UPLOAD

    2.2K

    AWS Machine Learning Services - Overview

    Published: March 18, 2019

    AWs Roadshow Presentation - Machine Learning / Artificial Intelligence / Deep Learning - Rekognition (Images / Video) - Polly - Lex - Translate - Comprehend - Transcribe - Forecast - Personalize

    Comments

    AWS Machine Learning Services - Overview

    • 1. AWSArtificial Intelligence | Machine Learning | Deep Learning © 2018, Amazon Web Services, Inc. or its Affiliates. All rights reserved. Amazon Trademark © 2018, Amazon Web Services, Inc. or its Affiliates. All rights reserved. Amazon Confidential and Trademark AWS Artificial Intelligence | Machine Learning | Deep Learning Centerpiece for digital transformation Halil BAHADIR Manager, Solutions Architect
    • 2. Slide4 © 2018, Amazon Web Services, Inc. or its Affiliates. All rights reserved. Amazon Trademark © 2018, Amazon Web Services, Inc. or its Affiliates. All rights reserved. Amazon Confidential and Trademark Centerpiece for digital transformation Customer experience Business operations Decision making Innovation Competitive advantage of digital transformation initiatives supported by AI in 2019 40% - IDC 2018
    • 3. Slide51 © 2018, Amazon Web Services, Inc. or its Affiliates. All rights reserved. Amazon Trademark © 2018, Amazon Web Services, Inc. or its Affiliates. All rights reserved. Amazon Confidential and Trademark Our mission at AWS Put machine learning in the hands of every developer
    • 4. The AI/ML Journey: Why AWS? © 2018, Amazon Web Services, Inc. or its Affiliates. All rights reserved. Amazon Trademark © 2018, Amazon Web Services, Inc. or its Affiliates. All rights reserved. Amazon Confidential and Trademark The AI/ML Journey: Why AWS? At Amazon, we’ve been making investments in ML for the last 20 years… New Retail Experience Search & Discovery Fulfilment & Logistics Existing Products New Initiatives
    • 5. Slide193 © 2018, Amazon Web Services, Inc. or its Affiliates. All rights reserved. Amazon Trademark © 2018, Amazon Web Services, Inc. or its Affiliates. All rights reserved. Amazon Confidential and Trademark Our unique approach Customer-focused 90%+ of our ML roadmap is defined by customers Breadth & depth More AI and ML services in production than any other provider Embedded R&D Customer-centric approach to advancing the state of the art Security & analytics Deepest set of security and encryption features, with robust analytics capabilities Multi-framework Support for the most popular frameworks Pace of innovation 200+ new ML launches in the last year
    • 6. Slide144 © 2018, Amazon Web Services, Inc. or its Affiliates. All rights reserved. Amazon Trademark © 2018, Amazon Web Services, Inc. or its Affiliates. All rights reserved. Amazon Confidential and Trademark More machine learning happens on AWS than anywhere else customers have used machine learning on AWS 10,000+ AWS holds the top spots on Stanford’s deep learning benchmark, DAWN, for fastest training time, lowest cost, lowest inference latency of TensorFlow projects in the cloud run on AWS 85% of deep learning in the cloud runs on AWS 81%
    • 7. Slide202 © 2018, Amazon Web Services, Inc. or its Affiliates. All rights reserved. Amazon Trademark © 2018, Amazon Web Services, Inc. or its Affiliates. All rights reserved. Amazon Confidential and Trademark More machine learning happens on AWS than anywhere else
    • 8. Slide235 © 2018, Amazon Web Services, Inc. or its Affiliates. All rights reserved. Amazon Trademark © 2018, Amazon Web Services, Inc. or its Affiliates. All rights reserved. Amazon Confidential and Trademark ML FRAMEWORKS & INFRASTRUCTURE The Amazon ML stack: Broadest & deepest set of capabilities AI SERVICES REKOGNITION IMAGE POLLY TRANSCRIBE TRANSLATE COMPREHEND LEX REKOGNITION VIDEO Vision Speech Language Chatbots AMAZON SAGEMAKER BUILD TRAIN FORECAST Forecasting TEXTRACT PERSONALIZE Recommendations DEPLOY Pre-built algorithms & notebooks Data labeling (GROUND TRUTH) One-click model training & tuning Optimization (NEO) One-click deployment & hosting ML SERVICES Frameworks Interfaces Infrastructure EC2 P3 & P3N EC2 C5 FPGAs GREENGRASS ELASTIC INFERENCE Reinforcement learning Algorithms & models (AWS MARKETPLACE FOR MACHINE LEARNING)
    • 9. AWS is framework agnostic © 2018, Amazon Web Services, Inc. or its Affiliates. All rights reserved. Amazon Trademark © 2018, Amazon Web Services, Inc. or its Affiliates. All rights reserved. Amazon Confidential and Trademark AWS is framework agnostic Choose from popular frameworks Run them fully managed Or run them yourself
    • 10. AWS: The platform of choice to run TensorFlow © 2018, Amazon Web Services, Inc. or its Affiliates. All rights reserved. Amazon Trademark © 2018, Amazon Web Services, Inc. or its Affiliates. All rights reserved. Amazon Confidential and Trademark AWS: The platform of choice to run TensorFlow 85% of all TensorFlow workloads in the cloud runs on AWS Source: Nucleus Research, November 2018
    • 11. AWS is the best platform for TensorFlow © 2018, Amazon Web Services, Inc. or its Affiliates. All rights reserved. Amazon Trademark © 2018, Amazon Web Services, Inc. or its Affiliates. All rights reserved. Amazon Confidential and Trademark AWS is the best platform for TensorFlow with 256 GPUs 7h 14m with 256 GPUs 6d 3.5h Fast Versatile Scalable Time-to-train (CNN, ResNet50) Time-to-train (LSTM, DeepSpeech2) Within 90% of ideal scaling Within 73% of ideal scaling
    • 12. Apache MXNet: Deep learning for enterprise developers © 2018, Amazon Web Services, Inc. or its Affiliates. All rights reserved. Amazon Trademark © 2018, Amazon Web Services, Inc. or its Affiliates. All rights reserved. Amazon Confidential and Trademark Apache MXNet: Deep learning for enterprise developers Start with off-the-shelf toolkits •Gluon CV and Gluon NLP Fast and scalable training •Keras-MXNet up to 2x faster than Keras-TensorFlow •Near-linear scalability up to 256 GPUs •Dynamic training Easy deployment •Java/Scala APIs •MXNet Model Server
    • 13. MXNet customer example: Chick-fil-A © 2018, Amazon Web Services, Inc. or its Affiliates. All rights reserved. Amazon Trademark © 2018, Amazon Web Services, Inc. or its Affiliates. All rights reserved. Amazon Confidential and Trademark MXNet customer example: Chick-fil-A TRACK WAFFLE FRY FRESHNESS •Identify fries that have exceeded the hold time USES DEEP LEARNING •Computer vision model for object detection and tracking BUILT THE MODEL USING MXNet •A team of enterprise developers with no ML expertise
    • 14. MXNet customer example: GumGum © 2018, Amazon Web Services, Inc. or its Affiliates. All rights reserved. Amazon Trademark © 2018, Amazon Web Services, Inc. or its Affiliates. All rights reserved. Amazon Confidential and Trademark MXNet customer example: GumGum
    • 15. Slide84 © 2018, Amazon Web Services, Inc. or its Affiliates. All rights reserved. Amazon Trademark © 2018, Amazon Web Services, Inc. or its Affiliates. All rights reserved. Amazon Confidential and Trademark
    • 16. Slide88 © 2018, Amazon Web Services, Inc. or its Affiliates. All rights reserved. Amazon Confidential and Trademark
    • 17. Slide90 © 2018, Amazon Web Services, Inc. or its Affiliates. All rights reserved. Amazon Confidential and Trademark
    • 18. Slide95 © 2018, Amazon Web Services, Inc. or its Affiliates. All rights reserved. Amazon Confidential and Trademark
    • 19. Slide96 © 2018, Amazon Web Services, Inc. or its Affiliates. All rights reserved. Amazon Confidential and Trademark
    • 20. Slide1576 © 2018, Amazon Web Services, Inc. or its Affiliates. All rights reserved. Amazon Trademark © 2018, Amazon Web Services, Inc. or its Affiliates. All rights reserved. Amazon Confidential and Trademark ML FRAMEWORKS & INFRASTRUCTURE The Amazon ML stack: Broadest & deepest set of capabilities AI SERVICES REKOGNITION IMAGE POLLY TRANSCRIBE TRANSLATE COMPREHEND LEX REKOGNITION VIDEO Vision Speech Language Chatbots AMAZON SAGEMAKER BUILD TRAIN FORECAST Forecasting TEXTRACT PERSONALIZE Recommendations DEPLOY Pre-built algorithms & notebooks Data labeling (GROUND TRUTH) One-click model training & tuning Optimization (NEO) One-click deployment & hosting ML SERVICES Frameworks Interfaces Infrastructure EC2 P3 & P3N EC2 C5 FPGAs GREENGRASS ELASTIC INFERENCE Reinforcement learning Algorithms & models (AWS MARKETPLACE FOR MACHINE LEARNING)
    • 21. Slide1577 © 2018, Amazon Web Services, Inc. or its Affiliates. All rights reserved. Amazon Trademark © 2018, Amazon Web Services, Inc. or its Affiliates. All rights reserved. Amazon Confidential and Trademark Custom machine learning for your business Amazon SageMaker Fast & accurate data labeling Built-in, high performance algorithms & notebooks BUILD 1 One-click training and tuning TRAIN Model optimization 2 Fully managed hosting with auto-scaling and elastic inference One-click deployment DEPLOY 3
    • 22. Amazon SageMaker © 2018, Amazon Web Services, Inc. or its Affiliates. All rights reserved. Amazon Confidential and Trademark Amazon SageMaker Collect and prepare training data Choose and optimize your ML algorithm Train and Tune ML Models Set up and manage environments for training Deploy models in production Scale and manage the production environment 1 2 3
    • 23. Slide232 © 2018, Amazon Web Services, Inc. or its Affiliates. All rights reserved. Amazon Trademark © 2018, Amazon Web Services, Inc. or its Affiliates. All rights reserved. Amazon Confidential and Trademark Custom machine learning for your business Amazon SageMaker Fast & accurate data labeling Built-in, high performance algorithms & notebooks BUILD 1 One-click training and tuning TRAIN Model optimization 2 Fully managed hosting with auto-scaling and elastic inference One-click deployment DEPLOY 3
    • 24. Slide1450 © 2018, Amazon Web Services, Inc. or its Affiliates. All rights reserved. Amazon Confidential and Trademark Collect and prepare training data Choose and optimize your ML algorithm Train and Tune ML Models Set up and manage environments for training Deploy models in production Scale and manage the production environment Pre-built notebooks for common problems Amazon SageMaker: Build, Train, and Deploy ML Models at Scale 1 2 3
    • 25. Slide1451 © 2018, Amazon Web Services, Inc. or its Affiliates. All rights reserved. Amazon Confidential and Trademark Collect and prepare training data Choose and optimize your ML algorithm Train and Tune ML Models Set up and manage environments for training Deploy models in production Scale and manage the production environment Pre-built notebooks for common problems Amazon SageMaker: Build, Train, and Deploy ML Models at Scale Built-in, high performance algorithms
    • 26. Slide1452 © 2018, Amazon Web Services, Inc. or its Affiliates. All rights reserved. Amazon Confidential and Trademark Collect and prepare training data Choose and optimize your ML algorithm Train and Tune ML Models Set up and manage environments for training Deploy models in production Scale and manage the production environment Pre-built notebooks for common problems Amazon SageMaker: Build, Train, and Deploy ML Models at Scale Built-in, high performance algorithms One-click training on the highest performing infrastructure
    • 27. Slide1453 © 2018, Amazon Web Services, Inc. or its Affiliates. All rights reserved. Amazon Confidential and Trademark Collect and prepare training data Choose and optimize your ML algorithm Train and Tune ML Models Set up and manage environments for training Deploy models in production Scale and manage the production environment Pre-built notebooks for common problems Amazon SageMaker: Build, Train, and Deploy ML Models at Scale Built-in, high performance algorithms One-click training on the highest performing infrastructure Model Optimization
    • 28. Slide1454 © 2018, Amazon Web Services, Inc. or its Affiliates. All rights reserved. Amazon Confidential and Trademark Collect and prepare training data Choose and optimize your ML algorithm Train and Tune ML Models Set up and manage environments for training Deploy models in production Scale and manage the production environment Pre-built notebooks for common problems Amazon SageMaker: Build, Train, and Deploy ML Models at Scale Built-in, high performance algorithms One-click training on the highest performing infrastructure Model Optimization One-click Deployment
    • 29. Slide1455 © 2018, Amazon Web Services, Inc. or its Affiliates. All rights reserved. Amazon Confidential and Trademark Collect and prepare training data Choose and optimize your ML algorithm Train and Tune ML Models Set up and manage environments for training Deploy models in production Scale and manage the production environment Pre-built notebooks for common problems Amazon SageMaker: Build, Train, and Deploy ML Models at Scale Built-in, high performance algorithms One-click training on the highest performing infrastructure Model Optimization One-click Deployment Fully managed with auto-scaling for 75% less
    • 30. Amazon SageMaker Ground Truth © 2018, Amazon Web Services, Inc. or its Affiliates. All rights reserved. Amazon Trademark © 2018, Amazon Web Services, Inc. or its Affiliates. All rights reserved. Amazon Confidential and Trademark Amazon SageMaker Ground Truth Build highly accurate training datasets and reduce data labeling costs by up to 70% using machine learning
    • 31. Slide1460 © 2018, Amazon Web Services, Inc. or its Affiliates. All rights reserved. Amazon Confidential and Trademark Raw Data NEW Human Annotations Automatic Annotations Training Data Human Annotations Amazon SageMaker Ground Truth How it works
    • 32. Slide1461 © 2018, Amazon Web Services, Inc. or its Affiliates. All rights reserved. Amazon Confidential and Trademark Mechanical Turk workers Private labeling workforce Third-party vendors NEW Amazon SageMaker Ground Truth How it works
    • 33. Slide1462 © 2018, Amazon Web Services, Inc. or its Affiliates. All rights reserved. Amazon Confidential and Trademark KEY FEATURES Automatic labeling via machine learning Ready-made and custom workflows Label management Private and public human workforce Amazon SageMaker Ground Truth Label machine learning training data easily and accurately NEW Easily integrate human labelers Get accurate results Quickly label training data
    • 34. Slide217 © 2018, Amazon Web Services, Inc. or its Affiliates. All rights reserved. Amazon Trademark © 2018, Amazon Web Services, Inc. or its Affiliates. All rights reserved. Amazon Confidential and Trademark Custom machine learning for your business One-click model training & deployment 10x better algorithm performance Train once run anywhere AMAZON SAGEMAKER 70% cost reduction for data labeling using Ground Truth 2x performance increases from model optimization with Neo 75% cost reduction for inference with Elastic Inference REDUCE COSTS INCREASE PERFORMANCE EASE-OF-USE
    • 35. Slide1578 © 2018, Amazon Web Services, Inc. or its Affiliates. All rights reserved. Amazon Trademark © 2018, Amazon Web Services, Inc. or its Affiliates. All rights reserved. Amazon Confidential and Trademark ML FRAMEWORKS & INFRASTRUCTURE The Amazon ML stack: Broadest & deepest set of capabilities AI SERVICES REKOGNITION IMAGE POLLY TRANSCRIBE TRANSLATE COMPREHEND LEX REKOGNITION VIDEO Vision Speech Language Chatbots AMAZON SAGEMAKER BUILD TRAIN FORECAST Forecasting TEXTRACT PERSONALIZE Recommendations DEPLOY Pre-built algorithms & notebooks Data labeling (GROUND TRUTH) One-click model training & tuning Optimization (NEO) One-click deployment & hosting ML SERVICES Frameworks Interfaces Infrastructure EC2 P3 & P3N EC2 C5 FPGAs GREENGRASS ELASTIC INFERENCE Reinforcement learning Algorithms & models (AWS MARKETPLACE FOR MACHINE LEARNING)
    • 36. Slide209 © 2018, Amazon Web Services, Inc. or its Affiliates. All rights reserved. Amazon Trademark © 2018, Amazon Web Services, Inc. or its Affiliates. All rights reserved. Amazon Confidential and Trademark Real-time Works with almost any product or content KEY FEATURES Responsive to changes in intent Automated machine learning Bring existing algorithms from Amazon SageMaker Deliver high quality recommendations Deep learning enabled algorithms Easy to Use Amazon Personalize Improve customer experiences with personalization and recommendations
    • 37. Slide208 © 2018, Amazon Web Services, Inc. or its Affiliates. All rights reserved. Amazon Trademark © 2018, Amazon Web Services, Inc. or its Affiliates. All rights reserved. Amazon Confidential and Trademark Activity stream Views, signups, conversion, etc. Inventory Videos, products, articles, etc. Customized Personalization API Demographics (optional) Name, age, location, etc. 1.Load data 2.Inspect data 3.Identify features 4.Select algorithms 5.Select hyperparameters 6.Train models 7.Optimize models 8.Build feature store 9.Deploy and host models 10.Create real-time caches Amazon Personalize Amazon Personalize – How it works
    • 38. Slide258 © 2018, Amazon Web Services, Inc. or its Affiliates. All rights reserved. Amazon Trademark © 2018, Amazon Web Services, Inc. or its Affiliates. All rights reserved. Amazon Confidential and Trademark Amazon Forecast: Machine learning time-series forecasting Any historical time-series Export to CSV to Integrate with SAP and Oracle Supply Chain Custom forecasts with 3 clicks Up to 50% more accurate 1/10th the cost Retail demand Travel demand AWS usage Revenue forecasts Web traffic Advertising demand Generate forecasts for:
    • 39. Amazon TextractExtract text and data from virtually any document © 2018, Amazon Web Services, Inc. or its Affiliates. All rights reserved. Amazon Trademark © 2018, Amazon Web Services, Inc. or its Affiliates. All rights reserved. Amazon Confidential and Trademark Amazon Textract Extract text and data from virtually any document Eliminate manual effort Lower document processing costs KEY FEATURES Extract data quickly and accurately Optical Character Recognition (OCR) Key-value pair detection Adjustable confidence thresholds Table detection Bounding box coordinates No ML experience required
    • 40. Amazon Rekognition Image © 2018, Amazon Web Services, Inc. or its Affiliates. All rights reserved. Amazon Trademark © 2018, Amazon Web Services, Inc. or its Affiliates. All rights reserved. Amazon Confidential and Trademark Amazon Rekognition Image Facial Analysis Face Recognition Text in Image Deep learning-based image analysis service Unsafe Image Detection Object & Scene Detection Celebrity Recognition
    • 41. Amazon Rekognition Video © 2018, Amazon Web Services, Inc. or its Affiliates. All rights reserved. Amazon Trademark © 2018, Amazon Web Services, Inc. or its Affiliates. All rights reserved. Amazon Confidential and Trademark Amazon Rekognition Video Object & Activity Detection Face Detection & Recognition Real-time Live Stream Deep learning-based video analysis service Unsafe Video Detection Celebrity Recognition Pathing
    • 42. Fighting Human Trafficking & Rescuing Victims © 2018, Amazon Web Services, Inc. or its Affiliates. All rights reserved. Amazon Trademark © 2018, Amazon Web Services, Inc. or its Affiliates. All rights reserved. Amazon Confidential and Trademark Fighting Human Trafficking & Rescuing Victims Uses Amazon Rekognition to power their Traffic Jam FaceSearch tool “Without Traffic Jam, investigators are left to manually sift through thousands of online ads. This means they sit at their computer, with a picture of the victim taped to their screen, and compare every photo they see online in the hope that they might find a match. Using AI technology, like Amazon Rekognition, this critical task can now be done with more accuracy and within seconds as compared to days, which is so important in cases where detectives have limited time to find the victim before he or she is moved to the next city.” - Emily Kennedy, President & Co-Founder, Marinus Analytics
    • 43. Fighting Child Sexual Abuse & Trafficking © 2018, Amazon Web Services, Inc. or its Affiliates. All rights reserved. Amazon Trademark © 2018, Amazon Web Services, Inc. or its Affiliates. All rights reserved. Amazon Confidential and Trademark Fighting Child Sexual Abuse & Trafficking Using AWS services, such as Amazon Rekognition, Thorn has seen a 65% reduction in investigation time “Amazon Rekognition has been an incredible partner to Thorn by helping us leverage their image and video analysis solutions in our mission to defend children from sexual abuse. Abusers have hijacked the most advanced technology to exploit children—selling children online for sex, circulating abuse images and videos, and engaging in live-streaming abuse. AWS has chosen to be a part of the solution—partnering to leverage their solutions to help find exploited children faster and stop abuse. Collaborating with leading technology companies like AWS is critical to building the tools we need to find these children faster, and end the spread of child sexual abuse material.” - Julie Cordua, CEO, Thorn
    • 44. Automating Footage Tagging © 2018, Amazon Web Services, Inc. or its Affiliates. All rights reserved. Amazon Trademark © 2018, Amazon Web Services, Inc. or its Affiliates. All rights reserved. Amazon Confidential and Trademark •Built in 3 weeks •Index created in one day •Saved ~9,000 hours a year in manual curation costs •Live video with frame sampling Automating Footage Tagging Previously, only about half of all footage was indexed due to the immense time requirements required by manual processes
    • 45. Slide93 © 2018, Amazon Web Services, Inc. or its Affiliates. All rights reserved. Amazon Trademark © 2018, Amazon Web Services, Inc. or its Affiliates. All rights reserved. Amazon Confidential and Trademark ml.aws