Design Thinking for Machine Learning
User-centered & data-driven. Leverage machine learning in smart product design.
Grainger Faculty Instructor: Ranjitha Kumar, Assistant Professor of Computer Science
Course Length: 3 days (approximately 18 contact hours)
Dates: By Request
CEUs: 1.8 (estimated)
Register Now">Register Now
Who Should Attend
This course was created for professionals developing smart products powered by machine learning, including entrepreneurs, user experience designers, product managers, and (full-stack) developers. Participants should have a basic understanding of programming and rudimentary mathematical maturity.
This course introduces techniques for designing effective user experiences powered by machine learning models. This course offers strategies for bootstrapping the design process in these situations, mapping user-centered design requirements to low-level modeling decisions, and explaining model behavior to users to engender trust.
Lectures present recent results from human-computer interaction (HCI) and artificial intelligence (AI) research, and case studies of real-world products. Instructor-led coding demonstrations and design exercises merge this theory with practice, reifying how user-centered design can be injected into a data-driven product’s development process.
This course teaches strategies for operationalizing user experience guidelines in the context of data-driven products, including how to
- identify user needs and automation/augmentation opportunities with Wizard-of-Oz prototypes
- design machine learning models that support explainability, personalization, and fairness
- iteratively improve user experience with human-machine co-learning systems
Computer and/or Software Requirements
Laptop with requisite open-source software installed (instructions will be provided a few weeks before the beginning of the course).
Resources & References
- Engelbart, D. “Augmenting Human Intellect: A Conceptual Framework”: http://dougengelbart.org/content/view/138
- “People + AI Guidebook.” https://pair.withgoogle.com/
- Amershi, S. et al. “Guidelines for Human-AI Interaction.” https://www.microsoft.com/en-us/research/publication/guidelines-for-human-ai-interaction/
- Kumar, R. and Vaccaro, K. “An Experimentation Engine for Data-Driven Fashion Systems.” http://ranjithakumar.net/resources/kumar-aaai17-fashion.pdf
Sample Course Outline
- Introduction to Design Thinking for Machine Learning
- Identifying User Needs and Automation/Augmentation Opportunities
- Design: Wizard-of-Oz Needfinding Systems
- Acquiring and Wrangling Data
- Code: Web Crawler for Mining Large Datasets
- Aligning Mental and Machine Learning Models
- Design: Models for Explainability, Personalization, and Fairness
- Training Machine Learning Models
- Code: Neural Network trained on Image Dataset
- Evaluating Data-Driven Products with User-Centered Metrics
- Design: Human-Machine Co-learning Systems
- Deploying Machine Learning Models and Staging Automation
- Code: Search Engine powered by Trained Model
- Wrap-Up Discussion
Register Now">Register Now
About the Instructor
Ranjitha Kumar is an Assistant Professor in the Department of Computer Science at the University of Illinois at Urbana-Champaign (UIUC), where she leads the Data-Driven Design group. She is a 2019 recipient of the Deans Award for Excellence in research from The Grainger College of Engineering, and the 2018 recipient of the C.W. Gear Outstanding Junior Faculty Award from the Department of Computer Science at Illinois. Her research has won best paper awards/nominations at premier conferences in HCI, and is supported by grants from the National Science Foundation, Google, Amazon, and Adobe. She received her PhD from the Computer Science Department at Stanford University in 2014, and was formerly the Chief Scientist at Apropose, Inc., a data-driven design company she founded that was backed by Andreessen Horowitz and New Enterprise Associates.