Seneca research is using automation to help Ontario tech company scale up

Vubble is collaborating with Seneca College to advance video categorization through machine learning and automation.

 

Collaborators: Vubble, Seneca College

In the age of misinformation and fake news, demand is growing for credible video content. Enter Vubble, a next-generation digital media company that provides customized and curated video content from reputable sources, all provided by a combination of algorithms and human editors.

Vubble has teamed up with Seneca’s School of Information and Communications Technology to incorporate artificial intelligence (AI) methods into the video curation process. The project brings together Vubble’s co-founders, former CBC managers and producers Katie MacGuire and Tessa Sproule, and Seneca’s Professor Vida Movahedi, in an industry-academic collaboration. Through accessing SOSCIP’s Cloud and GPU platforms, Professor Movahedi and her student trainees have taken on the challenge of helping to automate Vubble’s video metadata codification process.

Vubble uses next-generation media solutions to help media companies provide online users with personalized content, driven by the user’s interests and a well-rounded information diet, rather than advertising algorithms. Vubble’s founders are firm believers in the importance of broad exposure to quality information; Sproule calls it “foundational to democracy.” Her company has recently built a suite of online tools for a national media client and she is eager to see them put to use by the public.

“When you look at broadcasters and how they get information in front of people, there is an overt reliance on American companies driven by advertising,” Sproule said. “People need the serendipity of exposure to topics they didn’t even know they cared about.”

The Vubble and Seneca partners were initially introduced at an event sponsored by the Ontario Centres of Excellence. SOSCIP fostered the connection and the one-year project got underway in March 2018. The partners met to discuss project goals and subsequently met each quarter to discuss progress.

“There was a natural alignment,” said Sproule. “They believed in us and we needed their expertise.” The student trainees involved with the Vubble project benefit from exposure to both new technology and diversity. As the leader of Seneca’s first project collaboration with SOSCIP, Movahedi had her pick of talented students and selected six to work with her on the first phase of the project.

When you work with others from diverse backgrounds, you learn new things. You discover the strengths that others have and how to incorporate them into a team.

“They had good backgrounds in programming and some had taken a course in computer vision, but none had any experience with machine learning,” Movahedi said. “They learned a lot by working on this platform. I was happy to be able to introduce them to some of the Machine Learning concepts; they were so eager to learn.”

Members of the Seneca team (left to right): Wang (Abby) Pan, Eric Schvatzman, Jiel Selmanovski, Professor Vida Movahedi and Janice Ward (Vubble).

Movahedi says that working on a computer vision and machine learning project makes the students very attractive to employers; two of the six students involved have since gone on to do internships with BlackBerry. A third has been hired by Vubble.

She believes the students benefit from working as part of a diverse group. “We live in Canada and diversity is part of life,” she said. “When you work with others from diverse backgrounds, you learn new things. You discover the strengths that others have and how to incorporate them into a team. Whether we’re talking about people who come from different cultures or some who are introverted and others who are extroverted, diversity is part of the workplace.”

Not only did Movahedi and her students enjoy the project tasks; they enjoyed working with their Vubble partners. “They were friendly and reasonable and understood that with research, you can’t always make guarantees,” she said. “The Vubble team was also eager to understand what we were doing and to learn about the technical aspects of our work.”

For her part, Sproule believes that collaborating with academics makes sense for Vubble. “I love that they’re approaching this purely from a research perspective,” she said. “Their vision isn’t clouded by profits. Working with Seneca helps us build credibility with our clients and end-users, which is very important. We’re trying to make our platform transparent so the user can understand why certain pieces of content come their way.”

“We are big on explainable AI, and that approach aligns with academic research.”

A collaboration between female leaders is, unfortunately, still reasonably novel in technology fields, but Sproule and Movahedi believe that change is coming and is valuable. “It’s good to have women in leadership roles,” Movahedi said. “It’s a form of diversity, and our leadership style can be different sometimes. Even the large tech companies like Facebook and Yahoo are shifting toward including women as leaders, so there is obviously some value there.

The problem Vubble is trying to solve requires deliberate, thoughtful, plodding work. We aren’t moving fast and breaking things. We’re in it for the long run.

Female students often approach her to tell her that it gives them hope when they see her in a prominent position in the largely male field of artificial intelligence. “They say it helps them believe that they can do this too; that they are not in the wrong field,” Movahedi said. “They feel that they can approach me easily and I feel as if I’m setting an example.”

Tessa Sproule, Co-founder and CEO of Vubble

Sproule, who grew up with a mother working in a traditionally male field, agreed that “if you don’t see it modelled, you don’t feel as if it’s something you can do.” She believes that female leaders often bring additional talents to the table. “In my experience, when I had female leaders or mentors, I found them more open to new ideas and willing to let new voices be heard,” Sproule said, pointing to the problems some of the major technology companies have had when there were no women in the leadership mix.

“I hesitate to chalk it up to gender, but I do acknowledge that there is something different in Vubble’s approach to the business and technology of AI — one of our core values is in pushing for human-centred use of technology.”

Research has shown that women are stronger marathon runners than men. I think there’s something in that — the problem Vubble is trying to solve requires deliberate, thoughtful, plodding work. We aren’t moving fast and breaking things. We’re in it for the long run.”

Sproule is confident that opportunities for women in technology are burgeoning. “Vubble has participated in programs operating in the Waterloo-Toronto corridor that are specifically designed to help lift female-led companies,” she said. “Structurally, things are starting to change in terms of the business world not just accepting, but advocating for more women in the boardroom of every size of business.”

“There’s still a long way to go,” says Sproule, “but if I’m invited to participate in a panel discussing AI because I know what I’m talking about and I happen to be a woman, I’ll take that seat.” The initial research contract is coming to an end and Movahedi and her students will be providing their product to Vubble. However, both partners plan to continue the connection, since it should be possible to categorize videos even more accurately by improving the algorithm.

“The Seneca team has just cracked the surface of what’s possible and we want to keep growing the project,” Sproule said. Meanwhile, SOSCIP’s Vubble project has demonstrated the success of an underutilized leadership model: women in technology working together to forge new ideas and create technological breakthroughs.