SHENZHEN, China, June 17, 2021 /PRNewswire/ -- The 2021 DIGIX GLOBAL AI CHALLENGE was jointly launched by the Jiangsu Association of Artificial Intelligence (JSAI), Huawei Consumer Cloud Service Dept, and Huawei Nanjing Research Center in Beijing on June 10. At the launch ceremony, Liu Dongfang, Director of the Open Platform Business Dept,
Since debuting in 2019, the DIGIX GLOBAL AI CHALLENGE has attracted broad interest from student algorithm enthusiasts and industry experts alike. The contest's proposals, which encapsulate the latest AI theories and business practices, have attracted an increasing number of contestants. Contestants interested in putting their skills to the test can enroll on the HUAWEI Developers website.
More Proposals, Spurring the Imagination
This year's DIGIX GLOBAL AI CHALLENGE comes with more challenging proposals than ever before. The five proposals, which address predicting user retention during different periods based on multiple targets and views, article quality evaluation based on multi-model transfer learning, video recommendations based on multi-objective optimization, search rankings in multimodal and multilingual contexts, and menu identification leveraging few-shot learning models, all aim to resolve longstanding real world challenges.
The first proposal, predicting user retention during different periods based on multiple targets and views, is centered around methods for improving multi-day user retention through user information-based behavior chain vectorization, which is crucial for exploring next-generation machine learning models and multimodal vectorization.
The next proposal — article quality evaluation based on multi-model transfer learning — gets to the root of the data feed field. It is widely acknowledged that accurate article quality evaluation models, once available, will propel technological innovations and breakthroughs in the natural language processing (NLP) field, and the development of NLP algorithms as well.
The proposal addressing video recommendations based on multi-objective optimization tests contestants' ability to prioritize user implicit feedback, which is crucial for generating accurate recommendations.
Despite encountering a myriad of recommendation pop-ups, users tend to conduct searches to get the information they need. However, optimizing the user search experience and ranking search results remains a challenge. The proposal on search rankings in multimodal and multilingual contexts addresses this issue.
The final proposal, i.e. menu identification leveraging few-shot learning models, has been designed to boost image text recognition accuracy, explore new technologies in the image text recognition field, and contribute to advances in this field.
All five proposals are both pragmatic and forward-looking, requiring that contestants build algorithmic models based on given datasets, solve actual problems, and implement cutting-edge AI technologies. They aim to inspire innovation in the AI algorithm field, and create the conditions for a flourishing business environment.
Expert Mentorship, Empowering AI Talent
Contestants will enjoy the opportunity to pick the brains of noted authorities in AI, including Zhi-Hua Zhou, professor at Nanjing University, ACM/AAAI/IEEE Fellow; Pascal Van Hentenryck, professor at Georgia Institute of Technology and Program Co-Chair of Conference on Artificial Intelligence (AAAI) 2019; and Giuseppe De Giacomo, professor at Univ. Roma "La Sapienza" and Program Chair of European Conference on Artificial Intelligence (ECAI) 2020. In addition, 15 AI experts from China will form a distinguished judging panel, evaluating works during the different stages of the competition. These experts will offer invaluable feedback, helping contestants create more effective algorithms, and put what they have learned into practice.
US$305,000 in Prizes, Funding Tomorrow's AI Pioneers
A total of US$305,000 has been set aside as prize money for the contest, a substantial increase over previous years. For each of the five proposals, each champion, first runner-up, second runner-up, and honorary mention team will receive US$30,000, US$15,000, US$10,000 and US$2,000 respectively.
After signing up for the contest, contestants can opt to participate in the online competition by submitting entries for any number of proposals. 6 teams from each proposal and 30 teams in total will be selected for the elite final, based on the automatic scoring and expert review results. Finalists will participate in a special online competition, and give presentations.
As a world-renowned AI algorithm competition, the DIGIX GLOBAL AI CHALLENGE offers scholars and talented student algorithm developers a unique chance to explore futuristic technologies, and address challenges that hinder innovation in the digital era. It will continue to provide a platform for talented student algorithm developers to put their skills on full display and participate in technological and professional exchanges, while helping AI algorithm professionals think outside the box.
Click here to get started: https://developer.huawei.com/consumer/en/activity/digixActivity/digixdetail/201621219634131423