazon FreeRTOS (IOT338-R1) - A...
IoT Microcontrollers and Getting Started with Amazon FreeRTOS (IOT338-R1) - A...IoT Microcontrollers and Getting Started with Amazon FreeRTOS (IOT338-R1) - A...
IoT Microcontrollers and Getting Started with Amazon FreeRTOS (IOT338-R1) - A...
 
Monitoring Serverless Applications (SRV303-S) - AWS re:Invent 2018
Monitoring Serverless Applications (SRV303-S) - AWS re:Invent 2018Monitoring Serverless Applications (SRV303-S) - AWS re:Invent 2018
Monitoring Serverless Applications (SRV303-S) - AWS re:Invent 2018
 
Enterprise DevOps: Patterns of Efficiency (ENT311-R1) - AWS re:Invent 2018
Enterprise DevOps: Patterns of Efficiency (ENT311-R1) - AWS re:Invent 2018Enterprise DevOps: Patterns of Efficiency (ENT311-R1) - AWS re:Invent 2018
Enterprise DevOps: Patterns of Efficiency (ENT311-R1) - AWS re:Invent 2018
 
Data Lake Patterns for Voice, Vision, Advanced Analytics, & ML Using Serverle...
Data Lake Patterns for Voice, Vision, Advanced Analytics, & ML Using Serverle...Data Lake Patterns for Voice, Vision, Advanced Analytics, & ML Using Serverle...
Data Lake Patterns for Voice, Vision, Advanced Analytics, & ML Using Serverle...
 
Getting Started with AWS Greengrass (IOT215-R3) - AWS re:Invent 2018
Getting Started with AWS Greengrass (IOT215-R3) - AWS re:Invent 2018Getting Started with AWS Greengrass (IOT215-R3) - AWS re:Invent 2018
Getting Started with AWS Greengrass (IOT215-R3) - AWS re:Invent 2018
 
[NEW LAUNCH!] Scaling Tightly-coupled HPC workloads on HPC with Elastic Fabri...
[NEW LAUNCH!] Scaling Tightly-coupled HPC workloads on HPC with Elastic Fabri...[NEW LAUNCH!] Scaling Tightly-coupled HPC workloads on HPC with Elastic Fabri...
[NEW LAUNCH!] Scaling Tightly-coupled HPC workloads on HPC with Elastic Fabri...
 
Pause and Resume your EC2 Instances with Hibernate (CMP392) - AWS re:Invent 2018
Pause and Resume your EC2 Instances with Hibernate (CMP392) - AWS re:Invent 2018Pause and Resume your EC2 Instances with Hibernate (CMP392) - AWS re:Invent 2018
Pause and Resume your EC2 Instances with Hibernate (CMP392) - AWS re:Invent 2018
 
Making Hybrid Work for You: Getting into the Cloud Fast (GPSTEC308) - AWS re:...
Making Hybrid Work for You: Getting into the Cloud Fast (GPSTEC308) - AWS re:...Making Hybrid Work for You: Getting into the Cloud Fast (GPSTEC308) - AWS re:...
Making Hybrid Work for You: Getting into the Cloud Fast (GPSTEC308) - AWS re:...
 
Leadership Session: Using DevOps, Microservices, and Serverless to Accelerate...
Leadership Session: Using DevOps, Microservices, and Serverless to Accelerate...Leadership Session: Using DevOps, Microservices, and Serverless to Accelerate...
Leadership Session: Using DevOps, Microservices, and Serverless to Accelerate...
 
Operational Excellence with Containerized Workloads Using AWS Fargate (CON320...
Operational Excellence with Containerized Workloads Using AWS Fargate (CON320...Operational Excellence with Containerized Workloads Using AWS Fargate (CON320...
Operational Excellence with Containerized Workloads Using AWS Fargate (CON320...
 
AWS re:Invent 2018: Deep Dive: Hybrid Cloud Storage Arch. w/Storage Gateway, ...
AWS re:Invent 2018: Deep Dive: Hybrid Cloud Storage Arch. w/Storage Gateway, ...AWS re:Invent 2018: Deep Dive: Hybrid Cloud Storage Arch. w/Storage Gateway, ...
AWS re:Invent 2018: Deep Dive: Hybrid Cloud Storage Arch. w/Storage Gateway, ...
 
Continuous Compliance for Modern Application Pipelines (GPSWS402) - AWS re:In...
Continuous Compliance for Modern Application Pipelines (GPSWS402) - AWS re:In...Continuous Compliance for Modern Application Pipelines (GPSWS402) - AWS re:In...
Continuous Compliance for Modern Application Pipelines (GPSWS402) - AWS re:In...
 
Mythical Mysfits: Management and Ops with AWS Fargate (CON322-R1) - AWS re:In...
Mythical Mysfits: Management and Ops with AWS Fargate (CON322-R1) - AWS re:In...Mythical Mysfits: Management and Ops with AWS Fargate (CON322-R1) - AWS re:In...
Mythical Mysfits: Management and Ops with AWS Fargate (CON322-R1) - AWS re:In...
 
善用 GraphQL 與 AWS AppSync 讓您的 Progressive Web App (PWA) 加速進化 (Level 200)
善用  GraphQL 與 AWS AppSync 讓您的  Progressive Web App (PWA) 加速進化 (Level 200)善用  GraphQL 與 AWS AppSync 讓您的  Progressive Web App (PWA) 加速進化 (Level 200)
善用 GraphQL 與 AWS AppSync 讓您的 Progressive Web App (PWA) 加速進化 (Level 200)
 
AWS and Symantec: Cyber Defense at Scale (SEC311-S) - AWS re:Invent 2018
AWS and Symantec: Cyber Defense at Scale (SEC311-S) - AWS re:Invent 2018AWS and Symantec: Cyber Defense at Scale (SEC311-S) - AWS re:Invent 2018
AWS and Symantec: Cyber Defense at Scale (SEC311-S) - AWS re:Invent 2018
 
Deploying Your ONNX Deep Learning with Apache MXNet Model Server (AIM413) - A...
Deploying Your ONNX Deep Learning with Apache MXNet Model Server (AIM413) - A...Deploying Your ONNX Deep Learning with Apache MXNet Model Server (AIM413) - A...
Deploying Your ONNX Deep Learning with Apache MXNet Model Server (AIM413) - A...
 
Update Microcontroller Devices Over-the-Air with Amazon FreeRTOS (IOT304-R1) ...
Update Microcontroller Devices Over-the-Air with Amazon FreeRTOS (IOT304-R1) ...Update Microcontroller Devices Over-the-Air with Amazon FreeRTOS (IOT304-R1) ...
Update Microcontroller Devices Over-the-Air with Amazon FreeRTOS (IOT304-R1) ...
 
Day Two Operations of Kubernetes on AWS (GPSTEC309) - AWS re:Invent 2018
Day Two Operations of Kubernetes on AWS (GPSTEC309) - AWS re:Invent 2018Day Two Operations of Kubernetes on AWS (GPSTEC309) - AWS re:Invent 2018
Day Two Operations of Kubernetes on AWS (GPSTEC309) - AWS re:Invent 2018
 

Similar to Machine Learning at the IoT Edge (IOT214) - AWS re:Invent 2018

AWS Greengrass & Amazon FreeRTOS: Connectivity & Security at the Edge (IOT356...
AWS Greengrass & Amazon FreeRTOS: Connectivity & Security at the Edge (IOT356...AWS Greengrass & Amazon FreeRTOS: Connectivity & Security at the Edge (IOT356...
AWS Greengrass & Amazon FreeRTOS: Connectivity & Security at the Edge (IOT356...Amazon Web Services
 
Code in the Cloud- Deploy on Microcontroller and Edge Devices
Code in the Cloud- Deploy on Microcontroller and Edge DevicesCode in the Cloud- Deploy on Microcontroller and Edge Devices
Code in the Cloud- Deploy on Microcontroller and Edge DevicesAmazon Web Services
 
AWS IoT_Connected Home Solutions
AWS IoT_Connected Home Solutions AWS IoT_Connected Home Solutions
AWS IoT_Connected Home Solutions Amazon Web Services
 
Machine learning at the IoT Edge with AWS IoT Greengrass - SVC203 - Atlanta A...
Machine learning at the IoT Edge with AWS IoT Greengrass - SVC203 - Atlanta A...Machine learning at the IoT Edge with AWS IoT Greengrass - SVC203 - Atlanta A...
Machine learning at the IoT Edge with AWS IoT Greengrass - SVC203 - Atlanta A...Amazon Web Services
 
NEW LAUNCH! AWS Greengrass and Amazon FreeRTOS: Connectivity and Security at ...
NEW LAUNCH! AWS Greengrass and Amazon FreeRTOS: Connectivity and Security at ...NEW LAUNCH! AWS Greengrass and Amazon FreeRTOS: Connectivity and Security at ...
NEW LAUNCH! AWS Greengrass and Amazon FreeRTOS: Connectivity and Security at ...Amazon Web Services
 
Aws Tools for Alexa Skills
Aws Tools for Alexa SkillsAws Tools for Alexa Skills
Aws Tools for Alexa SkillsBoaz Ziniman
 
Leadership Session: AWS IoT (IOT218-L) - AWS re:Invent 2018
Leadership Session: AWS IoT (IOT218-L) - AWS re:Invent 2018Leadership Session: AWS IoT (IOT218-L) - AWS re:Invent 2018
Leadership Session: AWS IoT (IOT218-L) - AWS re:Invent 2018Amazon Web Services
 
Computing at the Edge with AWS Greengrass and Amazon FreeRTOS, ft. Enel (IOT2...
Computing at the Edge with AWS Greengrass and Amazon FreeRTOS, ft. Enel (IOT2...Computing at the Edge with AWS Greengrass and Amazon FreeRTOS, ft. Enel (IOT2...
Computing at the Edge with AWS Greengrass and Amazon FreeRTOS, ft. Enel (IOT2...Amazon Web Services
 
SRV201 Push Intelligence to the Edge Machine Learning on AWS Greengrass Devices
SRV201 Push Intelligence to the Edge Machine Learning on AWS Greengrass Devices SRV201 Push Intelligence to the Edge Machine Learning on AWS Greengrass Devices
SRV201 Push Intelligence to the Edge Machine Learning on AWS Greengrass Devices Amazon Web Services
 
IoT Compute at the Edge with AWS Greengrass - GOTO Amsterdam
IoT Compute at the Edge with AWS Greengrass - GOTO AmsterdamIoT Compute at the Edge with AWS Greengrass - GOTO Amsterdam
IoT Compute at the Edge with AWS Greengrass - GOTO AmsterdamBoaz Ziniman
 
Perform Machine Learning at the IoT Edge using AWS Greengrass and Amazon Sage...
Perform Machine Learning at the IoT Edge using AWS Greengrass and Amazon Sage...Perform Machine Learning at the IoT Edge using AWS Greengrass and Amazon Sage...
Perform Machine Learning at the IoT Edge using AWS Greengrass and Amazon Sage...Amazon Web Services
 
Introducing the New Features of AWS Greengrass (IOT365) - AWS re:Invent 2018
Introducing the New Features of AWS Greengrass (IOT365) - AWS re:Invent 2018Introducing the New Features of AWS Greengrass (IOT365) - AWS re:Invent 2018
Introducing the New Features of AWS Greengrass (IOT365) - AWS re:Invent 2018Amazon Web Services
 
AWS IoT Greengrass Workshop - SVC303 - Anaheim AWS Summit
AWS IoT Greengrass Workshop - SVC303 - Anaheim AWS SummitAWS IoT Greengrass Workshop - SVC303 - Anaheim AWS Summit
AWS IoT Greengrass Workshop - SVC303 - Anaheim AWS SummitAmazon Web Services
 
Using AWS Lambda as a Security Team (SEC322-R1) - AWS re:Invent 2018
Using AWS Lambda as a Security Team (SEC322-R1) - AWS re:Invent 2018Using AWS Lambda as a Security Team (SEC322-R1) - AWS re:Invent 2018
Using AWS Lambda as a Security Team (SEC322-R1) - AWS re:Invent 2018Amazon Web Services
 
Building IoT Applications for a Smart Home, ft. Vestel (IOT306-R1) - AWS re:I...
Building IoT Applications for a Smart Home, ft. Vestel (IOT306-R1) - AWS re:I...Building IoT Applications for a Smart Home, ft. Vestel (IOT306-R1) - AWS re:I...
Building IoT Applications for a Smart Home, ft. Vestel (IOT306-R1) - AWS re:I...Amazon Web Services
 
AWS IoT - from Cloud to Edge | AWS Floor28
AWS IoT - from Cloud to Edge | AWS Floor28AWS IoT - from Cloud to Edge | AWS Floor28
AWS IoT - from Cloud to Edge | AWS Floor28Amazon Web Services
 
Introduction to AWS IoT Greengrass - SVC305 - Chicago AWS Summit
Introduction to AWS IoT Greengrass - SVC305 - Chicago AWS SummitIntroduction to AWS IoT Greengrass - SVC305 - Chicago AWS Summit
Introduction to AWS IoT Greengrass - SVC305 - Chicago AWS SummitAmazon Web Services
 
Driving Innovation with Serverless Applications (GPSBUS212) - AWS re:Invent 2018
Driving Innovation with Serverless Applications (GPSBUS212) - AWS re:Invent 2018Driving Innovation with Serverless Applications (GPSBUS212) - AWS re:Invent 2018
Driving Innovation with Serverless Applications (GPSBUS212) - AWS re:Invent 2018Amazon Web Services
 

Similar to Machine Learning at the IoT Edge (IOT214) - AWS re:Invent 2018 (20)

AWS Greengrass & Amazon FreeRTOS: Connectivity & Security at the Edge (IOT356...
AWS Greengrass & Amazon FreeRTOS: Connectivity & Security at the Edge (IOT356...AWS Greengrass & Amazon FreeRTOS: Connectivity & Security at the Edge (IOT356...
AWS Greengrass & Amazon FreeRTOS: Connectivity & Security at the Edge (IOT356...
 
Code in the Cloud- Deploy on Microcontroller and Edge Devices
Code in the Cloud- Deploy on Microcontroller and Edge DevicesCode in the Cloud- Deploy on Microcontroller and Edge Devices
Code in the Cloud- Deploy on Microcontroller and Edge Devices
 
AWS IoT_Connected Home Solutions
AWS IoT_Connected Home Solutions AWS IoT_Connected Home Solutions
AWS IoT_Connected Home Solutions
 
Machine learning at the IoT Edge with AWS IoT Greengrass - SVC203 - Atlanta A...
Machine learning at the IoT Edge with AWS IoT Greengrass - SVC203 - Atlanta A...Machine learning at the IoT Edge with AWS IoT Greengrass - SVC203 - Atlanta A...
Machine learning at the IoT Edge with AWS IoT Greengrass - SVC203 - Atlanta A...
 
ML Inference at the Edge
ML Inference at the EdgeML Inference at the Edge
ML Inference at the Edge
 
NEW LAUNCH! AWS Greengrass and Amazon FreeRTOS: Connectivity and Security at ...
NEW LAUNCH! AWS Greengrass and Amazon FreeRTOS: Connectivity and Security at ...NEW LAUNCH! AWS Greengrass and Amazon FreeRTOS: Connectivity and Security at ...
NEW LAUNCH! AWS Greengrass and Amazon FreeRTOS: Connectivity and Security at ...
 
Aws Tools for Alexa Skills
Aws Tools for Alexa SkillsAws Tools for Alexa Skills
Aws Tools for Alexa Skills
 
Leadership Session: AWS IoT (IOT218-L) - AWS re:Invent 2018
Leadership Session: AWS IoT (IOT218-L) - AWS re:Invent 2018Leadership Session: AWS IoT (IOT218-L) - AWS re:Invent 2018
Leadership Session: AWS IoT (IOT218-L) - AWS re:Invent 2018
 
Computing at the Edge with AWS Greengrass and Amazon FreeRTOS, ft. Enel (IOT2...
Computing at the Edge with AWS Greengrass and Amazon FreeRTOS, ft. Enel (IOT2...Computing at the Edge with AWS Greengrass and Amazon FreeRTOS, ft. Enel (IOT2...
Computing at the Edge with AWS Greengrass and Amazon FreeRTOS, ft. Enel (IOT2...
 
IoT Made Easy | AWS IoT
IoT Made Easy | AWS IoTIoT Made Easy | AWS IoT
IoT Made Easy | AWS IoT
 
SRV201 Push Intelligence to the Edge Machine Learning on AWS Greengrass Devices
SRV201 Push Intelligence to the Edge Machine Learning on AWS Greengrass Devices SRV201 Push Intelligence to the Edge Machine Learning on AWS Greengrass Devices
SRV201 Push Intelligence to the Edge Machine Learning on AWS Greengrass Devices
 
IoT Compute at the Edge with AWS Greengrass - GOTO Amsterdam
IoT Compute at the Edge with AWS Greengrass - GOTO AmsterdamIoT Compute at the Edge with AWS Greengrass - GOTO Amsterdam
IoT Compute at the Edge with AWS Greengrass - GOTO Amsterdam
 
Perform Machine Learning at the IoT Edge using AWS Greengrass and Amazon Sage...
Perform Machine Learning at the IoT Edge using AWS Greengrass and Amazon Sage...Perform Machine Learning at the IoT Edge using AWS Greengrass and Amazon Sage...
Perform Machine Learning at the IoT Edge using AWS Greengrass and Amazon Sage...
 
Introducing the New Features of AWS Greengrass (IOT365) - AWS re:Invent 2018
Introducing the New Features of AWS Greengrass (IOT365) - AWS re:Invent 2018Introducing the New Features of AWS Greengrass (IOT365) - AWS re:Invent 2018
Introducing the New Features of AWS Greengrass (IOT365) - AWS re:Invent 2018
 
AWS IoT Greengrass Workshop - SVC303 - Anaheim AWS Summit
AWS IoT Greengrass Workshop - SVC303 - Anaheim AWS SummitAWS IoT Greengrass Workshop - SVC303 - Anaheim AWS Summit
AWS IoT Greengrass Workshop - SVC303 - Anaheim AWS Summit
 
Using AWS Lambda as a Security Team (SEC322-R1) - AWS re:Invent 2018
Using AWS Lambda as a Security Team (SEC322-R1) - AWS re:Invent 2018Using AWS Lambda as a Security Team (SEC322-R1) - AWS re:Invent 2018
Using AWS Lambda as a Security Team (SEC322-R1) - AWS re:Invent 2018
 
Building IoT Applications for a Smart Home, ft. Vestel (IOT306-R1) - AWS re:I...
Building IoT Applications for a Smart Home, ft. Vestel (IOT306-R1) - AWS re:I...Building IoT Applications for a Smart Home, ft. Vestel (IOT306-R1) - AWS re:I...
Building IoT Applications for a Smart Home, ft. Vestel (IOT306-R1) - AWS re:I...
 
AWS IoT - from Cloud to Edge | AWS Floor28
AWS IoT - from Cloud to Edge | AWS Floor28AWS IoT - from Cloud to Edge | AWS Floor28
AWS IoT - from Cloud to Edge | AWS Floor28
 
Introduction to AWS IoT Greengrass - SVC305 - Chicago AWS Summit
Introduction to AWS IoT Greengrass - SVC305 - Chicago AWS SummitIntroduction to AWS IoT Greengrass - SVC305 - Chicago AWS Summit
Introduction to AWS IoT Greengrass - SVC305 - Chicago AWS Summit
 
Driving Innovation with Serverless Applications (GPSBUS212) - AWS re:Invent 2018
Driving Innovation with Serverless Applications (GPSBUS212) - AWS re:Invent 2018Driving Innovation with Serverless Applications (GPSBUS212) - AWS re:Invent 2018
Driving Innovation with Serverless Applications (GPSBUS212) - AWS re:Invent 2018
 

More from Amazon Web Services

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...Amazon Web Services
 
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...Amazon Web Services
 
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 FargateAmazon Web Services
 
Costruire Applicazioni Moderne con AWS
Costruire Applicazioni Moderne con AWSCostruire Applicazioni Moderne con AWS
Costruire Applicazioni Moderne con AWSAmazon Web Services
 
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 Amazon Web Services
 
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...Amazon Web Services
 
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...Amazon Web Services
 
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 WorkloadsAmazon Web Services
 
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 sfatareAmazon Web Services
 
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 NodeJSAmazon Web Services
 
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 webAmazon Web Services
 
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 sfatareAmazon 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 AWSAmazon 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 DeckAmazon Web Services
 
Building a web application without servers
Building a web application without serversBuilding a web application without servers
Building a web application without serversAmazon 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
 
Introduzione a Amazon Elastic Container Service
Introduzione a Amazon Elastic Container ServiceIntroduzione a Amazon Elastic Container Service
Introduzione a Amazon Elastic Container ServiceAmazon 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
 

Machine Learning at the IoT Edge (IOT214) - AWS re:Invent 2018

  • 1.
  • 2. © 2018, Amazon Web Services, Inc. or its affiliates. All rights reserved. Machine Learning at the IoT Edge David Nunnerley AWS Senior Manager AWS IoT Greengrass I O T 2 1 4 Nobutaka Nakazawa CTO Brains Technology, Inc. Masanori Sato Group Manager Aisin AW LTD
  • 3. © 2018, Amazon Web Services, Inc. or its affiliates. All rights reserved. Agenda Why Machine Learning (ML) at the Edge? AWS IoT Greengrass overview ML Inference at the Edge with AWS IoT Greengrass New AWS IoT Greengrass ML capabilities Customer use case: Aisin AW (Masanori Sato) Brains Technology (Nobutaka Nakazawa)
  • 4. © 2018, Amazon Web Services, Inc. or its affiliates. All rights reserved.
  • 5. © 2018, Amazon Web Services, Inc. or its affiliates. All rights reserved. Medical equipment Industrial machinery Extreme environments Most machine data never reaches the cloud
  • 6. © 2018, Amazon Web Services, Inc. or its affiliates. All rights reserved. Why this problem isn’t going away Law of physics Law of economics Law of the land
  • 7. © 2018, Amazon Web Services, Inc. or its affiliates. All rights reserved.
  • 8. © 2018, Amazon Web Services, Inc. or its affiliates. All rights reserved. AWS IoT Greengrass
  • 9. © 2018, Amazon Web Services, Inc. or its affiliates. All rights reserved. AWS IoT Greengrass All AWS Cloud services e.g., Amazon S3, Amazon Kinesis, Amazon Redshift… AWS IoT services e.g., AWS IoT Core, AWS IoT Analytics, AWS IoT Device Defender…
  • 10. © 2018, Amazon Web Services, Inc. or its affiliates. All rights reserved. Data and State Sync Security Over the Air UpdatesConnectors Local Device Shadows Code Deployment Lambda Functions AWS-grade security Easily Update Greengrass Core Machine Learning Inference Local Execution of ML Models Local Resource Access Lambdas Interact With Peripherals Easy integrations with AWS services, protocol adaptors and other SaaS providers Local Messages and Triggers Local Message Broker Manage Secrets at the edge AWS Secrets Manager functionality at edge AWS Greengrass Extend AWS IoT to the Edge
  • 11. © 2018, Amazon Web Services, Inc. or its affiliates. All rights reserved. Data and State Sync Security Over the Air UpdatesConnectors Local Device Shadows Code Deployment Lambda Functions AWS-grade security Easily Update Greengrass Core Machine Learning Inference Local Execution of ML Models Local Resource Access Lambdas Interact With Peripherals Easy integrations with AWS services, protocol adaptors and other SaaS providers Local Messages and Triggers Local Message Broker Manage Secrets at the edge AWS Secrets Manager functionality at edge AWS Greengrass Extend AWS IoT to the Edge
  • 12. © 2018, Amazon Web Services, Inc. or its affiliates. All rights reserved. Data and State Sync Security Over the Air UpdatesConnectors Local Device Shadows Code Deployment Lambda Functions AWS-grade security Easily Update Greengrass Core Machine Learning Inference Local Execution of ML Models Local Resource Access Lambdas Interact With Peripherals Easy integrations with AWS services, protocol adaptors and other SaaS providers Local Messages and Triggers Local Message Broker Manage Secrets at the edge AWS Secrets Manager functionality at edge AWS Greengrass Extend AWS IoT to the Edge
  • 13. © 2018, Amazon Web Services, Inc. or its affiliates. All rights reserved. Data and State Sync Security Over the Air UpdatesConnectors Local Device Shadows Code Deployment Lambda Functions AWS-grade security Easily Update Greengrass Core Machine Learning Inference Local Execution of ML Models Local Resource Access Lambdas Interact With Peripherals Easy integrations with AWS services, protocol adaptors and other SaaS providers Local Messages and Triggers Local Message Broker Manage Secrets at the edge AWS Secrets Manager functionality at edge AWS Greengrass Extend AWS IoT to the Edge
  • 14. © 2018, Amazon Web Services, Inc. or its affiliates. All rights reserved. Data and State Sync Security Over the Air UpdatesConnectors Local Device Shadows Code Deployment Lambda Functions AWS-grade security Easily Update Greengrass Core Machine Learning Inference Local Execution of ML Models Local Resource Access Lambdas Interact With Peripherals Easy integrations with AWS services, protocol adaptors and other SaaS providers Local Messages and Triggers Local Message Broker Manage Secrets at the edge AWS Secrets Manager functionality at edge AWS Greengrass Extend AWS IoT to the Edge
  • 15. © 2018, Amazon Web Services, Inc. or its affiliates. All rights reserved. Data and State Sync Security Over the Air UpdatesConnectors Local Device Shadows Code Deployment Lambda Functions AWS-grade security Easily Update Greengrass Core Machine Learning Inference Local Execution of ML Models Local Resource Access Lambdas Interact With Peripherals Easy integrations with AWS services, protocol adaptors and other SaaS providers Local Messages and Triggers Local Message Broker Manage Secrets at the edge AWS Secrets Manager functionality at edge AWS Greengrass Extend AWS IoT to the Edge
  • 16. © 2018, Amazon Web Services, Inc. or its affiliates. All rights reserved. Data and State Sync Security Over the Air UpdatesConnectors Local Device Shadows Code Deployment Lambda Functions AWS-grade security Easily Update Greengrass Core Machine Learning Inference Local Execution of ML Models Local Resource Access Lambdas Interact With Peripherals Easy integrations with AWS services, protocol adaptors and other SaaS providers Local Messages and Triggers Local Message Broker Manage Secrets at the edge AWS Secrets Manager functionality at edge AWS Greengrass Extend AWS IoT to the Edge
  • 17. © 2018, Amazon Web Services, Inc. or its affiliates. All rights reserved. Data and State Sync Security Over the Air UpdatesConnectors Local Device Shadows Code Deployment Lambda Functions AWS-grade security Easily Update Greengrass Core Machine Learning Inference Local Execution of ML Models Local Resource Access Lambdas Interact With Peripherals Easy integrations with AWS services, protocol adaptors and other SaaS providers Local Messages and Triggers Local Message Broker Manage Secrets at the edge AWS Secrets Manager functionality at edge AWS Greengrass Extend AWS IoT to the Edge
  • 18. © 2018, Amazon Web Services, Inc. or its affiliates. All rights reserved. Data and State Sync Security Over the Air UpdatesConnectors Local Device Shadows Code Deployment Lambda Functions AWS-grade security Easily Update Greengrass Core Machine Learning Inference Local Execution of ML Models Local Resource Access Lambdas Interact With Peripherals Easy integrations with AWS services, protocol adaptors and other SaaS providers Local Messages and Triggers Local Message Broker Manage Secrets at the edge AWS Secrets Manager functionality at edge AWS Greengrass Extend AWS IoT to the Edge
  • 19. © 2018, Amazon Web Services, Inc. or its affiliates. All rights reserved. Data and State Sync Security Over the Air UpdatesConnectors Local Device Shadows Code Deployment Lambda Functions AWS-grade security Easily Update Greengrass Core Machine Learning Inference Local Execution of ML Models Local Resource Access Lambdas Interact With Peripherals Easy integrations with AWS services, protocol adaptors and other SaaS providers Local Messages and Triggers Local Message Broker Manage Secrets at the edge AWS Secrets Manager functionality at edge AWS Greengrass Extend AWS IoT to the Edge
  • 20. © 2018, Amazon Web Services, Inc. or its affiliates. All rights reserved.
  • 21. © 2018, Amazon Web Services, Inc. or its affiliates. All rights reserved. Inference Training Machine Learning at the Edge Local actions Edge Cloud
  • 22. © 2018, Amazon Web Services, Inc. or its affiliates. All rights reserved. Inference Training Machine Learning at the Edge Local actions Edge Cloud
  • 23. © 2018, Amazon Web Services, Inc. or its affiliates. All rights reserved. Inference Training Machine Learning at the Edge Local actions Edge Cloud
  • 24. © 2018, Amazon Web Services, Inc. or its affiliates. All rights reserved. Inference Training Machine Learning at the Edge Local actions Edge Cloud
  • 25. © 2018, Amazon Web Services, Inc. or its affiliates. All rights reserved. Inference Training Machine Learning at the Edge Local actions Edge Cloud
  • 26. © 2018, Amazon Web Services, Inc. or its affiliates. All rights reserved. Inference Training Machine Learning at the Edge Local actions Edge Cloud
  • 27. © 2018, Amazon Web Services, Inc. or its affiliates. All rights reserved. AWS IoT Greengrass Core Machine Learning Run ML Inference on Greengrass Deploy an ML model from Amazon SageMaker in the cloud to a target AWS Greengrass core device using the Greengrass console or Command Line Interface (AWS CLI) Install the necessary run-time for the model e.g., (TensorFlow, Apache MXNet, Chainer…) on the AWS Greengrass core Available for multiple hardware architectures: e.g., Intel x86-64, ARM v7 and Nvidia Jetson TX2 Code your Lambda to read from attached device/sensor (optionally from MQTT topic) and pass to the Lambda running the ML model. Take action based upon the inference.
  • 28. © 2018, Amazon Web Services, Inc. or its affiliates. All rights reserved. AWS IoT Greengrass Core 1.7 Machine Learning New Machine Learning capabilities Image Classification Connector (available for download from console) Pre-built Lambda to run the Image classification ML model Easy coding to bridge from input device to the supplied Lambda running the inference Image Classification Model can be trained to learn new image classifications in the cloud with Amazon SageMaker
  • 29. © 2018, Amazon Web Services, Inc. or its affiliates. All rights reserved. Code to use the Image Classification Connector import greengrass_machine_learning_sdk as ml with open('/test_img/test.jpg', 'rb') as f: content = f.read() def infer(): logging.info('invoking Greengrass ML Inference service') try: resp = client.invoke_inference_service( AlgoType='image-classification', ServiceName='imageClassification', ContentType='image/jpeg', Body=content ) except ml.GreengrassInferenceException as e: logging.info('inference exception {}("{}")'.format(e.__class__.__name__, e)) return except ml.GreengrassDependencyException as e: logging.info('dependency exception {}("{}")'.format(e.__class__.__name__, e)) return logging.info('resp: {}'.format(resp)) predictions = resp['Body'].read() logging.info('predictions: {}'.format(predictions))
  • 30. © 2018, Amazon Web Services, Inc. or its affiliates. All rights reserved. Code to use the Image Classification Connector import greengrass_machine_learning_sdk as ml with open('/test_img/test.jpg', 'rb') as f: content = f.read() def infer(): logging.info('invoking Greengrass ML Inference service') try: resp = client.invoke_inference_service( AlgoType='image-classification', ServiceName='imageClassification', ContentType='image/jpeg', Body=content ) except ml.GreengrassInferenceException as e: logging.info('inference exception {}("{}")'.format(e.__class__.__name__, e)) return except ml.GreengrassDependencyException as e: logging.info('dependency exception {}("{}")'.format(e.__class__.__name__, e)) return logging.info('resp: {}'.format(resp)) predictions = resp['Body'].read() logging.info('predictions: {}'.format(predictions))
  • 31. © 2018, Amazon Web Services, Inc. or its affiliates. All rights reserved. Code to use the Image Classification Connector import greengrass_machine_learning_sdk as ml with open('/test_img/test.jpg', 'rb') as f: content = f.read() def infer(): logging.info('invoking Greengrass ML Inference service') try: resp = client.invoke_inference_service( AlgoType='image-classification', ServiceName='imageClassification', ContentType='image/jpeg', Body=content ) except ml.GreengrassInferenceException as e: logging.info('inference exception {}("{}")'.format(e.__class__.__name__, e)) return except ml.GreengrassDependencyException as e: logging.info('dependency exception {}("{}")'.format(e.__class__.__name__, e)) return logging.info('resp: {}'.format(resp)) predictions = resp['Body'].read() logging.info('predictions: {}'.format(predictions))
  • 32. © 2018, Amazon Web Services, Inc. or its affiliates. All rights reserved. AWS IoT Greengrass Core 1.7 Machine Learning Other New Machine Learning capabilities Greengrass support for the new Amazon SageMaker Neo (Deep Learning Runtime) Optimize the model using Neo compiler in the cloud More performant without loss of accuracy Smaller memory footprint Deploy optimized Neo model to the Greengrass core device Install Neo run-time to the device Write a Lambda to run the Neo optimized ML model
  • 33. Smart Vending Machine -- R&D Innovation Team, from the SA Americas organization
  • 34. © 2018, Amazon Web Services, Inc. or its affiliates. All rights reserved. Camera Smart vending machine Object Detection and Image Classification models Load Sensors readings in local time series database Sensor fusion functions to detect removed items and strange objects Thanks for shopping! 3x Water Bottles USD 1.50 each Your total is $4.50
  • 35. © 2018, Amazon Web Services, Inc. or its affiliates. All rights reserved. Our Launch Partners “The addition of AWS IoT Greengrass with its latest ML Inference update running on ADLINK’s industrial vision systems makes for truly plug- and-play IoT. Now when we power-on an off-the-shelf ADLINK NEON smart camera running AWS IoT Greengrass with its latest ML Inference update, we can get to high-quality outcomes much, much faster. This allows us to further speed development of our IoT digital experiments for our logistics, quality inspection, industrial robotics, and other manufacturing customers.” - Elizabeth Campbell, General Manager, The Americas, ADLINK Technology “The potential of computer vision use cases enabled by IoT and AI is vast for businesses to exponentially improve productivity and efficiency. In this time of intelligent transformation, our premium industrial Think IoT cameras powered by AWS IoT Greengrass with the latest machine learning upgrades are engineered to make a notable difference to enterprise customers.” - Jon Pershke, Vice President of Strategy and Emerging Business, Intelligent Devices “The pervasiveness of artificial intelligence and the pace of digital transformation continues to grow at an astonishing rate. Innovations like the newest improvements to AWS IoT Greengrass Machine Learning that markedly decrease latency without decreasing the accuracy of ML inference accelerate new solutions to emerging industrial automation use cases for object identification and classification. AWS’ new machine learning solution integrated with Leopard Imaging’s AICam powered by NVIDIA® GPU will be a cornerstone in any edge to cloud Industrial and Smart City solution.” -Bill Pu, President and Co-Founder, Leopard Imaging “Vieureka of Panasonic is very pleased to utilize the application evolving functions of AWS’s machine learning as enabled by AWS IoT Greengrass. In order to offer Vieureka-Cameras and service management functions to all the partners of the AWS community, I would like to develop a Greengrass compatible version as soon as possible. We will create the environment for developers in the spring of 2019, with commercial versions available in autumn of the same year.” - Miyazaki, CEO of Vieureka Service, Panasonic
  • 36. © 2018, Amazon Web Services, Inc. or its affiliates. All rights reserved. Nobutaka Nakazawa Brains Technology, Inc. CTO Masanori Sato Aisin AW LTD Group Manager
  • 37. © 2018, Amazon Web Services, Inc. or its affiliates. All rights reserved. Machine state monitoring by cloud & edge computing AISIN AW CO.,LTD. Manufacturing Engineering Development Production System Innovation Group Masanori Sato
  • 38. © 2018, Amazon Web Services, Inc. or its affiliates. All rights reserved. Agenda ・ Company profile ・ About our production engineering ・ Action background ・ Action summary
  • 39. © 2018, Amazon Web Services, Inc. or its affiliates. All rights reserved. 9,830,000 NAVI Securing a Top Share of the Global Market with Our Innovative Manufacturing World No.1 World No.2 AT ■ Business summary of 2017(in March, 2018) Unit sales AT: 9,830,000units NAVI: 1,810,000units Sales amount A connection: 1,621,200 million yen AISIN AW CO.,LTD Company Profile
  • 40. © 2018, Amazon Web Services, Inc. or its affiliates. All rights reserved. Mission of manufacturing engineering Workplace skills that bringing value: Production engineering Production ◆ Finish of the SE ※ / drawing ※ Simultaneous Engineering ◆ Design of the product line / setup ◆ Design of facilities / production ◆ Development of the new production engineering ◆ Plan, design of the factory Ordering, suggestion Three-Pillar Manufacturing Suggestion Suggestion Suggestion Production engineering Product Design Trading company Equipment manufacturer Delivery of goods, suggestion Cooperate as a partner Vender
  • 41. © 2018, Amazon Web Services, Inc. or its affiliates. All rights reserved. Background of the action Way of thinking of Industrial IoT in AISIN AW Man Machine Material Method Human factor from who involved ・ Assembling ・ Machine setting ・ ・・etc. Factor hardware such as machines ・ Blade tool ・ Metal mold ・ ・・ etc. Factor from Materials (property value) ・ Ductility, toughness ・ Hardness ・ ・・etc. Factor from production method ・ Processing method ・ Processing order ・ ・・etc. Building a base of high level condition monitoring and control by using information technology The production revolution by IT is proposed in the world → Need to develop the Industrial IoT production system for AISIN AW
  • 42. © 2018, Amazon Web Services, Inc. or its affiliates. All rights reserved. Background of the action What we need for I-IoT We have started to develop I-IoT system that utilizes "cloud & edge" that can satisfy these requirements • Small start • Real-time detection • Scalability -Connectable with more than 20,000 machines -Easy deployment to each factory • Successful partner -Quick and challenge
  • 43. © 2018, Amazon Web Services, Inc. or its affiliates. All rights reserved. Cloud & edge computing analysis base ◯◯◯ Factory ◯◯◯ Factory Analysis monitor Edge device AWS Services - Storage, Managed Services, etc Cloud Edge device • Dashboard • Simulation model making • Algorithm development • Edge device management Factory ANotice monitor Real-time monitoring & ML detection Factory B Analysis monitor Notice monitor Machines Machines
  • 44. © 2018, Amazon Web Services, Inc. or its affiliates. All rights reserved. Cause inquire-able detection ML algorithm Develop machine learning algorithm that person can understand a results and can improve immediately If not If cause is clear Data A Machine learning Not good Data A Machine learning It’s not red enough, and It is not ripe What is ? What part ? I see! I can take action immediately Not good [No reason] [Reason]
  • 45. © 2018, Amazon Web Services, Inc. or its affiliates. All rights reserved. Cause inquire-able detection ML algorithm Good point ・ Got good result in a month after using →Because satisfaction of detection result, it‘s actively used → Leads to expansion ・The model can be constructed with high precision at an early stage →Model making took three months ⇒ one week Difficult point ・Since it’s new initiative, it will not be adopted unless it is indicated by the result →Need one year temporary use for the use at the mass production line
  • 46. © 2018, Amazon Web Services, Inc. or its affiliates. All rights reserved. Outcomes of cloud & edge system in AISIN AW Development of the I-IoT future ・ Further high precision monitoring by algorithm development ・ Expanding to other factories and processes and supervising management ・ Training of workers to increase IoT talent Time[msec] Time[msec] value value unusual point The system detected “anomaly” state and Suppress the cost of long line stop
  • 47. © 2018, Amazon Web Services, Inc. or its affiliates. All rights reserved. Machine Learning at the IoT Edge Nobutaka Nakazawa CTO Brains Technology, Inc. I O T 2 1 4
  • 48. © 2018, Amazon Web Services, Inc. or its affiliates. All rights reserved. Agenda Company profile Overview of impulse Algorithm Summary
  • 49. © 2018, Amazon Web Services, Inc. or its affiliates. All rights reserved.
  • 50. © 2018, Amazon Web Services, Inc. or its affiliates. All rights reserved. Company profile Company name Brains Technology, Inc. Founded August 8, 2008 CEO Sawako Hamanaka Capital 110 million yen (Including capital reserve) Address Shinagawa Center Building 3-23-17, Takanawa, Minato-Ku, TOKYO, Japan URL https://www.brains-tech.co.jp Provide innovative service and bring about technological innovations with open technology Providing innovative service for business enterprises, improve the productivity of corporate activities dramatically Our Mission
  • 51. © 2018, Amazon Web Services, Inc. or its affiliates. All rights reserved.
  • 52. © 2018, Amazon Web Services, Inc. or its affiliates. All rights reserved. Overview of Impulse 20 41 84 0 20 40 60 80 100 FY2015 (〜2016/7) FY2016 (〜2017/7) FY2017 (〜2018/7) 145+ Predictive Maintenance Quality Management - Plant equipment(power, chemical, bio) - Co-generation system - Industrial machinery (robots) - Construction machines (crane, elevator) - Electrical equipment (air conditioners, water heaters) - Auto parts (transmissions, gears, drive shafts, bumpers) - Electrical equipment (LED) - Chemical products - Casting Impulse is the IoT ML edge platform for the manufacturing industry for any kind of time series data built on top of AWS. https://impulse-cloud.com
  • 53. © 2018, Amazon Web Services, Inc. or its affiliates. All rights reserved. Architecture File Monitor Output result Raw data Anomaly Detection Post Process Factory AWS IoT Amazon S3 AWS Lambda Amazon DynamoDB AWS Batch UI Dashboard Simulation Line A Output result Line B Raw data Edge PC Greengrass Core Thing Thing Amazon Athena Amazon S3 Raw data Model Amazon SageMaker
  • 54. © 2018, Amazon Web Services, Inc. or its affiliates. All rights reserved. Deploy Models using ML Inference Three modules to deploy ML Library (MXNet, Tensorflow, scikit-learn) AWS Greengrass ML Inference or Lambda Your Code Lambda Model AWS Greengrass ML Inference / S3 Steps to deploy • Create AWS Lambda functions for ML inference. • Create Models by Amazon SageMaker or AWS Batch and upload the models to Amazon S3. • AWS IoT fully manages the whole deployment process. Upload model files to S3 Setting up local resource in AWS IoT
  • 55. © 2018, Amazon Web Services, Inc. or its affiliates. All rights reserved.
  • 56. © 2018, Amazon Web Services, Inc. or its affiliates. All rights reserved. Automated algorithm and parameter optimization Analyzing the characteristics of the data and auto-selecting the optimum algorithms and parameters Anomaly detection models Input gaussian periodic correlation independent Mahalanobis S-H-ESD One Class SVM Sparse Coding Recommended parameters Breakout LOF Gaussian Process Data Characteristics Algorithm Parameter Output Recommended parameters Recommended parameters Recommended parameters Recommended parameters Recommended parameters Recommended parameters
  • 57. © 2018, Amazon Web Services, Inc. or its affiliates. All rights reserved. Sparse coding • Sparse coding is a class of unsupervised methods for learning sets of over-complete bases to represent data efficiently • Finds a sparse representation of data against a fixed, precomputed dictionary • Works well for high-speed time- series signals with periodic pattern Dictionary Leaning Decoding from dictionary
  • 58. © 2018, Amazon Web Services, Inc. or its affiliates. All rights reserved. LOF (Local Outlier Factor) • Calculate anomaly scores considering the density and distance of the surrounding data • Works well for high-dimensional correlated data with dimensionality reduction technique (PCA, GPLVM, etc.) • No need to assume a distribution and it can be applied even when the density has multimodality Dimensionality reduction LOF anomaly detection
  • 59. © 2018, Amazon Web Services, Inc. or its affiliates. All rights reserved.
  • 60. © 2018, Amazon Web Services, Inc. or its affiliates. All rights reserved. Summary • Fully managed IoT ML edge platform • Highly-scalable, easy to process ML leaning and model deployment cycle • Deployment of the algorithms and the models from AWS platform can eliminate the need to go on-site to update the algorithm or the ML model • Lambda function with additional libraries (scikit-learn, numpy, pandas, etc.) can run any ML logic you created. (unless exceeding Lambda size limitation) • Some limitations still exist • It is necessary to consider the fault tolerance at the edge • Greengrass only runs on recent Linux environment • Not all regions support AWS Greengrass yet • Time series analysis needs data cache mechanism on the edge
  • 61. Thank you! © 2018, Amazon Web Services, Inc. or its affiliates. All rights reserved. David Nunnerley Masanori Sato Nobutaka Nakazawa
  • 62. © 2018, Amazon Web Services, Inc. or its affiliates. All rights reserved.