Contenu connexe Similaire à ENT301_Real-World AI For the Enterprise (20) Plus de Amazon Web Services (20) ENT301_Real-World AI For the Enterprise1. @dmbanga @apwenchel© 2017, Amazon Web Services, Inc. or its Affiliates. All rights reserved. @dmbanga @apwenchel
Real-World AI For the Enterprise
D a n R . M b a n g a
B u s i n e s s D e v e l o p m e n t M a n a g e r
A I P l a t f o r m s a n d E n g i n e s
@dmbanga
E N T 3 0 1
N o v e m b e r 2 7 , 2 0 1 7
AWS re:Invent
3. @dmbanga @apwenchel© 2017, Amazon Web Services, Inc. or its Affiliates. All rights reserved.
A Cycle for AI
Perception
Learning
Knowledge
Representation
(KR)
4. @dmbanga @apwenchel© 2017, Amazon Web Services, Inc. or its Affiliates. All rights reserved.
A Cycle for AI
Perception
Learning
Knowledge
Representation
(KR)
Reasoning
5. @dmbanga @apwenchel© 2017, Amazon Web Services, Inc. or its Affiliates. All rights reserved.
A Cycle for AI
Perception
Learning
Knowledge
Representation
(KR)
Reasoning
Planning
Execution
6. @dmbanga @apwenchel© 2017, Amazon Web Services, Inc. or its Affiliates. All rights reserved.
A Cycle for AI
Perception
Learning
Knowledge
Representation
(KR)
Reasoning
Planning
Execution
Vision
Language
Speech
Sensors
7. @dmbanga @apwenchel© 2017, Amazon Web Services, Inc. or its Affiliates. All rights reserved.
A Cycle for AI
Perception
Learning
Knowledge
Representation
(KR)
Reasoning
Planning
Execution
Vision
Language
Speech
Sensors
Brain (adaptive)
Memory (static)
8. @dmbanga @apwenchel© 2017, Amazon Web Services, Inc. or its Affiliates. All rights reserved.
A Cycle for AI
Perception
Learning
Knowledge
Representation
(KR)
Reasoning
Planning
Execution
Vision
Language
Speech
Sensors
Brain (adaptive)
Memory (static)
Actuators
9. @dmbanga @apwenchel© 2017, Amazon Web Services, Inc. or its Affiliates. All rights reserved.
A Cycle for AI
Perception
Learning
Knowledge
Representation
(KR)
Reasoning
Planning
Execution
Brain (adaptive)
Memory (static)
Actuators
Camera data
Audio data
Sensor data
Clicks
User activity
10. @dmbanga @apwenchel© 2017, Amazon Web Services, Inc. or its Affiliates. All rights reserved.
A Cycle for AI
Perception
Learning
Knowledge
Representation
(KR)
Reasoning
Planning
Execution
Brain (adaptive)
Actuators
Camera data
Audio data
Sensor data
Clicks
User activity
Knowledge Graph
11. @dmbanga @apwenchel© 2017, Amazon Web Services, Inc. or its Affiliates. All rights reserved.
A Cycle for AI
Perception
Learning
Knowledge
Representation
(KR)
Reasoning
Planning
Execution Actuators
Camera data
Audio data
Sensor data
Clicks
User activity
Knowledge Graph
Machine (Deep) Learning
12. @dmbanga @apwenchel© 2017, Amazon Web Services, Inc. or its Affiliates. All rights reserved.
A Cycle for AI
Perception
Learning
Knowledge
Representation
(KR)
Reasoning
Planning
Execution
Camera data
Audio data
Sensor data
Clicks
User activity
Machine (Deep) Learning
Knowledge Graph
AI
14. @dmbanga @apwenchel© 2017, Amazon Web Services, Inc. or its Affiliates. All rights reserved.
The Promise of Artificial Intelligence:
INNOVATION
15. @dmbanga @apwenchel© 2017, Amazon Web Services, Inc. or its Affiliates. All rights reserved.
Solving Some Of The Hardest Problems In Computer Science
Learning Language Perception Problem
Solving
Reasoning
19. @dmbanga @apwenchel© 2017, Amazon Web Services, Inc. or its Affiliates. All rights reserved.
• Applied Research
• Core Research
• Alexa
• Demand Forecasting
• Risk Analytics
• Search
• Recommendations
• AI Services
• Q&A Systems
• Supply Chain Optimization
• Advertising
• Machine Translation
• Video Content Analysis
• Robotics
• Lots of Computer Vision..
• NLP / NLU
Over 20 years of AI at Amazon…
24. @dmbanga @apwenchel© 2017, Amazon Web Services, Inc. or its Affiliates. All rights reserved.
TP2-AI
Tools, People and Processes for AI
25. @dmbanga @apwenchel© 2017, Amazon Web Services, Inc. or its Affiliates. All rights reserved.
Organizational Processes
How does machine learning fit in your organization?
26. @dmbanga @apwenchel© 2017, Amazon Web Services, Inc. or its Affiliates. All rights reserved.
Data Visualization &
Analysis
Business Problem –
ML problem framing Data Collection
Data Integration
Data Preparation &
Cleaning
Feature Engineering
Model Training &
Parameter Tuning
Model Evaluation
Are business
goals met?
Model Deployment
Monitoring &
Debugging
– Predictions
YesNo
DataAugmentation
Feature
Augmentation
The Machine Learning Process
Re-training
27. @dmbanga @apwenchel© 2017, Amazon Web Services, Inc. or its Affiliates. All rights reserved.
Data Visualization &
Analysis
Business Problem –
ML problem framing Data Collection
Data Integration
Data Preparation &
Cleaning
Feature Engineering
Model Training &
Parameter Tuning
Model Evaluation
Are business
goals met?
Model Deployment
Monitoring &
Debugging
– Predictions
YesNo
DataAugmentation
Feature
Augmentation
Discovery: The Analysts
Re-training
• Help formulate the right
questions
• Domain Knowledge
28. @dmbanga @apwenchel© 2017, Amazon Web Services, Inc. or its Affiliates. All rights reserved.
Data Visualization &
Analysis
Business Problem –
ML problem framing Data Collection
Data Integration
Data Preparation &
Cleaning
Feature Engineering
Model Training &
Parameter Tuning
Model Evaluation
Are business
goals met?
Model Deployment
Monitoring &
Debugging
– Predictions
YesNo
DataAugmentation
Feature
Augmentation
Integration: The Data Architects
Re-training
• Build the data platform:
• Amazon S3
• AWS Glue
• Amazon Athena
• Amazon EMR
• Amazon Redshift
& Spectrum
29. @dmbanga @apwenchel© 2017, Amazon Web Services, Inc. or its Affiliates. All rights reserved.
Data Visualization &
Analysis
Business Problem –
ML problem framing Data Collection
Data Integration
Data Preparation &
Cleaning
Feature Engineering
Model Training &
Parameter Tuning
Model Evaluation
Are business
goals met?
Model Deployment
Monitoring &
Debugging
– Predictions
YesNo
DataAugmentation
Feature
Augmentation
Re-training
Modeling: The Data Scientists
• Builds the ML Models:
• Deep Learning AMI
• SparkML on EMR
• AI Services
• Amazon ML
30. @dmbanga @apwenchel© 2017, Amazon Web Services, Inc. or its Affiliates. All rights reserved.
Data Visualization &
Analysis
Business Problem –
ML problem framing Data Collection
Data Integration
Data Preparation &
Cleaning
Feature Engineering
Model Training &
Parameter Tuning
Model Evaluation
Are business
goals met?
Model Deployment
Monitoring &
Debugging
– Predictions
YesNo
DataAugmentation
Feature
Augmentation
Re-training
Production: The SDE and the DevOps
• Build Smart Apps
• AWS Lambda
• Amazon S3
• Amazon API Gateway
• AWS IoT
• Amazon Kinesis
• Amazon ECS/ECR
• Mobile Hub
• AWS KMS
• Amazon EC2
• More…
31. @dmbanga @apwenchel© 2017, Amazon Web Services, Inc. or its Affiliates. All rights reserved.
Security & Encryption Integrated Across The Platform
Fine grained
access controls
Broad KMS
integration
Server-side
encryption
with CMK
Audit key
usage by
user & role
Import
keys
Policy
validation
& simulation
32. @dmbanga @apwenchel© 2017, Amazon Web Services, Inc. or its Affiliates. All rights reserved.
Key AWS Certifications and Assurance Programs
33. @dmbanga @apwenchel© 2017, Amazon Web Services, Inc. or its Affiliates. All rights reserved.
Intelligent Services Powered By Deep Learning
Amazon AI
34. @dmbanga @apwenchel© 2017, Amazon Web Services, Inc. or its Affiliates. All rights reserved.
More in
2017
Infrastructure CPU
Engines MXNet TensorFlow Caffe Theano Pytorch CNTK
Services
Amazon Polly
Platforms
IoT
Speech
Mobile
Amazon
ML
Spark &
EMR
Kinesis Batch ECS
GPU
More in
2017
Chat
Amazon LexAmazon Rekognition
Vision
Amazon AI: Machine Learning In The Hands Of Every Developer
35. @dmbanga @apwenchel© 2017, Amazon Web Services, Inc. or its Affiliates. All rights reserved.
Machine Learning in the Enterprise
…but how?
37. @dmbanga @apwenchel© 2017, Amazon Web Services, Inc. or its Affiliates. All rights reserved.
“The future is already here,
it’s just not very evenly distributed.”
—William Gibson
39. @dmbanga @apwenchel© 2017, Amazon Web Services, Inc. or its Affiliates. All rights reserved.
Remember the infrastructure Built on AWS ML takes time, and technical debt
Democratize AI, responsibly Maximize scarce experts’ productivity There is a lot more than the ML model
Machine Learning @ Capital One
40. @dmbanga @apwenchel© 2017, Amazon Web Services, Inc. or its Affiliates. All rights reserved.
The road to production is long and arduous
Prepare for the journey
41. @dmbanga @apwenchel© 2017, Amazon Web Services, Inc. or its Affiliates. All rights reserved.
Model selection and Hyper-
parameter Tuning
GPU Optimization Self-service
Production is a hard place Continuous Monitoring Rapid ML model refit/deploy
Capital One AI: The Road to Production
42. @dmbanga @apwenchel© 2017, Amazon Web Services, Inc. or its Affiliates. All rights reserved.
Machine Learning Technology @ Capital One
Built on AWS
43. @dmbanga @apwenchel© 2017, Amazon Web Services, Inc. or its Affiliates. All rights reserved.
Amazon AI AI Platforms AI Engines & InfraAI Services
Capital One AI: Democratizing ML in a Well-Managed Way
Experiment
Management
Logging Versioning Reproducibility Monitoring
Optimize &
Scale
Model Optimization GPU saturation
Data/Compute
coordination
CapitalOne
44. @dmbanga @apwenchel© 2017, Amazon Web Services, Inc. or its Affiliates. All rights reserved.
Organizing for Success
You want a Data Science Team, then what ?
45. @dmbanga @apwenchel© 2017, Amazon Web Services, Inc. or its Affiliates. All rights reserved.
People
Machine Learning is inherently Interdisciplinary
Physical
Scientists
Analysts
Architects/
Integrators
Computer
Scientists
46. @dmbanga @apwenchel© 2017, Amazon Web Services, Inc. or its Affiliates. All rights reserved.
People
Physical
Scientists
Analysts
Architects/
Integrators
Computer
Scientists
Used to solving
problems by
applying ML to
sensors data
Machine Learning is inherently Interdisciplinary
47. @dmbanga @apwenchel© 2017, Amazon Web Services, Inc. or its Affiliates. All rights reserved.
People
Machine Learning is inherently Interdisciplinary
Physical
Scientists
Analysts
Architects/
Integrators
Computer
Scientists
Used to solving
problems by
applying ML to
sensors data
Understand
Business
Problem
Framing
48. @dmbanga @apwenchel© 2017, Amazon Web Services, Inc. or its Affiliates. All rights reserved.
People
Machine Learning is inherently Interdisciplinary
Physical
Scientists
Analysts
Architects/
Integrators
Computer
Scientists
Used to solving
problems by
applying ML to
sensors data
Understand
Business
Problem
Framing
Understand
Computer
Logic and ML
Science
49. @dmbanga @apwenchel© 2017, Amazon Web Services, Inc. or its Affiliates. All rights reserved.
People
Machine Learning is inherently Interdisciplinary
Physical
Scientists
Analysts
Architects/
Integrators
Computer
Scientists
Used to solving
problems by
applying ML to
sensors data
Understand
Business
Problem
Framing
Understand
Computer
Logic and ML
Science
Can put
everything
together
50. @dmbanga @apwenchel© 2017, Amazon Web Services, Inc. or its Affiliates. All rights reserved.
Capital One AI: How We Organize for Success
Create a positive work environment for
ML talent
Centralize new technologiesAttract and motivate the right talent
52. @dmbanga @apwenchel© 2017, Amazon Web Services, Inc. or its Affiliates. All rights reserved.
Opportunity: Transform Financial Services
Increase efficiency, creating value for our
customers
New ways of empowering customers
to take control of their financial lives
Create amazing customer experiences
54. The Deep Learning Revolution
Terrence Sejnowski, The Salk Institute for Biological Studies
Eye, Robot: Computer Vision and Autonomous
Robotics
Aaron Ames & Pietro Perona, California Institute of Technology
Exploiting the Power of Language
Alexander Smola, Amazon Web Services
Reducing Supervision: Making More with Less
Martial Hebert, Carnegie Mellon University
Learning Where to Look in Video
Kristen Grauman, University of Texas
Look, Listen, Learn: The Intersection of Vision and
Sound
Antonio Torralba, MIT
Investing in the Deep Learning Future
Matt Ocko, Data Collective Venture Capital
Thursday, November 30th
1:00 - 5:00pm | Venetian, Ballroom F
https://reinvent.awsevents.com/learn/deep-learning-summit/
55. @dmbanga @apwenchel© 2017, Amazon Web Services, Inc. or its Affiliates. All rights reserved. @dmbanga @apwenchel
Thank you!
d m m b a n g a @ a m a z o n . c o m