Machine Learning Everywhere
1 like412 views
The document outlines Oracle's strategy for integrating machine learning (ML) across various business functions to enhance decision-making, efficiency, and outcomes. It identifies the challenges faced in enterprise ML, such as fragmented data and collaboration inefficiencies, and promotes Oracle's cloud-based solutions to simplify ML processes for different user roles. Additionally, it highlights pre-built applications and tools designed to improve user experiences and deliver actionable insights across multiple domains.
Machine Learning Everywhere
- 1. Copyright © 2019, Oracle and/or its affiliates. All rights reserved. | 1 Machine Learning Everywhere Stephen Weingartner Solution Engineer Oracle Analytics / Data / IaaS February 6, 2019
- 2. Copyright © 2019, Oracle and/or its affiliates. All rights reserved. | Safe Harbor Statement The following is intended to outline our general product direction. It is intended for information purposes only, and may not be incorporated into any contract. It is not a commitment to deliver any material, code, or functionality, and should not be relied upon in making purchasing decisions. The development, release, timing, and pricing of any features or functionality described for Oracle’s products may change and remains at the sole discretion of Oracle Corporation.
- 3. Copyright © 2019, Oracle and/or its affiliates. All rights reserved What’s the Big Appeal for ML? • Automate thinking, analysis & reason • Faster and better decisions • Improved outcomes • Competitive advantage
- 4. Copyright © 2019, Oracle and/or its affiliates. All rights reserved. | Example Benefits of ML Within an Organisation Marketing HR Individualize product recommendations Identify attrition candidates Manufacturing Autonomously identify root causes to throughput issues Service Automate the repair vs replace decisions IT Operations Detect threats and stay ahead of malicious attacks
- 5. Traditional ML Can Be Expensive, Time Consuming and Complex Copyright © 2019, Oracle and/or its affiliates. All rights reserved • Train model and score performance Model Training and Evaluation • Monitor performance of model versions • Retrain model when performance drops Monitor and Manage Lifecycle • Select algorithms and parameters • Select and transform the data for the algorithm if needed Model Creation • Connect to data and collect data • Cleanse and enrich data in preparation for analysis Data Ingest and Prep • Identify business objectives Business Understanding • Review data to inform model creation and data strategy Data Exploration • Deploy model production into a business process (API, integration, report, etc) Deploy
- 6. Copyright © 2019, Oracle and/or its affiliates. All rights reserved.Copyright © 2019, Oracle and/or its affiliates. All rights reserved. Enterprise ML Challenges Fragmented Usage Data silos limit effectiveness Hundreds of Technologies Multiple incompatible platforms Specialized Skills Expensive and scarce talent Domain knowledge gap Complex model building Technology and Infrastructure Scalability Specialized hardware Model management
- 7. Copyright © 2019, Oracle and/or its affiliates. All rights reserved. Enterprise ML Collaboration Challenges InefficientworkflowsareholdingcompaniesbackfromrealizingthefullpotentialofML Data Scientists Cannot work efficiently App Developers Cannot access usable machine learning IT Admins Spend too much time on support Business Execs Not enough live solutions. Low ROI
- 8. Copyright © 2019, Oracle and/or its affiliates. All rights reserved.Copyright © 2019, Oracle and/or its affiliates. All rights reserved. Oracle’s ML Strategy The more data, the better Lower the skills requirement so it is easily actionable by all Connected entire value chain while optimizing locally Human oversight and control Right domain of expertise “Embed ML in everything that we do”
- 9. PaaS Business Analytics Content and Experience App Dev Data Mngmnt Management IT Analytics Application Performance Monitoring Infrastructure Monitoring Log Analytics Orchestration SecurityIdentity Configuration and Compliance Security Monitoring and Analytics CASB Integration WebCenter Portal Cloud DIVA Cloud Content and Experience Data VisualizationAnalytics Cloud Business Intelligence EssbaseData Science Cloud Machine Learning DeveloperAI Platform BlockchainJava Application Container Mobile and Chatbots Visual Builder API Catalog Messaging Autonomous NoSQL Database Cloud Event Hub MySQLDatabase Database Backup Hadoop / Big Data Sparkline Spark Data Hub Autonomous Data Warehouse Cloud API Platform Apiary Internet of Things Oracle Data Integrator GoldenGate Data Integration Platform Cloud Oracle Integration Cloud Process Automation Managed File Transfer Common SaaS Analytics Cloud Marketplace Adaptive Intelligent Apps API Catalog SaaS ERP Risk MngmntProject Mngmnt ProcurementFinancials Revenue Mngmnt Accounting Hub Project Financial Mngmnt SCM IoT Apps HCM Analytics CRM Analytics ERP Analytics SCM Analytics Industry EPM HCM CX Workforce Mngmnt Work life Solutions Talent Mngmnt Workforce Rewards Global Human Resources Supply Chain Collaboration and Visibility Product Master Data Mngmnt Procurement Product Lifecycle Mngmnt Inventory Mngmnt Logistics Manufac- turing Order Mngmnt Maintenance Supply Chain Planning In-Mem Cost Mngmnt IoT Fleet Monitoring IoT Connected Worker IoT Production Monitoring IoT Asset Monitoring Asset Service Monitoring Planning and Budgeting Profitability and Cost Mngmnt Enterprise Data Mngmnt Enterprise Performance Reporting Enterprise Planning Financial Consolidation and Close Tax Reporting Account Reconciliation Industrial Manufacturing High Technology HospitalityCommunications Consumer Goods And Retail Education and Research Financial Services Utilities Sales Performance Mngmnt CPQ Customer Data Mngmnt Marketing Engagement Sales Service LoyaltyCommerce Social Data Cloud IaaS Infrastructure Classic Dedicated Compute SPARC Model 300 Object Storage Classic Archive Storage Classic Storage Software Appliance VPN for Compute Classic VPN for Dedicated Compute Classic Dedicated Compute Classic Virtual Machine Compute Classic FastConnect Classic Container Service Classic Archive Storage Data Transfer Service Compute Bare Metal GPU Instance Virtual Cloud Network Block Volumes Object Storage Identity and Access Mngmnt Compute VM Instance Compute Bare Metal Instance Audit Tagging Database Bare Metal Exadata Bare Metal Load Balancing DNS Email Delivery Ravello FastConnect Container Pipelines Container Registry Container Engine Cloud at Customer Oracle Cloud: SaaS, PaaS, IaaS ML Everywhere
- 10. Copyright © 2019, Oracle and/or its affiliates. All rights reserved Making ML Easier for: App Developers IT Managers Data ScientistsBusiness Users Business Analysts DBA’s
- 11. Copyright © 2019, Oracle and/or its affiliates. All rights reserved Making ML Easier for Business Users
- 12. Copyright © 2019, Oracle and/or its affiliates. All rights reserved Business Users Can Benefit from Machine Learning Need better predictions and decisions Need faster predictions and decisions Avoid costly and complex ML projects Challenges MACHINE LEARNING
- 13. Copyright © 2019, Oracle and/or its affiliates. All rights reserved.Copyright © 2019, Oracle and/or its affiliates. All rights reserved. Ready-to-Go: Adaptive Intelligence Applications (AIA) Pre-built, packaged ML and data-driven SaaS applications • Purpose-built, ready-to-use • Embedded ML • Enrich with 3rd-party data • Connected intelligent outcomes across separate functions Optimized supplier terms Demand sensing Next best offers and actions Best-fit candidates Your Data 3rd Party Data Connected Outcomes
- 14. Oracle Sales Cloud Next Best Action Win Probability
- 15. 15 AI Apps for CX B2B Marketer, Sales, and Service Professional Available Today Win Probability • AI and data-driven win/loss probability scores that help identify pipeline risk for improved forecast accuracy Next Best Sales Action • AI optimized next best actions based on sales stage, win probability and best practices learned from across other opportunities Coming Soon Optimized Marketing Orchestrations (B2B) • Automatically identify the optimal message, channel and timing to optimally convert prospects and drive pipeline velocity Lead Optimization • AI-prioritized leads and next best actions to effectively convert into qualified sales opportunities Automated Answers • Generate best-fit answers to customer queries across channels to reduce support costs
- 16. Copyright © 2019, Oracle and/or its affiliates. All rights reserved Oracle Commerce Cloud Individualized product recommendations Real-time Click stream Past transactions Past product views Data enrichment Benefits • More individualized experience • Increased basket size • More revenue per transaction ML MODEL
- 17. 17 AI Apps for CX B2C Commerce, Marketer, Service Professional Available Today Next Best Offers and Recommendations • AI-driven personalized product recommendations online Coordinated Open-Time Content • AI-driven personalized product recommendations across outbound channels Intuitive Search Experiences • Utilize data to provide personalized search results and blended product recommendations Coming Soon Optimized Marketing Orchestrations (B2C) • Use AI to define the optimal message, channel and timing for your digital communications Connected Audiences • Re-target customers and find new audiences that exhibit similar customer profiles Automated Answers • Provide best-fit answers to queries automatically via AI
- 18. Copyright © 2019, Oracle and/or its affiliates. All rights reserved Oracle HCM Cloud Automatically optimize candidates-to-jobs and candidates-to-candidates comparison Benefits • Improve overall pipeline quality • Reduce time-to-fill openings • Predict highest likelihood to succeed Predictions on best match and likely tenure in role Processed job descriptions Processed resumes Social profiles Historic performance ML MODEL
- 19. AI Apps for HCM Available Now Best-Fit Candidates • Optimize candidates-to-jobs and candidates-to- candidates matches with the best candidates to reduce time-to-fill and improve overall pipeline quality Coming Soon & Future Best Candidate Experience • Provide AI-driven personalized job recommendations to candidates visiting company job site Personalized Onboarding Actions • Deliver AI-driven personalized onboarding actions to employees and managers Team Mix Modeling • Efficiently staff programs and projects with existing, new and on-demand hires to meet organizational goals Time Entry and Absence Approvals • Intelligently automate time entry approvals and provide alerts to anomalies or compliance issues Proactive Advice • Deliver more timely, accurate and compliant compensation offers Recommended Actions (Employees and Managers) • Optimize employee health and growth through guided actions HR Professionals, Employees, Hiring Managers, and Candidates
- 20. Copyright © 2019, Oracle and/or its affiliates. All rights reserved Oracle ERP Cloud Negotiate with suppliers more effectively Benefits • Improve supplier negotiations • Identify and minimize supply chain risk • Support strategic suppliers Recommendations on best negotiation strategy with suppliers Historic performance Industry news Third-party credit rating Interest rates ML MODEL
- 21. Copyright © 2019, Oracle and/or its affiliates. All rights reserved. | Confidential – Oracle Internal/Restricted/Highly Restricted
- 22. Copyright © 2019, Oracle and/or its affiliates. All rights reserved Oracle Manufacturing Cloud Predict quality outcomes, quickly find root causes Human factors Work orders Materials & source data Factory IoT Data streams 101010 Predictions on quality and yield problems with root causes Benefits • Predict quality outcomes • Identify most-likely root causes • Suggested corrective action ML MODEL
- 23. Making ML Easier for Business Analysts Copyright © 2019, Oracle and/or its affiliates. All rights reserved
- 24. Copyright © 2019, Oracle and/or its affiliates. All rights reserved Machine Learning Can Make More Analysts More Productive Time Bound Too Much Coding Difficult Collaboration Challenges MACHINE LEARNING
- 25. Confidential – Oracle Internal/Restricted/Highly Restricted 25
- 26. Confidential – Oracle Internal/Restricted/Highly Restricted 26
- 27. Making ML Easier for Data Scientists Copyright © 2019, Oracle and/or its affiliates. All rights reserved
- 28. Custom Machine Learning Development Challenges Copyright © 2019, Oracle and/or its affiliates. All rights reserved. ingest, transform and govern data from multiple sources drive more collaboration around ML deploy completed models and manage lifecycle free ML developers from IT tasks Data Ingest and Prep Business Understanding Addressing the end-to-end challenges of custom ML development across full lifecycle quickly mine data for useful features Data Exploration Model Creation Model Training and Evaluation Monitor and Manage Lifecycle Deploy
- 29. Ready-to-Build: Oracle’s PaaS ML Platform Components Copyright © 2019, Oracle and/or its affiliates. All rights reserved Model Creation Data Ingest and Prep Business Understanding Data Exploration DeployModel Training and Evaluation analytics / visualization / reporting Oracle Analytics Cloud (OAC) deployment standard GraphPipe (open source project) data ingest Oracle Data Integration Cloud (DIPC) prep & ETL batch & stream Oracle Data Flow Cloud* (serverless Spark) Monitor, Manage and Secure Oracle Cloud IaaS Oracle Management Cloud Oracle Security Clouddata processing and data store Big Data Cloud (Hadoop) object store Oracle Data Catalog Cloud* ADW (Database) Oracle Data Flow Cloud* (serverless) data science development environment Oracle Data Science Cloud* notebooks Git development automation deployment automation lifecycle automation curated suite of open source ML tools delivered as a service* *future ***
- 30. Speeding and Simplifying Data Ingest and Prep Oracle Data Integration Platform Cloud Data Lake Builder – quickly instantiate and copy data into a data lake. Data Preparation – simplified data ingestion focused on an intuitive interface ETL Data Pipelines – run transformations in a data lake or a data warehouse Stream Analytics – event processing on real- time data streams Copyright © 2019, Oracle and/or its affiliates. All rights reserved
- 31. Deploy a model as an API Collaborate in Projects Easily handoff models from data scientist to developer The Oracle Data Science Platform
- 32. Oracle Data Science Cloud Workflow 32 Reproducibility Data Versioning Code Versioning Model Versioning Environment Management Model Deployment Operationalize Models as Scalable APIs Model Management Monitor and Optimize Model Performance Data Exploration Collaborative Data Analysis / Feature Engineering Model Build and Train with Open Source Frameworks Collaborators · Data Scientists · Business Stakeholders · App Developers · IT Admins
- 33. Oracle Data Science Cloud More ML models, better ML models, less effort Curated suite of cloud-native open source tools • Delivered as a managed service, or add your own Drives more collaboration • Notebooks and repositories based on Jupyter and Git Simple, push-down processing • SQL, Hadoop or serverless Spark* Native deployment integration with Oracle Cloud • Containers, Kubernetes, serverless* … Lifecycle Management • Versioning, monitoring, automatic retraining* * future direction Copyright © 2019, Oracle and/or its affiliates. All rights reserved data science development environment Oracle Data Science Cloud notebooks Git development automation deployment automation lifecycle automation curated suite of open source ML tools delivered as a service* * * * *
- 34. Copyright © 2019, Oracle and/or its affiliates. All rights reserved. Choice of Technology Stack </> Programming Languages Java SE, Java EE, Ruby, Python, PHP, Go, JavaScipt,… Databases MySQL, Cassandra, Mongo,… Tools Apollo Angular, Ionic,… Cloud Services using Open Source AppDev Docker, Kubernetes, Node.js Eclipse,... Big Data Hadoop, Spark, Kafka,... Blockchain Hyperledger Fabric AI / Deep Learning TensorFlow, Caffe, DL4J Support for a Broad Set of Technologies 3rd Party Applications Salesforce, Microsoft, SAP, ServiceNow,… Security OAuth 2.0, SAML, SCIM, OpenID Connect Systems Management Cisco, Dell, F5, LAMP,… Channels Amazon Alexa, Facebook Messenger, Slack,… Supporting Open Source NetBeans, Eclipse, Spark, Hadoop, OpenJDK,… Contributions to Open Source Oracle Cloud is OPEN
- 35. Copyright © 2019, Oracle and/or its affiliates. All rights reserved Making ML Easier for the App Developers
- 36. Developers Can Benefit From Machine Learning Delivering Chat Interfaces Integration Scaleability Challenges MACHINE LEARNING
- 37. Oracle Integration Cloud Uses Machine Learning to Make Data Mapping Easier Copyright © 2019, Oracle and/or its affiliates. All rights reserved Learns to recommend data mapping between applications • Auto mapping & translation across apps for faster connectivity • Machine Learning guidance in the Oracle Integration Cloud • Yelp-like “star” ratings feature proven and popular recommendations
- 38. Oracle Intelligent Bots: Pre-Trained ML with integrations Copyright © 2019, Oracle and/or its affiliates. All rights reserved Custom Chat bots for the complexity of enterprise • OOTB ML - Easily leverage pre-built conversational AI and integrations without having to learn data science • Conversational AI platform – easily tunable ML service that turns conversations into actions on backend systems • Enterprise data integration • Bot builder UI HCMERP CX 3rd Party Digital Assistant Individual Skill Bots
- 39. Copyright © 2019, Oracle and/or its affiliates. All rights reserved. Using Machine Learning to Speed Everyday Tasks HCM 3rd Party CX ERP Expenses • Create expense report • Add items to existing report • Get new/aged credit card transactions Sample Interactions Application Systems Oracle Digital Assistant Projects • Review project status • Get overdue task list • Report time worked on project Finance • Run financial consolidation • Run planning and budgeting • Run tax reporting Human Resources • Application status • Fnd jobs for a specific location • Show jobs within range of user location Sales • Get next appointment summary • Get a list of tasks which are due • Get quote for opportunity Understands your context and can act across multiple applications. • text • email • messaging • voice • images
- 40. 40Copyright © 2019, Oracle and/or its affiliates. All rights reserved. University of Adelaide 47% reduction in call volume 96% reduction in call center wait time 60% ”Awesome" rating from students Provides”GradeA"studentexperience withOracleIntelligentBots CUSTOMER PERSPECTIVE Our Adjusted ATAR chatbot meant students didn’t have to wait – in the busiest hour we had approximately one user every five seconds! The response has been fantastic, with more than 60 percent of users rating their interaction as ‘Awesome’. Catherine Cherry, Associate Director, Prospect Management, University of Adelaide
- 41. Copyright © 2019, Oracle and/or its affiliates. All rights reserved Enterprise Class IaaS for Modern ML • The fastest x86 compute – Unmatched bare metal performance • GPU as a service – Use our tools, or yours • High bandwidth network – Connect compute and storage • High performance flash storage – Primary access, fast load from archive • Low-cost archival storage – For data not in immediate use Oracle Cloud Infrastructure The highest performance at the lowest cost
- 42. Making ML Easier for DBA’s Copyright © 2019, Oracle and/or its affiliates. All rights reserved
- 43. Database administrators spend a significant amount of time on repetitive, tedious and error-prone tasks Challenges Tuning Securing Monitoring and Trouble Shooting Copyright © 2019, Oracle and/or its affiliates. All rights reserved MACHINE LEARNING
- 44. Copyright © 2019, Oracle and/or its affiliates. All rights reserved. Ready-to-Work: Oracle Autonomous Database Embedded ML in Databases to enable predictive enterprise applications Self-Driving Automates database and infrastructure management, monitoring, tuning Self-Securing Protects from both external attacks and malicious internal users Self-Repairing Protects from all downtime including planned maintenance All Analytic Workloads • Data Warehouse, Data Mart • Data Lake, Machine Learning Autonomous Data Warehouse (ADW) Online TP & Mixed Workloads • Transactions, Batch, Reporting, IoT • Application Dev, Machine Learning Autonomous Transaction Processing (ATP) ORACLE AUTONOMOUS DATABASE
- 46. Copyright © 2019, Oracle and/or its affiliates. All rights reserved. Oracle Database: In-Database Machine Learning • Wide range of algorithms and models with Oracle Advanced Analytics option • Flexible model building - SQL, R or Python - Oracle Data Miner - Oracle Auto ML • Machine learning models and algorithms run inside the Oracle Database – not externally • Massively parallel execution • Data stays in-place
- 47. Making ML Easier for IT Managers Copyright © 2019, Oracle and/or its affiliates. All rights reserved
- 48. Deeper insights Maintenance at scale Security at scale Copyright © 2019, Oracle and/or its affiliates. All rights reserved. IT Managers need to deal with increasing complexity at scale Challenges MACHINE LEARNING
- 49. Copyright © 2019, Oracle and/or its affiliates. All rights reserved. Bringing Machine Learning To IT Management and Security Data Sources user activity application middleware/API database hypervisor/container/ operating system infrastructure external threat feeds data lake Build Your Own ML Models* Analytics Cloud Data Science Cloud Data Integration Platform Cloud Pre-Built ML Models (examples) identity access management capacity planning problem remediation problem prediction anomalous behavior Oracle Management Cloud Oracle Security Cloud * future direction Oracle Data Catalog Cloud*
- 50. Copyright © 2019, Oracle and/or its affiliates. All rights reserved. Why Use Oracle to Deploy ML Everywhere? Pre-Built SaaS Smarts. Immediate and Easy Benefits Many Choices in Open Technologies Runs on Enterprise Grade SaaS, PaaS, IaaS Oracle Runs and Maintains Designed by Specialists. Used by others. Lowered Risk