Deep Learning to assist in Human Learning

When we observe thousands and sometimes millions of students learning simultaneously what patterns unfold? 

Being able to autonomously understand students as they learn, both in terms of assessing knowledge and being able to provide feedback, is a grand challenge in education and one that has been studied for hundreds of years. The simultaneous advent of massive education datasets, and machine learning methods which are able to understand complex data provides a unique opportunity to readress these hard problems in education.

When we apply artificial neural network methods to data from thousands and sometimes millions of students learning, clear and useful patterns unfold. (1) Using recurrent neural networks we are able to predict how students will respond to future questions, and (2) through the use of tree neural networks we can understand commonalities in student code as they learn to program.

While deep learning has served as a breakthrough for massive online education, most learning occurs in small contexts. The final part of this talk will speak to the open machine learning problem of how we can provide artificial intelligence for educational settings with few or even no historical data.

Dr. Chris Piech is a Lecturer of Computer Science at Stanford University where he teaches both introduction to programming and introduction to artificial intelligence and runs a research lab. He is a recipient of the George E. Forsyth Memorial Teaching Award.
 
In his research Chris has developed algorithms, especially deep learning architectures, to better understand how humans learn both generally and in the context of programming education. He is especially known for pioneering Deep Knowledge Tracing and Tuned Peer Grading. The algorithms that he has developed are run at Stanford, on Code.org, on Khan Academy and on Coursera in order to teach and provide feedback to millions of students.  Chris is currently working on methods for "zero shot artificial intelligence tutors," with the goal of developing autonomous tutors that can effectively understand and help students who are working through assignments when there is no prior student data. 
 
Chris was born in Nairobi, Kenya and spent his childhood between Nairobi and Kuala Lumpur, Malaysia.