Upgrade to Pro — share decks privately, control downloads, hide ads and more …

[第43回 Machine Learning 15minutes! × 2] Azure AI...

[第43回 Machine Learning 15minutes! × 2] Azure AI Updates

[第43回 Machine Learning 15minutes! × 2] Azure AI Updates

https://satonaoki.wordpress.com/2020/01/27/ml15min-azure-ai-updates/

第43回 Machine Learning 15minutes! × 2 (2020/01/25)

https://machine-learning15minutes.connpass.com/event/154965/

SATO Naoki (Neo)

January 25, 2020
Tweet

More Decks by SATO Naoki (Neo)

Other Decks in Technology

Transcript

  1. 90% reduced time from documents to insights $300K Cost savings

    per day 40x More insurance cases reviewed 200K Customers interacting with chatbots +20% Increase in customer satisfaction +32% Increase in sales
  2. MSR Beijing MSR Cambridge MSR Redmond MSR Montreal MSR New

    England MSR New York Fueled by Microsoft breakthrough research 96% on RESNET vision test 94.9% on Switchboard test 89.4% on Stanford CoQA test 69.9% with MT Research system 39.5 Teraflops with Intel Stratix 10 MSR India MSR Shanghai Machine translation human parity Object detection human parity Switchboar d Switchbo ard cellular Meeting speech IBM Switchboard Broadcast speech Speech recognition human parity Conversational Q&A human parity First FPGA deployed in a datacenter
  3. 435K Ad Insight recommendations daily 1B PowerPoint Designer slides used

    Tested at scale in Microsoft solutions 80M Personalized experiences delivered daily Machine translation human parity Object detection human parity Switchboard Switchboard cellular Meeting speech IBM Switchboard Broadcast speech Speech recognition human parity Conversational Q&A human parity First FPGA deployed in a datacenter
  4. For all skill levels Automated ML + drag & drop

    + code Industry leading MLOps Open & Interoperable Trusted
  5. For all skill levels Industry leading MLOps Open & Interoperable

    Trusted Model explainability and fairness
  6. Stream processing Dashboard & application Application Data lake storage Data

    IoT Hub Device at well site Azure Machine Learning Step 1 Build and train
  7. Stream processing Dashboard & application Application Data lake storage Data

    IoT Hub Device at well site Azure Machine Learning Step 1 Build and train Step 2 Package and deploy
  8. Stream processing Dashboard & application Application Data lake storage Data

    IoT Hub Device at well site Azure Machine Learning Step 1 Build and train Step 2 Package and deploy
  9. Stream processing Dashboard & application Application Data lake storage Data

    IoT Hub Device at well site Azure Machine Learning Model telemetry data Step 1 Build and train Step 2 Package and deploy Step 3 Monitor
  10. Stream processing Dashboard & application Application Data lake storage Data

    IoT Hub Device at well site Azure Machine Learning Model telemetry data Step 3 Monitor Step 2 Package and deploy Step 1 Build and train Sensor data
  11. Azure Machine Learning Step 1 Build and train POS data

    Power BI Inventory ordering system Data Lake Storage Azure Synapse Analytics
  12. POS data Power BI Inventory ordering system Data Lake Storage

    Azure Synapse Analytics Azure Machine Learning Step 1 Build and train Step 2 Package and deploy
  13. POS data Power BI Inventory ordering system Data Lake Storage

    Azure Synapse Analytics Azure Machine Learning Step 1 Build and train Step 2 Package and deploy Forecasting
  14. Step 1 Build and train Step 2 Package and deploy

    Step 3 Monitor POS data Power BI Inventory ordering system Data Lake Storage Azure Synapse Analytics Azure Machine Learning Forecasting
  15. Web search Enable your apps to see, hear, speak, understand,

    and interpret user needs Vision Speech Decision Language
  16. Web search Vision Speech Decision Personalizer Text Analytics D.O.B.: 08/04/57

    Phone: 555-8367 Email: [email protected] Language New sentiment analysis model Industry-leading NLP capabilities
  17. Web search Vision Speech Decision Personalizer X Y Introducing Form

    Recognizer Feedback Loop Language New sentiment analysis model 100+ new entity types
  18. Kubernetes Web search Vision Speech Decision Personalizer “Kubernetes” Cucumbers Kubernetes

    Introducing Custom Speech with Office 365 data Language New sentiment analysis model 100+ new entity types Form Recognizer Feedback Loop
  19. Web search Vision Speech Decision Personalizer Custom Speech with Office

    365 data Introducing Custom Neural Voice capability Language New sentiment analysis model 100+ new entity types Form Recognizer Feedback Loop
  20. Announcing new capabilities Power Virtual Agents Build with most popular

    open source tools Deploy and manage across multiple channels
  21. Azure Storage or file system Customer calls Speech to Text

    Text Analytics Sentiment Analysis Key Phrase Extraction Call intervention or escalations Batch processing Power BI App Portal Speech Real-time processing Named Entity Recognition
  22. Ingest Enrich New data sources TranslationNEW AML models contents 2

    3 1 document primaryImage secondaryImage History Origin Construction Design Material Aesthetics Tourism Access Popularity
  23. Ingest Background on Eiffel Tower Enrich Explore contents 2 3

    1 document primaryImage secondaryImage History Origin Construction Design Material Aesthetics Tourism Access Popularity
  24. Key phrase extraction Language detection Automatic translation Named entity extraction

    Clause extraction Risk identifier Azure Web App Enrich Explore Cognitive skills Custom skills Azure Cognitive Search Search Index PDFs Ingest Images Azure Blob Storage
  25. Ingest Explore Magazine metadata Magazine images Image tagging Geo-point extractor

    Custom object detector Similar image tagger Biography enricher Search Index Azure Web App Enrich Azure Cognitive Search Azure Blob Storage