This report summarizes the most interesting exhibits and seminars at AI Expo Tokyo, held on April 3 to 5, 2019.
Many Lionbridge AI members attended AI Expo and we learned a lot about the interesting applications of artificial intelligence. If you’re interested in attending AI conferences too, here’s a complete schedule of the upcoming AI conferences worldwide in 2019.
AI Expo is the largest artificial intelligence exhibition and conference in Japan, in reference to exhibit space and number of exhibitors. This year was the 3rd annual AI Expo, and there were about 250 exhibitors and 50,000 participants over three days.
AI Expo Exhibits
AI Expo had hundreds of exhibits about how to organize and use big data. In addition to the main conference, many booths had small stages where the exhibitors gave short presentations for anyone passing by that was interested in learning more about their service.
GAUSS is an AI startup company that uses machine learning in the marketing and sports industries. For example, they created an algorithm to predict the winner in horse races. GAUSS also received media attention recently because they used machine learning to predict the new Japanese era name when Crown Prince Naruhito accedes to the Imperial Throne. (Unfortunately, they did not use Japanese literature works to train the model and their prediction was incorrect.)
SAKURA internet is a data center services business that house their clients’ servers and provide related services such as maintenance and management on the client’s behalf. Their booth had a human recognition demo that worked in real time, attracting many of the participants that walked past. Their demo used face recognition to predict people’s gender, age, and emotion Other booths also had similar face recognition demos, with different features such as counting the number of times each person walked by their booth.
The image below is SAKURA internet’s demo. As you can see, it was so popular and crowded that it’s hard to read the captions for each person…
AI Expo was also encouraging the attendees to learn more about artificial intelligence and machine learning. One of the event sponsors, Japan Deep Learning Association, is an organization that aims to develop and promote the AI industry to both beginners and experienced engineers. Japan Deep Learning Association’s booth had AI textbooks, including official study materials for the AI general aptitude exam that the organization administers three times a year.
SkillUp AI was another booth at AI Expo that provided practical machine learning courses, including courses for people preparing for the AI general aptitude exam.
AI Expo Conference
The Outlook Presented by AI Business Leaders
At the keynote session, Yuki Hamada, CPO of Mercari, Inc., discussed how recent advances in technology have shaped Mercari’s business, services, values, and customer support. We’ve already entered the new technological era, and it’s important to prepare for AI so that you don’t get left behind. Hamada encouraged people with a background in mathematics to be proactive about studying machine learning. Even if you aren’t a math person, Hamada recommended online courses where you can begin the basics of machine learning.
The Future Prospect of AI Strategy in Japan
Yuichiro Anzai from the Japan Society for the Promotion of Science also gave an enlightening presentation about the future perspectives of how AI will change society. He focused on the industrial, employment, and educational structures, research and development, and relationship between AI and humans.
We’re currently repeating some of the same habits as we had in previous technological revolution eras, such as underestimating human knowledge and debating about what is artificial intelligence, whether artificial intelligence will become superior to humanity, and the threat to humanity. But there are some new concepts in this current artificial intelligence boom. For example, we haven’t run into any clear limitations of machine learning, and we’re able to think about use cases for artificial intelligence in a wide range of industries such as legal, finance, and education. To prepare for the new technological era, the Japanese government recently announced a new plan to require all university students to take a beginner AI course.
Insights of AI Ventures: The Future of Artificial Intelligence
Yousuke Okada, founder and director of ABEJA, gave his presentation around the concept of “technopreneurship,” a term coined by combining the words technology and entrepreneurship. ABEJA highly values the spirit of mastering both technology and business.
ABEJA provides an end-to-end system for artificial intelligence, and supports companies in providing AI tools to their customers. Okada introduced some of the use cases of ABEJA. For example, they used skeleton annotation on workers to improve the process at a large Japanese factory. The project involved monitoring the workers and measuring the whole operation, to find areas for improvement.
ABEJA also worked with a major household appliances company to automate their customer support. For example, if a customer had a broken air conditioning system, it’s important to know which piece is broken so they can order a replacement. With deep learning, the client company could better pinpoint the issues, reduce the amount of clarifying questions with their customers, and fix the problem in the first attempt. Of course, all of this reduces costs and makes customers happier.
Introduction on AI, its recent trends, and future
Naohiko Uramoto, President of the Japanese Society for Artificial Intelligence, gave a presentation that covers the latest trends in AI research, development, and applications. He discussed the future of AI and the impact on society.
He introduced Andrew Ng’s criteria for how to choose your first AI project:
- Does the project give you a quick win?
- Is the project either too trivial or too unwieldy in size?
- Is your project specific to your industry?
- Are you accelerating your pilot project with credible partners?
- Is your project creating value?
Uramoto also discussed the common problems with AI data. Efficient, high-quality algorithms can only be created from high-quality training data. But it’s difficult for data scientists to collect and organize large amounts of high-quality, unbiased data that is representative of all possible scenarios. Perhaps some of the exhibitors’ services can make a significant impact for artificial intelligence by helping to solve this problem.