How Machine Learning Will Transform Finance
Download as PPTX, PDF5 likes1,041 views
The document discusses the transformative potential of machine learning (ML) in finance, highlighting its ability to enhance productivity, optimize processes, and improve decision-making through advanced analytics. It explores various use cases, common obstacles, and the changing role of data scientists, as well as the necessity for organizations to adapt and integrate these technologies effectively. Key recommendations include digitizing data, forming data ethics committees, and building advanced analytic competencies within finance.
![Image Recognition: More Room to Grow
Muffin or Chihuahua Apple or Owl
[Both are sweet; only one is good to eat]](https://image.slidesharecdn.com/clayton-howmlwilltransformfinance-slideshare-171115215641/85/How-Machine-Learning-Will-Transform-Finance-25-320.jpg)
How Machine Learning Will Transform Finance
- 1. How Machine Learning Will Transform Finance Rich Clayton Vice President Product Strategy Big Data Analytics Oracle Corporation @richardmclayton
- 2. • What is Machine Learning? • What are high yield use cases in finance? • What are the common obstacles? • How has technology advanced? • What you do to get started? In This Session….
- 3. IDC Organizations That Analyze All Relevant Data and Deliver Actionable Information Will Achieve Extra $430 Billion in Productivity Gains Over Their Less Analytically Oriented Peers by 2020
- 4. What is Machine Learning?
- 5. Robotic Bosses Unlikely in Next 10 Years
- 6. Copyright © 2016 Oracle and/or its affiliates. All rights reserved. | 47% US Employment @ Risk of Automation in Next 20 Years 6 Source: Oxford University The Nature of Work Will Most Definitely Change
- 7. Machine Learning is based on algorithms that can learn from data without relying on rules-based programming. McKinsey The New Technology Enabler
- 8. Machine Learning Goes Beyond Predictive Analytics Build Model Predict Outcome Take Action Learn/ Adapt MachineHuman YESTERDAY’S PREDICTIVE ANALYTICS
- 9. Machine Learning Goes Beyond Predictive Analytics Build Model Predict Outcome Take Action Learn/ Adapt MachineHuman YESTERDAY’S PREDICTIVE ANALYTICS TODAY’S MACHINE LEARNING Build Model Predict Outcome Take Action Learn/ Adapt
- 10. WORK LIFE Next Best Offer Recruiting Best Candidates Fraud Detection Supply Chain Automation Predictive Maintenance Optimize Payment Terms Boost Data Center Efficiency Personalized Shopping Personalized Medicine Personalized Entertainment Improved Navigation Self-Driving Cars Virtual Assistants Wealth Management Machine Learning: It’s Everywhere
- 11. Copyright © 2016 Oracle and/or its affiliates. All rights reserved. | From the Frontlines of Machine Learning Innovation Large bank used ML to analyze its collection activities and learned it could eliminate more than 40% of customer calls with better outcome. Retail Telecom Financial Services Global retailer used advanced ML to forecast customer demand cutting forecasting error in half. Telecom company found that their ML methods yielded a 75x reduction in ‘false-alarms’ for churn; focusing resources on those truly at risk of leaving.
- 12. Confluence of Events Happening Simultaneously
- 13. • Predictive models will improve over time • Business Applications will embed self-learning algorithms • From experiment designer to process configurator • Monitor models in production • Scale labs beyond prototypes • Operationalize insights The Role of the Data Scientist Will Change Source: Forrester, Massive Machine Learning
- 14. MACHINE AUTOMATION Simplify execution of repetitive computational, statistical processes PEOPLE JUDGEMENT Model inputs factors and training data, improve data imperfections, model usage (ethics) Adaptive Intelligence – Best of Both Worlds
- 15. Consolidate Many Sources Interactive Visualization Defined Models Adaptive Intelligence Platform Human to Machine Auto Explain Contextual Insights Self-Learning Algorithms Programmatic Integration Recommend Best Action Machine to Human 15
- 16. How can ML be applied at all levels of the finance organization?
- 17. Copyright © 2016 Oracle and/or its affiliates. All rights reserved. | Framing the Opportunity Space Process Transformation Enterprise Protection Finance Productivity ML• Employee safety • Consumer fraud • Employee theft • Physical & Cyber-security • Optimized Payment Terms • Best-Fit Suppliers • Optimized Demand Sensing • Inventory Management • Autonomous Discovery • Narrative Automation • Expense Audit • Tax Analytics • Smart Forecasts • Account Reconciliations
- 18. What are the common obstacles?
- 19. “Bottleneck is management, implementation and imagination.” Andrew McAfee Principal Research Scientist MIT Sloan School of Management Source: HBR, The Business of Artificial Intelligence
- 20. Copyright © 2016 Oracle and/or its affiliates. All rights reserved. | Hurdles to Overcome on the Path to ML Adoption Oracle Confidential – Internal 20 Accessing & managing data Disparate technologies Missing skill sets Bringing together enterprise and third-party data meaningfully Enterprises do not have the technologists and data scientists needed Non-integrated stack within the enterprise presents challenges Limited data, limited skills and complex technology stack make ML adoption difficult
- 21. How is ERP embracing Machine Learning?
- 22. • Purpose-Built and Ready-To-Go • Enriched with 3rd-Party Data • Bundled Decision Science • Built on the Modern Cloud • Connected Intelligent Outcomes Oracle’s Adaptive Applications 1st-Party Data Connected Outcomes Decision Science Machine Learning 3rd-Party Data Optimized Supplier Terms Demand Sensing Next Best Offers and Actions Best-Fit Candidates ERP SCM HCMCX
- 23. How Would ML Work in an ERP Process? Insights Discount Offers Accepted / Rejected, Financial Impact Supervisory Controls Program Definition, Supplier Eligibility, Supplier Opt-in Decision Science Artificial Intelligence, Machine Learning Algorithms Oracle ERP Cloud Feedback Discount Offers Accepted / Rejected Surfaced Outputs Stack Ranked Supplier Eligibility and Discount Offers 1st-Party Data Payables, Suppliers, Treasury Position, 3rd-Party Data Company Profiles, Business News, Regulatory Actions, Channel Delivery Web, Email, Mobile, Digital
- 24. How are analytics platforms embedding machine learning to serve finance?
- 25. Image Recognition: More Room to Grow Muffin or Chihuahua Apple or Owl [Both are sweet; only one is good to eat]
- 26. Copyright © 2017, Oracle and/or its affiliates. All rights reserved. 26 Designers Are Thinking About…. • Why do you receive 65 push notifications per day on your cell phone but 0 from your business? • What your analytics systems are doing while you are sleeping? • Why you spend so much time explaining simple variances when much can be automated? • Why you can ask our mobile for directions to find nearest restaurant but you can’t ask our system how revenues are trending in Italy?
- 27. Copyright © 2017, Oracle and/or its affiliates. All rights reserved. 27 Oracle Analytics Cloud Autonomous Data Discovery • In-line Machine Learning for the business user with no specialty skills required • Uncovers hidden drivers and recommends areas of exploration without user actively analyzing specific factors 27 What is it? Why does it matter? • Guide the user to areas of interest they might not know to look at • Freedom to explore data more fully without specialist intervention • In-depth statistical analysis on contextual data enriches the interactive analytic experience
- 28. Copyright © 2017, Oracle and/or its affiliates. All rights reserved. 28 Oracle Analytics Cloud Self-Learning Contextual Insights Anticipates questions through self-learning • Infuse data-based insights into daily activities • Get customized feeds based on what you are interested in, when and where you are interested in it, and who you collaborate with • Anticipates your needs and delivers appropriate information to help you make better informed decisions throughout the day • Use your device’s voice capabilities to obtain answers 28
- 29. Copyright © 2017, Oracle and/or its affiliates. All rights reserved. 29 Oracle Analytics Cloud Voice Activated Conversational Analytics Powerful, intuitive keyword search • Interprets semantic layer, user private data, expression library and catalog artifacts • Voice-enabled • Fuzzy match, stemming, natural language processing • Generates on-the-fly queries - visualizations are auto-created while user types • Mobile • Browser • Desktop Available on all platforms 29
- 30. Copyright © 2017, Oracle and/or its affiliates. All rights reserved. 30 Oracle Analytics Cloud Natural Language Generation • NLG as standard visualization type • Narrative auto-generated based on visual grammar • Building block for other conversational interface capabilities 30 What is it? Why it matters? • As analytic interfaces become more conversational, you need words in natural language to enhance the meaning of visualizations • Summarizing and highlighting salient points on a chart allows users to focus on what matters and helps filter out the noise
- 31. Oracle Confidential – Internal/Restricted/Highly Restricted 3 How do you get started with ML?
- 32. Copyright © 2016 Oracle and/or its affiliates. All rights reserved. | 3 Your Action Plan • Digitize and save all data! • Create data lab for experimentation • Form data ethics committee • Build advanced analytic competencies and finance • Embrace cloud; it’s faster!
- 33. Copyright © 2016 Oracle and/or its affiliates. All rights reserved. | EXPERIMENT PROCESS DISCOVERPREDICT WRANGLE • Build a system of innovation, not just system of record • Understand data potential • Enrich the data and make it better • Unlock insights and share the value • MIT rule of thumb: 6 x 6 x 6 Form a Data Lab: A Fundamental New Approach
- 34. Copyright © 2016 Oracle and/or its affiliates. All rights reserved. | Deepen Analytic Competencies in Finance Storytelling Scenario Analysis Analytic Process Design Visualization Machine Learning Data Modeling Data Wrangling Text Analysis
- 35. “Computers are useless. They can only give you answers.” – Picasso …so ask smarter questions.
Editor's Notes
- #15: @ Oracle, we believe the future is Adaptive Intelligence, Not artificial. Technology while may be artificial, never replaces human judgement So what is Adaptive Intelligence? Adaptive intelligence is the intersection of people judgement and machine automation. Consider this: machines can memorize more in 1 second than people can in 10 years. Without forgetting and without fatigue. This new collaboration creates many opportunities to improve our forecasting abilities and improve our intuition as uncertainty grows.
- #16: By contrast, Adaptive Intelligence New data is created, in the form of observations learned from data The machine prompts the consumer with relevant insight It’s active - It works by recommending an action Context is all important – who am I, where am I, what am I doing, what do I need to know right now. The models are self-learning; they take the observations on data and optimize the recommended action over time.