Machine learning and imaging technology address key barriers in e-learning; Impartus to bring technology to market
Ever try to learn something technical or complex using a video lecture? College students will tell you it’s frustrating. There’s no table of contents to pinpoint where the key information exists in a particular video, and it’s almost impossible to quickly find topics in the massive pile of video material now stored on websites like YouTube. New data analytics technology from Xerox addresses both of these issues and makes it possible to find content that is tailored to a particular student’s learning patterns.
Using computer vision and multimedia analytics techniques the technology removes key barriers in e-learning such as quickly finding and accessing information contained in massive collections of instructional videos and the need to personalize the educational experience. The technology, invented by researchers at Xerox Research Centre India (XRCI) has been licensed to Impartus, a video learning solutions provider which plans to incorporate it into its existing e-learning offerings.
“YouTube alone contains more than 20,000 hours of classroom video lectures, but unlike textbooks, these resources do not come with navigational tools like a table of contents,” said Manish Gupta, Xerox vice president and Xerox Research Centre India director. “By automatically combining written, spoken and supplemental content, these methods turn video into next generation textbooks. We’re also applying machine learning to tailor the learning experience to the student.”
Personalized E-Learning
E-learning makes video coursework from top universities available to anyone anywhere in the world. Often students come from widely varying educational backgrounds. The Xerox technology allows the same content to be tailored to each students’ needs and can even step in as a sort of personal e-tutor, recognizing when a student is struggling to understand a concept and suggesting additional resources.
If a student, after reaching the tenth minute in a video, repeatedly goes back to the fifth minute to view the material again, the solution will infer that the student might need help understanding the concepts presented between the fifth and tenth minute, and offer up suggestions for additional content.
Video Indexing
The technology also uses multimedia imaging algorithms to improve video search navigation by making it really easy for students to get right to the exact spot within a video lesson they want to view. The technology creates an easy-to-use subject index and also a word cloud that hyperlink to the exact spot in the video where that concept is being discussed.
“There are enormous opportunities for educational institutions to improve the learning outcomes for their students via a more effective use of videos,” Gupta said. “This partnership with Impartus reflects our belief in open innovation, and demonstrates how innovation can help expand access to education and personalized learning to anyone with access to the Internet.”
Impartus will license and integrate Xerox technology into its interactive video e-learning solution worldwide allowing Impartus users to benefit from the power of advanced machine learning and multimedia analytics.
“Xerox has assembled a team of the best research talent in India, and they are developing technology that is not currently found in other parts of the world,” said Amit Mahensaria, Impartus Co-founder. “Partnering with Xerox enhances our award-winning video learning platform and furthers our own mission of providing better learning, particularly in developing countries like India, Kenya, and Malaysia.”