Layer 6, a Toronto-based AI firm owned by TD Bank Group, has won the international Association for Computing Machinery (ACM) RecSys Challenge for the second year running.
“TD has great support for these types of competitions and our motivation here is to continue to challenge ourselves to be able to think differently, with new data sets and of course to be able to track world-class talent,” Tomi Poutanen, TD’s chief AI officer and Layer 6 co-founder told ITBusiness.ca
According to a release from TD, the competitors had to create personalized music playlists based on Spotify’s newly released Million Playlist Dataset, which includes 1 million playlists created by users.
The AI systems were given limited information about the playlists, in some cases 10 songs, and in others only one song, or simply a title.
By winning the competition, Layer 6 was in good company with previous winners including Chinese ecommerce giant Alibaba and major Russian search engine Yandex.
Each year, ACM RecSys holds a different challenge at their annual conference, with the challenge being run by different companies offering unique datasets.
This year’s challenge hosted by Spotify was actually a perfect fit for Poutanen and the Layer 6 team.
The company may have only been created in 2016, but its team has been working on personalization systems for over seven years, through its company Milq Inc., which is a website that curates personalized playlists for everything including music, sports, and literature.
Poutanen said during the challenge they used AI, in collaboration with the University of Toronto professor Scott Sanner and his research student, to understand the sequential nature of music playlists and create a natural flow, by using a strong deep learning component.
Banking and AI
As ITBusiness.ca reported, TD collaborates with Layer 6 to improve customer experience.
“Banking is all about putting the right offer and conversation in front of the customer at the right time and via the right channel. That means being able to anticipate customer needs and future actions the customers may take to direct the conversation and experience in that direction,” said Poutenan.
He stated AI, specifically deep and machine learning, allows companies like TD to take all the information they have on a customer and put it into a framework that analyzes the data without any the guesswork associated with the traditional analytic approaches.
“Consumer expectations now are largely driven by their experience with big tech companies like Netflix, iTunes or Amazon, all these services are highly personalized, so the expectation of the customer of what a digital experience should be continues to go up. It’s important for banks to keep up with consumer expectations,” said Poutanen.
Poutanen and fellow Layer 6 co-founder Jordan Jacobs are also co-founders of the Vector Institute for Artificial Intelligence.
He said they “had a vision for creating an institute that was larger than the University of Toronto, that could more closely collaborate with the industry and really help lift the whole technology sector and economic sector with the application of AI.”
The pair have been major players in the Canadian AI industry every since and it was winning last year’s RecSys 2017 challenge, that got Layer 6’s AI system some notoriety.
“Last year we won by a landslide,” said Poutanen, “showing that our framework is quite unique and that attracted a lot of attention to us. We were really the first company that publicly went out and spoke about deep learning used in recommendations setting.