Consumer products built with what we loosely term artificial intelligence are growing exponentially. As computing and processing power increases, it’s now possible to build technologies based on the idea of neural networks centred around learning and learning algorithms. This technology did exist a few decades ago, but were too costly and time-consuming to run.
“AI — and specifically deep learning in our case — is important because it provides a much faster and objective approach to solving challenging imaging problems.” – Algolux CEO Allan Benchetrit
Montreal, in particular, is seeing a massive surge of these types of technologies thanks to institutions supporting disruptive companies. TandemLaunch is a startup accelerator set up more like a startup studio that builds companies internally. Scouting for university technologies that could have a high impact when commercialized, they perform a matchmaking process with the inventors of these technologies and other recent graduates and PhDs, who then come together as the founding team of a tech startup. TandemLaunch mentors the team, provides them with seed investment, and guides them in the process of building their company over a 15-18 month incubation period.
While they don’t specifically focus on artificial intelligence, a large number of the companies within the TandemLaunch portfolio have created products using AI or neural networks. Omar Zahr, venture developer at TandemLaunch, notes that this is ultimately a consequence of the direction that you see the industry moving in, as well as academia.
“What’s happened recently is the performance of our computers; the processing has improved,” said Zahr. “Especially with the advent of the GPU, the graphics processing unit, which is very good at running calculations in parallel — something that these neural networks need. This essentially enables our ability to utilize these networks in an efficient way.”
Among the startups that have moved through the TandemLaunch program are Sportlogiq, a player tracking and activity recognition system for hockey; Algolux, a machine learning software platform that optimizes camera/vision systems; and Landr, a music-mastering platform based on machine learning.
When asked about the trend toward artificial intelligence, Algolux CEO Allan Benchetrit agreed with Zahr that the key is the speed of solving problems.
“AI — and specifically deep learning in our case — is important because it provides a much faster and objective approach to solving challenging imaging problems. Computer vision applications count on the camera being able to see and understand in a way similar to the way humans do,” Benchetrit said. “When we think about applications like facial detection for security or traffic sign detection for self-driving cars, the key for the underlying system is not just to detect, but rather to make interpretations and decisions based on that detection. Using neural networks, deep learning allows systems to quickly analyze millions of images (e.g. stop sign) in order to detect regardless of weather condition or obstruction, and then make instantaneous critical decisions.”
Sportlogiq also takes advantage of the speed of analysis in order to give teams, broadcasters, fans and analysts a new way to understand hockey. Since their tenure in the TandemLaunch program in 2014, this team of fifteen computer vision and machine learning PhDs, MScs, and software engineers has raised $1.7 million in seed funding and seen a lot of interest in their technology.
By leveraging computer vision, which is a form of artificial intelligence, they are able to classify patterns on images in order to identify people and objects. This is seen as incredibly valuable in the professional sports world, and they have garnered interest from teams and broadcasters alike.
“In our first year, we have brought on seventeen NHL teams and four major broadcast networks as clients, in addition to filing five patents around the technology,” said CEO and co-founder Craig Buntin.
They have also brought on a number of high-profile investors, including Mark Cuban, one of the main “shark” investors on the ABC reality television series, Shark Tank. Cuban takes a particular interest in sports-related startups and led Sportlogiq’s seed round.
TandemLaunch expects to see a large number of AI-type startups in their program in coming years. They are always recruiting for new inventors, graduate students, and entrepreneurs and recently held a project showcase to attract new students.
“The process of selecting projects is really the process of locating technologies that represent great opportunities,” said Zahr.
With the continued improvements in processing power, it’s likely that we will see many more of these types of technologies emerge in the near future. Whether it’s the mastering of a track, the optimization of vision systems, or the analysis of the movement of a player on the ice, the applications of artificial intelligence are seemingly limitless.