Tell us a bit about yourself before you enrolled in the Visual Computing program at SFU.

I used to work as a data scientist in the San Francisco Bay Area for three years. My background is in applied math and I started teaching myself fundamentals of data science a few months before getting my first job in the Bay Area. Teaching myself new concepts has been an ongoing theme throughout my career and academic life. While I was working I spent a lot of time outside my job to learn about cutting-edge developments in machine learning. In particular, a machine learning conference at Google led by Ian Goodfellow and his research group scratched my curiosity to learn more about deep learning and its applications to computer vision. The conference as well as the readings I did after that were the main factors for me to consider pursuing a graduate degree in visual computing.

What attracted you to the Visual Computing program?

There are a few factors such as SFU’s ranking in visual computing research as well as the fact that the program is unique in Canada. However, for me the most important factor was SFU's focus on preparing students for industry. Here at SFU we have practical labs in which we learn to build 2D indoor maps using a robot and to build a model that detects vehicles in real time. Our courses highly focuses on the practical applications of the concepts we are learning and that prepares students well for industry.

Why did you choose to come to Metro Vancouver?

The reason I decided to come to Burnaby was primarily SFU's expertise in Visual Computing. However, I have lived in Metro Vancouver before and I loved it. It's perfect for people who enjoy nature and outdoor activities. So deep down I think my decision was also a bit influenced by my love for Metro Vancouver.

What is the most important thing you have learned in the program so far?

I have learned about a few types of generative adversarial networks (GAN) in my Machine Learning course that I'm quite excited about. GANs are primarily used for generating real looking images. One of the great applications of GANs which I learned during my machine learning project is image translation. Let's say for self-driving cars the car needs to make decisions at night even when there is little light. I learned about models that translate night time images to day time images with stunning accuracy.

Why do you think this program is significant?

I think this program is significant if you enjoy learning about cutting edge algorithms and, most importantly, you enjoy developing and applying them.

What types of jobs are you looking for? How does this program fit with your career goals?

I'm looking for jobs with a big focus on applications of deep learning to computer vision. We have learned a lot about this topic in our first lab course and we’ve built two projects related to this topic that I'm proud of.

First, we built a model to automatically find facial landmarks given images of human faces. In the second project we built a model which finds vehicles in an image and draws bounding boxes around them in real time. The second project is my personal favourite because the algorithm is quite powerful and has a wider application than just Autonomous Driving. For instance, the algorithm is also used to analyze Medical Images for detecting tumors.

What would you tell potential students looking at SFU's visual computing program?

I think one important thing to have in mind is that the lab projects that we work on might look very difficult at the beginning. However, you will be amazed how after a week you make so much progress in building an amazing application such as the vehicle detection model I mentioned earlier. Persistence to learn and working with other classmates is key.

Also it's absolutely critical to make sure you are comfortable with linear algebra and vector calculus. These are the building blocks of all the great material we learn during the program and having a good knowledge of those topics would help you get a deeper understanding.