Become a machine learning developer using AWS Machine Learning Services [AWS Summit @ Warsaw]

Transcript

1. © 2019, Amazon Web Services, Inc. or its affiliates. All rights reserved.
   S U M M I T
   Become a machine learning
   developer using AWS Machine
   Learning Services
   Alex Casalboni
   Technical Evangelist, AWS
   

   View Slide

2. S U M M I T
   About me
   • Software Engineer & Web Developer
   • Data science background
   • Worked in a startup for 4.5 years
   • AWS Customer since 2013
   

   View Slide

3. S U M M I T © 2019, Amazon Web Services, Inc. or its affiliates. All rights reserved.
   Put machine learning in the
   hands of every developer
   Our mission at AWS
   

   View Slide

4. © 2019, Amazon Web Services, Inc. or its affiliates. All rights reserved.
   S U M M I T
   The picture can't be displayed.
   Some of our machine learning customers…
   

   View Slide

5. S U M M I T
   M L F R A M E W O R K S &
   I N F R A S T R U C T U R E
   The Amazon ML Stack
   A I S E R V I C E S
   R E K O G N I T I O N
   I M A G E
   P O L L Y T R A N S C R I B E T R A N S L A T E C O M P R E H E N D
   C O M P R E H E N D
   M E D I C A L
   L E X
   R E K O G N I T I O N
   V I D E O
   Vision Speech Chatbots
   A M A Z O N S A G E M A K E R
   B U I L D T R A I N
   F O R E C A S T
   T E X T R A C T P E R S O N A L I Z E
   D E P L O Y
   Pre-built algorithms & notebooks
   Data labeling (G R O U N D T R U T H )
   One-click model training & tuning
   Optimization ( N E O )
   One-click deployment & hosting
   M L S E R V I C E S
   F r a m e w o r k s I n t e r f a c e s I n f r a s t r u c t u r e
   E C 2 P 3
   & P 3 d n
   E C 2 C 5 F P G A s G R E E N G R A S S E L A S T I C
   I N F E R E N C E
   Models without training data (REINFORCEMENT LEARNING)
   Algorithms & models ( A W S M A R K E T P L A C E )
   Language Forecasting Recommendations
   NEW NEW
   NEW
   NEW
   NEW
   NEW
   NEW
   NEW
   NEW
   RL Coach
   

   View Slide

6. © 2019, Amazon Web Services, Inc. or its affiliates. All rights reserved.
   S U M M I T
   Machine Learning
   Training &
   Certification
   AWS DeepRacer
   AWS DeepLens
   Amazon SageMaker
   

   View Slide

7. S U M M I T © 2019, Amazon Web Services, Inc. or its affiliates. All rights reserved.
   Over 230 algorithms and models that
   can be deployed directly to Amazon SageMaker
   

   View Slide

8. AWS Marketplace for Machine Learning
   ML algorithms and models available instantly
   K E Y F E A T U R E S
   Automatic labeling via machine learning
   IP protection
   Automated billing and metering
   S E L L E R S
   Broad selection of paid, free, and
   open-source algorithms and models
   Data protection
   Discoverable on your AWS bill
   B U Y E R S
   

   View Slide

9. S U M M I T © 2019, Amazon Web Services, Inc. or its affiliates. All rights reserved.
   Pre-configured environments to quickly
   build deep learning applications
   

   View Slide

10. © 2019, Amazon Web Services, Inc. or its affiliates. All rights reserved.
    S U M M I T
    AWS is framework agnostic
    Choose from popular frameworks
    Run them fully managed Or run them yourself
    

    View Slide

11. © 2019, Amazon Web Services, Inc. or its affiliates. All rights reserved.
    S U M M I T
    The best place to run TensorFlow
    Fastest time
    for TensorFlow
    65% 90%
    30m 14m
    • 85% of TensorFlow workloads in the
    cloud runs on AWS (2018 Nucleus
    report)
    • Available w/ Amazon SageMaker and
    the AWS Deep Learning AMIs
    

    View Slide

12. S U M M I T © 2019, Amazon Web Services, Inc. or its affiliates. All rights reserved.
    Reduce deep learning inference costs up to 75%
    

    View Slide

13. © 2019, Amazon Web Services, Inc. or its affiliates. All rights reserved.
    S U M M I T
    Amazon Elastic Inference
    Reduce deep learning inference costs up to 75%
    K E Y F E A T U R E S
    Integrated with
    Amazon EC2 and
    Amazon SageMaker
    Support for TensorFlow,
    Apache MXNet -
    PyTorch coming soon
    Single and
    mixed-precision
    operations
    

    View Slide

14. S U M M I T © 2019, Amazon Web Services, Inc. or its affiliates. All rights reserved.
    

    View Slide

15. Machine Learning Life Cycle
    Experimentation
    • Setup and manage Notebooks
    • Get data to notebooks securely
    Training
    • Setup and manage clusters
    • Scale/distribute ML algorithms
    Deployment
    • Setup and manage
    inference clusters
    • Manage and auto scale
    inference APIs
    • Testing, versioning,
    and monitoring
    Fetch data
    Clean & format
    data
    Prepare &
    transform data
    Train model
    Evaluate model
    Integrate with
    prod
    Monitor/
    debug/refresh
    6–18
    months
    Data Wrangling
    • Manage data ingestion
    • Execute ETL
    

    View Slide

16. Amazon SageMaker
    Fully managed
    hosting with auto-
    scaling
    One-click
    deployment
    Pre-built
    notebooks for
    common problems
    Built-in, high
    performance
    algorithms
    One-click
    training
    B U I L D T R A I N & T U N E D E P L O Y
    Hyperparameter
    optimization
    

    View Slide

17. Amazon SageMaker algorithms
    BlazingText
    DeepAR Forecasting
    Image Classification
    Object Detection
    IP Insights
    K-Means & K-Nearest Neighbors
    Latent Dirichlet Allocation (LDA)
    Principal Component Analysis (PCA)
    Linear Learner
    Neural Topic Model (NTM)
    Factorization Machines
    Object2Vec
    Random Cut Forest (RCF)
    Semantic Segmentation
    Sequence-to-Sequence
    XGBoost
    docs.aws.amazon.com/sagemaker/latest/dg/algos.html
    

    View Slide

18. But what happens before building
    and training a model?
    

    View Slide

19. Types of Machine Learning
    Reinforcement learning Supervised learning
    Unsupervised learning
    AMOUNT OF DATA
    SOPHISTICATION
    Supervised learning
    

    View Slide

20. Supervised Learning
    «The task of learning a function that
    maps an input to an output based on
    example input-output pairs»
    «The task of learning a function that
    maps an input to an output based on
    example input-output pairs»
    

    View Slide

21. Public datasets on the web
    registry.opendata.aws (100+, already on Amazon S3)
    github.com/awesomedata/awesome-public-datasets (600+)
    archive.ics.uci.edu (400+)
    www.data.gov (200k+)
    openneuro.org (200+)
    

    View Slide

22. S U M M I T © 2019, Amazon Web Services, Inc. or its affiliates. All rights reserved.
    Build highly accurate training datasets and
    reduce data labeling costs by up to 70%
    

    View Slide

23. Data in S3 Mechanical Turk (public)
    Your own employees (private)
    Third-party labelers (vendor)
    Human
    annotations
    Training
    data
    How it works (1)
    What if we have 1M+ images to annotate?
    

    View Slide

24. Data in S3
    Automatic
    annotations
    Human
    annotations
    Training
    data
    Active
    Learning
    model
    >80% confidence
    <80% confidence
    How it works (2)
    Goal: reduce data labeling costs by up to 70%
    

    View Slide

25. View Slide

26. View Slide

27. Types of Machine Learning
    Reinforcement learning Supervised learning
    Unsupervised learning
    AMOUNT OF DATA
    SOPHISTICATION
    Reinforcement learning
    

    View Slide

28. Reinforcement Learning Use Cases
    AUTONOMOUS CARS FINANCIAL TRADING DATACENTER COOLING
    FLEET LOGISTICS
    

    View Slide

29. S U M M I T © 2019, Amazon Web Services, Inc. or its affiliates. All rights reserved.
    

    View Slide

30. © 2019, Amazon Web Services, Inc. or its affiliates. All rights reserved.
    S U M M I T
    AWS DeepRacer Car Specifications
    CAR 18th scale 4WD with monster truck chassis
    CPU Intel Atom™ Processor
    MEMORY 4GB RAM
    STORAGE 32GB (expandable)
    WI-FI 802.11ac
    CAMERA 4 MP camera with MJPEG
    DRIVE BATTERY 7.4V/1100mAh lithium polymer
    COMPUTE BATTERY 13600mAh USB-C PD
    SENSORS Integrated accelerometer and gyroscope
    PORTS 4x USB-A, 1x USB-C, 1x Micro-USB, 1x HDMI
    SOFTWARE Ubuntu OS 16.04.3 LTS, Intel® OpenVINO™
    toolkit, ROS Kinetic
    

    View Slide

31. 3D simulator with
    virtual track
    Reward
    function
    Reinforcement
    learning algorithm
    How does it work?
    

    View Slide

32. Build reinforcement learning model
    DeepRacer League Races at AWS Summits
    Winners of each DRL Race and top points getters
    compete in Championship Cup at re:Invent 2019
    Virtual tournaments through the year
    AWS DeepRacer League
    World’s first global autonomous
    racing league, open to anyone
    

    View Slide

33. AWS DeepRacer League
    World’s first global autonomous
    racing league, open to anyone
    aws.amazon.com/deepracer/league/points-and-prizes/
    

    View Slide

34. Rf(x) {
    "all_wheels_on_track": bool,
    "x": float,
    "y": float,
    "distance_from_center": float,
    "is_left_of_center": bool,
    "heading": float,
    "progress": float,
    "steps": int,
    "speed": float,
    "steering_angle": float,
    "track_width": float,
    "waypoints": [[float, float], … ],
    "closest_waypoints": [int, int]
    }
    

    View Slide

35. Rf(x) def reward_function(params):
    # Read input parameters
    track_width = params['track_width’]
    distance_from_center = params['distance_from_center’]
    all_wheels_on_track = params['all_wheels_on_track’]
    speed = params['speed']
    marker = 0.2 * track_width
    if not all_wheels_on_track:
    reward = 0.001
    else:
    if distance_from_center <= marker:
    reward = 1.0
    else:
    reward = 0.1
    reward *= speed
    return float(reward)
    

    View Slide

36. View Slide

37. ai.aws & ml.aws
    aws.training/machinelearning
    github.com/awslabs/amazon-sagemaker-examples
    

    View Slide

38. S U M M I T © 2019, Amazon Web Services, Inc. or its affiliates. All rights reserved.
    Alex Casalboni
    Technical Evangelist, AWS
    

    View Slide

39. S U M M I T © 2019, Amazon Web Services, Inc. or its affiliates. All rights reserved.
    

    View Slide