A three-way collaboration project dubbed “machine teaching” works to find alternative teaching styles to help students learn effectively.
But don’t let the name “machine teaching” fool you. This project is an interactive learning project for humans by humans.
Engineering Professor Robert Nowak has been working with Jerry Zhu, a professor of computer science and leader of the “machine teaching” project, both have been working on machine learning. Zhu and Nowak try to train machines to recognize images or interpret forms of social media.
Nowak and Zhu then use this knowledge to work with cognitive and educational psychologists at UW to examine how people reason and learn. The machine learning techniques enable researchers to personalize learning systems to effectively educate students, Nowak said.
Tim Rogers, a UW cognitive psychology professor, has also been working on the project. Rogers said machine teaching is a complex series of steps that involves communication between computer science, cognitive psychology and educational psychology professors.
Essentially, the educational psychologists will look at problems educators and students face within the classroom. These problems in turn are given to the cognitive psychologist who examines the best model of a learner in a given subject like mathematics, Rogers said.
“There’s nice interplay between the mathematical computer science world of machine learning and the psychology-cognitive side of the human learning,” Nowak said.
If a machine is taught these concepts, this information can be used to determine the concept or lesson in which students should learn, Nowak said.
Then Zhu processes and configures the information in order to determine the best learning experience that will allow students to learn information as quickly and efficiently as possible, Rogers said.
These results are then handed back to the cognitive psychologists and the educational psychologists to apply in their work, he said.
Rogers said the company is not trying to replace teachers, but rather help them.
“We are trying to use what we know about human learners to work out alternative ways of teaching that may be more effective that would be implemented by teachers,” Rogers said.
While the group of researchers are trying to teach and understand the best possible learning methods for students, they are also learning themselves.
Between the three departments’ communication with one another, there can be difficulties working together because each department has their own approach, jargon and ways of understanding and interpreting certain aspects of the research, Rogers said.
Despite this difficulty, Rogers said the three groups meet frequently to discuss their different approaches and thoughts about the project in order to best understand how to bear on subjects of importance like mathematical learning.
What is important for moving this project forward is to ensure the science is communicated to people so they realize this project is about helping teachers understand ways to teach kids that may be more effective, Rogers said.
“The idea of this teaching is to design these optimal sets of examples that are tuned to an individual’s learning style in order to help them master new concepts as quick as possible,” Nowak said.
