Keep it simple series machine learning
0 likes110 views
This document is a 10-part series from Predictive Layer on machine learning. It defines machine learning, compares traditional programming to machine learning, and provides examples of business uses cases for machine learning like predictive maintenance, demand forecasting, and dynamic pricing. It emphasizes that machine learning can generate adaptive applications and models in real-time from data to improve business performance without extensive programming. The series advocates that combining people and machine learning in teams can produce creative, productive, and efficient decision making.
Keep it simple series machine learning
- 1. Predictive Layer “Keep It Simple Series” Kiss 2018 “Machine Learning” A series of 10 episodes Credits /need to thank : Andew Ng – Standford class of Machine Learning Eric Grimson - MIT Class Machine Learning PredictiveLayers Gurus © Predictive Layer 2018
- 2. Arthur Samuel's definition from 1959: Machine Learning: Field of study that gives computers the ability to learn without being explicitly programmed Cognitive Computing Keep It Simple Smart - Machine Learning #1/10 © Predictive Layer 2018
- 3. Tom Mitchell’s definition in 1998: Well posed Learning Problem: A computer program is said to learn from experience E with respect to some task T and some performance measure P, if its performance on T, as measured by P, improves with experience E. Cognitive Computing Keep It Simple Smart - Machine Learning #2/10 © Predictive Layer 2018
- 4. Keep It Simple Smart - Machine Learning #3/10 The magic of Machine Learning : generate programs based on data that enable a totally adaptive set of business applications that can serve a truly agile business strategy. In a way, Machine Learning generates on a real time basis a new component of your IS based on internal and external data without depending on massive complex expensive software development or configuration. It is a major game changer for business Credits /need to thank: Eric Grimson - MIT Class Machine Learning Traditional Programming Machine Learning Program Data OutPut Big Data Program Computing Platform OutPut Computing Platform © Predictive Layer 2018
- 5. Keep It Simple Smart - Machine Learning #4/10 © Predictive Layer 2018
- 6. Keep It Simple Smart - Machine Learning #4 /10 – Add on Comparison of the expected revenue from advanced machine learning (reinforced with with eligibility trace) as a function of the number of episodes (ie weeks for transportation) Advanced Machine Learning is a powerful game changer for yield and revenue management and provide impressive financial returns Correlation of historical big data (tickets, price, timestamp, route and timetable, client profiles, airplane, options & extras) and external influence data (weather, vacation, economics, strikes, events, calendar & seasons, location, transport price index etc.) to generate automatically predictive and prescriptive models for volume, value and pricing on a real time basis that outperform all other approaches Predictive Layer Software Solution for Yield Management © Predictive Layer 2018 From “ Real-Time Dynamic Pricing in a Non-Stationary Environment using Model-Free Reinforcement Learning”
- 7. Mega Trend For The Next 10 Years: COGNIFICATION OF BUSINESS NEXT(X) = X + AI* (*) AI: Artificial Intelligence , Machine Learning, Deep Learning (*) X: Anything related to business performance, issue, challenge or process Keep It Simple Smart - Machine Learning #5/10 © Predictive Layer 2018
- 8. FUTURE OF BUSINESS = CURRENT + USE CASE (ARTIFICIAL INTELLIGENCE*) (*) AI: Artificial Intelligence , Machine Learning, Deep Learning (mostly for image and language) Keep It Simple Smart - Machine Learning #6/10 SAMPLE BUSINESS USE CASES (MACHINE LEARNING) Dynamic Demand Planning Dynamic Forecasting / Forecast to plan Dynamic Pricing / Yield Management (Prescriptive) Dynamic Customer Segmentation Predictive Maintenance (Prescriptive) Predictive Usage of Resources & Assets Predictive Trafic for people, passengers or vehicles Prescriptive Workforce planning Risk Management Credit Scoring © Predictive Layer 2018
- 9. Keep It Simple Smart - Machine Learning #7/10 Select the right use case that creates value! Align use case with Company Strategy and Purpose… Value created should then become « obvious » « Best Price For our Us and our Client » Yields Mngt / Dynamic Pricing (B2B and B2C) « No queing for eating » Demand Planning & Forecasting « The right product at the right place » « No down time » Predictive Maintenance « Less Trade & Promotion » Dynamic Forecasting and prescriptive A&P © Predictive Layer 2018
- 10. TEAM of ARTIFICIAL INTELLIGENCE PEOPLE & MACHINE LEARNING CREATIVE Thinking PRODUCTIVE DECISION Making EFFICIENT Quality Time Time pressure BEST TEAM Keep It Simple Smart - Machine Learning #8/10 © Predictive Layer 2018
- 11. AI Positive Loop AI/ML Product Users Initial Data Set Defendable Business Position With unique data assets initially build with MVP. Then expand with users (Data Accumulation as a core Capability), and use cases. Big data & Open Data with AI/ML become a critical differentiator for the enterprise More data Keep It Simple Smart - Machine Learning #9/10 Using AI for sustainable business differentiation Example: Yield Management / Dynamic Pricing Value MVP (Minimum Viable Product)AI/ML: Artifcial Intelligence & Machine Learning © Predictive Layer 2018
- 12. After the industrial revolution of artificial physical power that is evolving dramatically with the shift from fossile to renewable energy Time has come to expand with the revolution of artificial intelligence and brain power that will disrupt business and society THE TIME IS NOW ! Keep It Simple Smart - Machine Learning #10/10 © Predictive Layer 2018
- 13. Credits /need to thank : Andew Ng – Standford class of Machine Learning Eric Grimson - MIT Class Machine Learning PredictiveLayers Gurus Predictive Layer “Keep It Simple Series” Kiss 2018 “Machine Learning” A series of 10 episodes Thank You © Predictive Layer 2018