The 2018 Toronto – Tsinghua AI Squared Forum attracted researchers, entrepreneurs, innovators, and students to Canada’s largest city to “discuss opportunities for research collaboration and partnerships in areas of artificial intelligence.”
Organized by the North America Federation of Tsinghua Alumni Associations and Tsinghua Alumni Association of Southern Ontario Canada with the assistance of the Association of Chinese Senior IT Professionals and Synced, the event ran May 5 at University of Toronto’s Bahen Center. It was the first such collaboration between the two universities. Speakers also included professors from MIT, Western Ontario, and the University of Waterloo.
Hongqiang Zhao, CFO of Chinese fintech company and event sponsor 100 Credit, addressed AI & Big Data Applications in Financial Services” in his morning keynote speech. Zhao told Synced, “Toronto is a well-known AI research hub. We are looking for technical, talent, and programme exchanges between Toronto and Beijing. As US trading and immigration policies tighten, we want to seek opportunities here. China has a lot of application scenarios for AI technologies and these need to be backed up by Canada’s research capabilities.”
Tsinghua University Professor Minlie Huang opened the morning technical session with an overview of conversational AI and challenges facing chatbots in terms of emotion, personality, and common sense. Elaborating on his AAAI 2018 paper Emotional Chatting Machine: Emotional Conversation Generation with Internal and External Memory, Huang said challenges facing bots today include coming up with many possible responses to a single question, real-world understanding, and awareness of situations and moods.
Building on the topic of conversational AI, University of Toronto Professor Yang Xu briefed the audience on The Fluidity of Word Meaning: a Challenge for NLP. Yang’s recent research deals with shifting semantic meanings through time, wherein the evolution from initial word meaning to yet-to-emerge usages is turned into a computation problem that can help advance natural language understanding.
Among the afternoon’s multi-track sessions were presentations from Canadian legal tech startups Blue J Legal and ROSS Intelligence. Blue J’s Anthony Niblett explained how his company is using machine learning to predict legal outcomes, while Jimoh Ovbiagele spoke on how natural language processing is changing legal searches and providing easy and low-cost public access to legal documents. Both companies emerged from the IBM Watson 2014 Challenge at the University of Toronto.
Highlights from the technical sessions included MIT Professor and Co-founder of Beijing-based DeePhi Tech Song Han, who spoke on Bandwidth-Efficient Deep Learning. Song’s work focuses on energy-efficient deep learning that optimizes computer architecture. He has previously proposed Deep Compression, which aims to compress neural nets by an order of magnitude without losing the prediction accuracy. During his undergraduate study Song was an exchange student at the University of Toronto.
The afternoon session closed with Nicholas Frosst from Google Brain speaking on Geoffrey Hinton’s work with Capsules, which is the UofT professor’s most recent deep learning research direction.
The event attracted 300 attendees interested in deep learning and AI. Since the founding of the AI-focused Vector Institute in 2017, the University of Toronto has garnered international attention. The visit from Tsinghua University, a member of China’s elite “C9 League” of higher learning institutions, will further Canada-China exchanges in AI.
Journalist: Meghan Han | Editor: Michael Sarazen