How Machine Learning Works for Business
1 like847 views
The document discusses how machine learning can serve as a tool and opportunity for businesses, emphasizing its ability to automate and enhance software through trained logic derived from historical data. It covers the complexities of choosing the right machine learning architecture and highlights the importance of data management and asking quality questions. Lastly, it underscores the potential of 'expertise on demand' to facilitate continual improvements and personalized experiences in various business processes.
How Machine Learning Works for Business
- 1. Building expertise on demand. HOW MACHINE LEARNING WORKS FOR BUSINESS
- 2. TODAY’S GOAL Leave with a firm grasp on how machine learning can be a tool and an opportunity for your business.
- 3. OUR JOURNEY ● Context ● Choosing the right tool ● Putting it to work ● The hurdles
- 4. WHAT IS MACHINE LEARNING? A brief introduction.
- 5. SOFTWARE IS EATING THE WORLD Everything is becoming code. Software automates, simplifies, and accelerates business.
- 6. BUT... Someone needs to write the code/logic.
- 7. THE TRADITIONAL WAY Handwritten logic. If / Then / Else
- 8. THE PROBLEM Complex situations require complex2 software logic.
- 9. THE SOLUTION Trained logic using historical data. Model tripID hasEggs eggsBought milkBought 1 1 6 1 2 0 0 2 3 1 6 1 4 1 8 1 5 0 0 1 6 1 6 1 7 0 0 3 0011001101011101 0101100111010 11001001101 Stored as a mathematical model. Finds patterns in the data.
- 10. WHAT IT LOOKS LIKE console.aws.amazon.com/machinelearning/home?region=us-east-1#/datasources
- 11. ML IS EATING THE SOFTWARE All applications are becoming “smart” — with unprecedented complexity in logic. Machine learning automates, simplifies, and accelerates software.
- 12. MURKY LANDSCAPE Artificial Intelligence Machine Understanding (?) Pattern recognition Classification Prediction Can only do one thing Brute-force approach Autonomous decisions Universally applicable Intuition approach Google DeepMind Amazon Machine Learning Natural language processing Computer vision Optimization IBM Watson Classic learning Multi-tiered deep learning neural networks Deep learning neural network Explicit ProgrammingHandwritten Machine Learning logiccomplexity
- 13. CHOOSING THE RIGHT TOOL Exploring the major machine learning architectures.
- 15. Fast, simple predictions. Product recommender Price prediction Behavior prediction Typically, features are predefined. “CLASSIC” LEARNING
- 17. Deep learning has eclipsed. “SHALLOW” LEARNING
- 19. Complex, contextual understanding. Computer vision Natural language processing Uncovering hidden insights Often, features are learned directly from the data. “DEEP” LEARNING Many Different Types ● Deep Belief (DBN) ● Convolutional (CNN) ● Recurrent (RNN)
- 22. FINDING A CONTEXT FOR MACHINE LEARNING How to wrap your head around this massive subject.
- 23. Expertise on demand. Train your ideal decision-making processes and tasks, then execute them anytime, anywhere, at any scale. PUT IT TO WORK
- 24. TOOL OR OPPORTUNITY? How can you create expertise on demand for you or your customers?
- 26. UNIVERSAL EXCELLENCE A proven path to business success is staffing “the best.” And, imagine your “best” continually getting better every day. ● Hiring expert ● SEO expert ● PR expert ● Social media expert ● Researching expert And more...
- 27. EXPERT MARKETER Consistently presents the right message, to the right person, at the right time ● Understands your customers better than any high- priced agency ever could ● Dynamically personalizes each individual experience ● Accurately predicts audience behavior
- 28. EXPERT FORECASTER ● Revenue ● Inventory ● Taxes ● Expenses ● Staffing
- 29. EXPERT WRITER From SEO meta tags to novels. ● Can be trained to any “voice” ● Deep topic research and digestion ● Instantly learn from world experts ● Accurate fact checks ● No more hiring writers ● No more typos
- 30. EXPERT ADVISOR Imagine what you could accomplish with your own personal JARVIS. He’s closer than you think.
- 32. WHY WE LOVE COMPUTERS Any time, any where, any scale. No vacations, no sleep, no sick days.
- 33. THE HURDLES Exploring the common sticking points.
- 34. IT’S GETTING EASIER EVERY DAY ● Purpose-built APIs ● Pre-trained models ● Open source projects But...
- 35. We’re headed into a world, where we can know anything, anytime. So the quality of the questions you ask determines your success. “Wrong questions get wrong answers.” ~Gregory The Seventh Son (2014) QUESTIONS MATTER
- 36. FEED THE BEAST Build your data stockpile. Historically we don’t save data, because we didn’t look at it. Now we can learn even more from it — without looking at it. So hoard as much as you can.
- 37. RECAP ● Train complex software logic ● Deploy expertise on-demand ● Ask empowering questions ● Hoard your data
- 38. GOAL CHECK Feel like you have a firm grasp on how machine learning can be a tool and an opportunity for your business?
- 39. THOUGHTS? Upsides & downsides to “expertise on demand?”
- 40. UNLEASH YOUR BUSINESS EMBRACE EXPONENTIAL 10xnation.com