Almost a thousand attendees filled the century-old St James Theater in the heart of Old Montréal for the World Summit AI Americas. Held April 10 – 11, the conference welcomed AI leaders from industry and academia, who shared their AI experiences, insights, and analysis.
Sessions included deep dive tech talks that broke down technical details for a general knowledge audience, as well as detailed use cases examining practical applications of AI in enterprise and how the tech has transformed business. There were also workshops hosted by startups, scale-ups and unicorns; which addressed challenges such as machine learning in drug development, the use of AI for social good, and how governments can harness the power of AI to improve infrastructure.
On day one attendees filled the main conference room to hear Turing Award 2018 Honoree Yoshua Bengio. a Université de Montréal professor and head of the Montréal Institute of Learning Algorithms (MILA). An AI chic atmosphere was informed by multicoloured stained glass windows, loud DJ music and a dazzling welcoming video.
Bengio dived straight into his topic, Moving Beyond Supervised Deep Learning, arguing what is missing from current machine learning are understanding and generalizations beyond the training distribution: “So long as machine learning models ‘cheat’ by relying only on superficial statistical regularities, they remain vulnerable to out-of-distribution examples, adversarial examples.”
Bengio added “Humans generalize better than other animals thanks to a more accurate internal model of the underlying causal relationships.”
Gary Marcus, the New York University Professor with a history of concerns and criticisms regarding deep learning, took to the same stage on day two. He acknowledged deep learning is good for certain things, such as object, face and speech recognition; finding patterns in board games and radiology; or guessing the colors of patterns in old black and white movies. But that’s basically where it ends.
Marcus argued that cognition is complicated, and we should “stop looking for silver bullets … learning isn’t just about big data and number-crunching.” To illustrate his point Marcus showed pictures of his young daughter attempting to squeeze through the back of her chair: “(Learning) is also about things like learning through exploration, problem solving, and an intuitive understanding of how the world works.”
The World Summit AI Americas 2019 attracted 955 attendees from 38 countries across enterprise, big tech, startups, investors and science and academia. Next year’s World Summit AI Americas will return to Montréal, one of the AI superclusters advocated by the Canadian government and a growing international AI hotbed with multidimensional collaborations uniting universities, governments, and the private sector.
Journalist: Fangyu Cai | Editor: Michael Sarazen
Pingback: 機器學習江湖的見樹見林與因果 – 樹洞 Tree Hole 2.0
Im so happy I have finally reached your site.
Thanks a lot for your function, it seems an individual spent hours writing it.
The final time I had the same impression regarding the particular
article was right here http://www.rtjournal.org/should-i-go-to-college/. It’s today very hard to locate something so deeply researched.
You manufactured my day!