870 likes | 878 Views
This comprehensive guide covers the practical applications of machine learning in industrial automation, including predictive maintenance, quality control, demand forecasting, and training industrial robots. Learn about the implementation options, main branches of machine learning, algorithms, and steps to start a machine learning project.
E N D
Practical Uses of Machine Learning and Ignition Kathy Applebaum Co-Director of Sales Engineering / Inductive Automation Kevin McClusky Senior Software Engineer / Inductive Automation
Machine Learning Practical uses of machine learning What is machine learning Before you start Steps to a machine learning project Conclusion Resources to learn more Agenda
Practical Uses Predictive maintenance Quality control Demand forecasting Training industrial robots Self-driving vehicles Machine Learning in Industrial Automation
Implementation Options More Automated More Control
Implementation Options More Automated More Control
What is Machine Learning? Analytics Machine Learning Artificial Intelligence Three main branches
What is Machine Learning? Analytics • Knowledge discovery Three main branches
What is Machine Learning? Analytics • Knowledge discovery • Descriptive Three main branches
What is Machine Learning? Analytics • Knowledge discovery • Descriptive • Diagnostic Three main branches
What is Machine Learning? Analytics • Knowledge discovery • Descriptive • Diagnostic • Predictive Three main branches
What is Machine Learning? Analytics • Knowledge discovery • Descriptive • Diagnostic • Predictive • Prescriptive Three main branches
What is Machine Learning? Analytics • Knowledge discovery Machine Learning • Learn and improve from experience Three main branches
What is Machine Learning? Analytics • Knowledge discovery Machine Learning • Learn and improve from experience Artificial Intelligence • Tasks that simulate human intelligence Three main branches
What is Machine Learning? Classifiers – predict a category Regression – predict a value Main types of machine learning
What is Machine Learning? Algorithms
What is Machine Learning? K-means - Clustering
What is Machine Learning? K-means - Clustering Categorization for Defect Analysis
What is Machine Learning? Decision trees Should I bring an umbrella? Cloudy? Yes Rain in forecast? Yes No
What is Machine Learning? Decision trees Predictive Maintenance Should I bring an umbrella? Cloudy? Yes Rain in forecast? Yes No
What is Machine Learning? Regression analysis
What is Machine Learning? Regression analysis Process Tuning Production Forecasting
What is Machine Learning? Neural Networks
What is Machine Learning? Neural Networks Process Simplification Vision Systems
Getting Started What things should we have before starting a machine learning project? Prerequisites
Getting Started Data Prerequisites
Getting Started Data • Sources of data Prerequisites
Getting Started Data • Sources of data • Quality data Prerequisites
Getting Started Data • Sources of data • Quality data • Labeled vs. unlabeled data Prerequisites
Getting Started Data Statistics knowledge Prerequisites
Getting Started Data Statistics knowledge • Sampling techniques Prerequisites
Getting Started Data Statistics knowledge • Sampling techniques • Correlation vs. causation Prerequisites
Getting Started Data Statistics knowledge • Sampling techniques • Correlation vs. causation • How good are your results? Prerequisites
Getting Started Data Statistics knowledge Domain knowledge Prerequisites
Getting Started Data Statistics knowledge Domain knowledge • In-depth knowledge about your process Prerequisites
Getting Started Data Statistics knowledge Domain knowledge • In-depth knowledge about your process • Know what types of data are promising Prerequisites
Getting Started Data Statistics knowledge Domain knowledge • In-depth knowledge about your process • Know what types of data are promising • Know when results don’t make sense Prerequisites
Machine Learning Steps Pick a question to answer
Machine Learning Steps Pick a question to answer • High value vs. easy
Machine Learning Steps Pick a question to answer • High value vs. easy • Cost function
Machine Learning Steps Pick a question to answer Use domain knowledge
Machine Learning Steps Pick a question to answer Use domain knowledge • Pick useful data
Machine Learning Steps Pick a question to answer Use domain knowledge • Pick useful data • Acquire missing data
Machine Learning Steps Pick a question to answer Use domain knowledge • Pick useful data • Acquire missing data • Quality data
Machine Learning Steps Pick a question to answer Use domain knowledge • Pick useful data • Acquire missing data • Quality data • Dependent variables
Machine Learning Steps Pick a question to answer Use domain knowledge ETL
Machine Learning Steps Pick a question to answer Use domain knowledge ETL • Automate steps
Machine Learning Steps Pick a question to answer Use domain knowledge ETL • Automate steps • Acquire new data automatically
Machine Learning Steps Pick a question to answer Use domain knowledge ETL • Automate steps • Acquire new data automatically • Clean up data