To code or not to code: the way forward for machine learning

In a rapidly changing word of technology and engineering advancements, we’re reminded of Charles Darwin’s words it’s not the strongest that survive, but the most adaptable. For humans this means learning from our errors, one painful mistake at a time, and fixing our approach so we do not stumble again. We’re limited by our personal experiences so we can only adapt once we approach a problem; but by then it may be too late. Imagine having the collective wisdom and understanding of everyone’s experiences so that you know how to solve problems you’ve never seen before. This is the beauty of machine learning.

 

Behrooz hanging out in front of the Magnolia’s in the MU

If you haven’t heard of machine learning, then it’s just a matter of time. These techniques are already involved in highly complex board games, advertising optimization, and especially self-driving cars. It’s difficult to say how impactful machine learning will be to our everyday lives because the applications of this field are still being discovered. One of the primary foundations of machine learning is researching how computers interpret visual information so computers can make on-the-fly adjustments to stop for a pedestrian or speed up to merge on the freeway.

Behrooz Mahasseni recently finished his Ph.D. in Electrical Engineering and Computer Science where his research focused on how computers interpret video recordings. As part of his research, he worked on a project to analyze football videos to identify specific patterns like huddles, punts, and special teams plays. This is specifically useful for football recruiters who don’t have time to watch 3.5-hour football games when they’re looking for a good wide-receiver for their team. Behrooz’s work helps the computer understand when passing plays occurred so the football recruiter can watch the ‘highlights’ reel for five minutes and get all the information they need to make a hiring decision. This seems rather easy, but Behrooz worked on this for high school football games where the video is not in high definition, from an oblique angle instead of a birds-eye-view, and probably has a very excited parent-videographers jumping up and down for major plays. Obviously teaching a computer to understand videos is easier said than
done, but Behrooz was able to get all this accomplished with a high degree of accuracy that helped him land a job with Apple. He’s described this job as research and development using the skills he learned in graduate school (that’s about all he can say) but it took him many years of school to finally realize he had the skills to act
as the spearhead of technological innovation.

Behrooz’s family including his wife Mitra and Behrad celebrating the Persian New Year March 2016

There is so much more to discuss with Behrooz, especially about where the field of machine learning and artificial intelligence is moving. We will also discuss his first experience with a robotic competition in Tehran, his decision to move to the United States, and his never-ending drive for finding and solving new problems. Be sure to listen in Sunday September 3rd at 7PM on 88.7 KBVR Corvallis!
Print Friendly, PDF & Email