The document discusses the development of recommendation engines using Amazon SageMaker, detailing its techniques and customer use cases, including improvements in system building and training time. It emphasizes the importance of recommendation engines for enhancing customer experience across various fields like e-commerce, finance, and education. Additionally, it highlights how Amazon SageMaker simplifies the machine learning process for developers and data scientists.

1. © 2018, Amazon Web Services, Inc. or its affiliates. All rights reserved.
Yotam Yarden
Data Scientist, Amazon Web Services
AWS Pop UP Loft Berlin 2018
Build a Recommendation Engine on AWS
Today

2. © 2018, Amazon Web Services, Inc. or its affiliates. All rights reserved.
Agenda
• Recommendation Engine – Why?
• Recommendation Engine – Common Techniques
• Introducing Amazon SageMaker
• Customer use cases
• Develop, Train & Deploy a Recommendation Engine

3. © 2018, Amazon Web Services, Inc. or its affiliates. All rights reserved.
Artificial Intelligence
At Amazon (1995)

4. © 2018, Amazon Web Services, Inc. or its affiliates. All rights reserved.
And today…
My Profile – amazon.de My Profile – amazon.com

5. © 2018, Amazon Web Services, Inc. or its affiliates. All rights reserved.
• Personalize and enhance customer
experience
• Different goals:
• Increased time spent on a platform
• Suggest complementary items
• Customer satisfaction
Motivation

6. © 2018, Amazon Web Services, Inc. or its affiliates. All rights reserved.
Use Cases
Ecommerce:
• Amazon.com
Content:
• Movies (Netflix)
• Music (Amazon Music)
• Articles (The Global And Mail)
Finance:
• Services Recommendation
• Stocks buying / selling
• Relevant news and stock related data
Education:
• Courses recommendations
Legal:
• Similar cases

7. © 2018, Amazon Web Services, Inc. or its affiliates. All rights reserved.
Agenda
• Recommendation Engine – Why?
• Recommendation Engine – Common Techniques
• Introducing Amazon SageMaker
• Customer use cases
• Develop, Train & Deploy a Recommendation Engine

8. © 2018, Amazon Web Services, Inc. or its affiliates. All rights reserved.
https://www.oreilly.com/ideas/deep-matrix-factorization-using-apache-mxnet?cmp=tw-data-na-article-engagement_sponsored+kibird

9. © 2018, Amazon Web Services, Inc. or its affiliates. All rights reserved.
Supervised Machine Learning
All Labeled Data
Train Test
Model Training Model
Labels
Test Set
Predictions
|Predictions – True Labels|
= Accuracy

10. © 2018, Amazon Web Services, Inc. or its affiliates. All rights reserved.
https://www.oreilly.com/ideas/deep-matrix-factorization-using-apache-mxnet?cmp=tw-data-na-article-engagement_sponsored+kibird
Testset
Test / Validation

11. © 2018, Amazon Web Services, Inc. or its affiliates. All rights reserved.
Naïve approach
Linear model? [type of user, movie genre, etc.]
Polynomial model? [+interactions]

12. © 2018, Amazon Web Services, Inc. or its affiliates. All rights reserved.
Matrix Factorization
X≈
UserEmbeddings
Item Embeddings

13. © 2018, Amazon Web Services, Inc. or its affiliates. All rights reserved.
Matrix Factorization – “Neural Networks”
Representation

14. © 2018, Amazon Web Services, Inc. or its affiliates. All rights reserved.
Deep Matrix Factorization

15. © 2018, Amazon Web Services, Inc. or its affiliates. All rights reserved.

16. © 2018, Amazon Web Services, Inc. or its affiliates. All rights reserved.
Binary Predictions

17. © 2018, Amazon Web Services, Inc. or its affiliates. All rights reserved.
Binary Predictions
+Negative Sampling
Negative

18. © 2018, Amazon Web Services, Inc. or its affiliates. All rights reserved.
Most of the Data is Still Untapped
• Images
• Titles
• Descriptions
• Reviews
• Episode Names

19. © 2018, Amazon Web Services, Inc. or its affiliates. All rights reserved.
DSSM – Deep Structures Semantic
Models
User
Embedding
Item
Embedding
⨀
⨀⨁ score
output
user Search
BOW
title words
BOW
resnet: imgs
dropout
dense densedensedensedense
concat concat
densedense

20. © 2018, Amazon Web Services, Inc. or its affiliates. All rights reserved.
Which Technique to Choose? Roadmap Matrix
Iterative
process
   
Data Available Limited user data
Binary user-item
interaction
User data
Additional user-item interaction
More user data
Extensive item
data
Extensive user
data
Extensive item
data
Relevant
Algorithms
Matrix Factorization
Binary
Matrix Factorization
Factorization Machines
DiFacto
DSSM Customized and
more advanced
DSSM
Relative
Complexity
2 4 5 5
Deployment
Considerations
 Historical data size – 30d / 60d / 1y…
 Fine-tuning techniques (daily, weekly..)
 Inference - compressed model? Tradeoff between model complexity and inference latency
 Validation system setup
 Iterate fast and simple

21. © 2018, Amazon Web Services, Inc. or its affiliates. All rights reserved.
Agenda
• Recommendation Engine – Why?
• Recommendation Engine – Common Techniques
• Introducing Amazon SageMaker
• Customer use cases
• Develop, Train & Deploy a Recommendation Engine

22. © 2018, Amazon Web Services, Inc. or its affiliates. All rights reserved.
ML @ AWS: Our mission
Put machine learning in the hands of every developer
and data scientist

23. © 2018, Amazon Web Services, Inc. or its affiliates. All rights reserved.
Customer Running ML on AWS Today

24. © 2018, Amazon Web Services, Inc. or its affiliates. All rights reserved.
ML is still too complicated for developers and data
scientists
Collect and prepare
training data
Choose and
optimize your ML
algorithm
Set up and manage
environments for
training
Train and tune model
(trial and error)
Deploy model
in production
Scale and manage
the production
environment

25. © 2018, Amazon Web Services, Inc. or its affiliates. All rights reserved.
A m a z o n S a g e M a k e r
Eas ily build, tr ain, and deploy
mac hine lear ning models

26. © 2018, Amazon Web Services, Inc. or its affiliates. All rights reserved.
Amazon SageMaker
Pre-built
notebooks for
common
problems
BUILD
Choose and
optimize your ML
algorithm
Set up and manage
environments for
training
Train and tune model
(trial and error)
Deploy model
in production
Scale and manage
the production
environment

27. © 2018, Amazon Web Services, Inc. or its affiliates. All rights reserved.
Pre-built
notebooks for
common
problems
K-Means Clustering
Principal Component Analysis
Neural Topic Modelling
Factorization Machines
Linear Learner - Regression
XGBoost
Latent Dirichlet Allocation
Image Classification
Seq2Seq
Linear Learner - Classification
ALGORITHMS
Apache MXNet
TensorFlow
Caffe2, CNTK,
PyTorch, Torch
FRAMEWORKS
Set up and m anage
environments for
training
Train and tune
m odel (trial and
error)
Deploy m odel
in production
Scale and m anage the
production environment
Built-in, high
performance
algorithms
BUILD
Amazon SageMaker

28. © 2018, Amazon Web Services, Inc. or its affiliates. All rights reserved.
Pre-built
notebooks for
common
problems
Built-in, high
performance
algorithms
One-click
training
BUILD TRAIN
Train and tune model
(trial and error)
Deploy model
in production
Scale and manage
the production
environment
Amazon SageMaker

29. © 2018, Amazon Web Services, Inc. or its affiliates. All rights reserved.
Pre-built
notebooks for
common
problems
Built-in, high
performance
algorithms
One-click
training
Hyperparameter
optimization
BUILD TRAIN
Deploy model
in production
Scale and manage
the production
environment
Amazon SageMaker

30. © 2018, Amazon Web Services, Inc. or its affiliates. All rights reserved.
Pre-built
notebooks for
common
problems
Built-in, high
performance
algorithms
One-click
deployment
One-click
training
Hyperparameter
optimization
Scale and manage
the production
environment
BUILD TRAIN DEPLOY
Amazon SageMaker

31. © 2018, Amazon Web Services, Inc. or its affiliates. All rights reserved.
Fully managed
hosting with auto-
scaling
One-click
deployment
Pre-built
notebooks for
common
problems
Built-in, high
performance
algorithms
One-click
training
Hyperparameter
optimization
BUILD TRAIN DEPLOY
Amazon SageMaker

32. © 2018, Amazon Web Services, Inc. or its affiliates. All rights reserved.
Agenda
• Recommendation Engine – Why?
• Recommendation Engine – Common Techniques
• Introducing Amazon SageMaker
• Customer Use Cases
• Develop, Train & Deploy a Recommendation Engine

33. © 2018, Amazon Web Services, Inc. or its affiliates. All rights reserved.
Customers Use Case - erento is using Amazon
SageMaker to develop, train, and deploy
recommendation systems in their market place
Erento’s in-house Data Science team is using Amazon SageMaker to build and deploy
ML models to solve item availability and decrease the enquiry-to-offer time through a
recommendation system, which suggests similar items that are available and increases
the chance for a successful booking. Using Amazon SageMaker reduced our
recommendation system building time from half a year to few weeks and
reduced the algorithm training time from hours to few seconds. It also helped us reduce
dependencies between projects, which has streamlined our whole pre-deployment
process.
- Wassim Zoghlami, Data Scientist Engineer at Erento
“
“

34. © 2018, Amazon Web Services, Inc. or its affiliates. All rights reserved.
Agenda
• Recommendation Engine – Why?
• Recommendation Engine – Common Techniques
• Introducing Amazon SageMaker
• Customer use cases
• Develop, Train & Deploy a Recommendation Engine

35. © 2018, Amazon Web Services, Inc. or its affiliates. All rights reserved.
console

36. © 2018, Amazon Web Services, Inc. or its affiliates. All rights reserved.
References
• https://www.oreilly.com/ideas/deep-matrix-factorization-using-
apache-mxnet
• https://github.com/apache/incubator-mxnet
• https://github.com/awslabs/amazon-sagemaker-examples
• https://www.csie.ntu.edu.tw/~b97053/paper/Rendle2010FM.pdf
• https://www.youtube.com/watch?v=cftJAuwKWkA
• https://www.youtube.com/watch?v=1cRGpDXTJC8&t=640s