NAVER Proves Global AI Leadership, Presenting 18 Papers at World’s Top AI Conference ‘CVPR’
NAVER Proves Global AI Leadership, Presenting 18 Papers at World’s Top AI Conference ‘CVPR’
NAVER Proves Global AI Leadership, Presenting 18 Papers at World’s Top AI Conference ‘CVPR’
- NAVER CLOVA, NAVER’s AI division, presented 14 papers, three workshop papers and made one oral presentation at CVPR
- NAVER’s AI division introduced computer vision-based technologies and study results that are applicable on real services
- Three papers from SNU-NAVER collaboration were also presented at the conference, showing results from the joint AI R&D centers on hyperscale AI
- NAVER CLOVA published 60 papers at global top-tier conferences in the first half of 2022
2022-06-24
NAVER CLOVA, the artificial intelligence (AI) division of NAVER Corp. (KRX: 035420), reaffirmed NAVER’s global competitiveness in AI technology as it presented 18 papers at the world’s top AI conference, Computer vision and Pattern Recognition (CVPR) 2022.
NAVER CLOVA is the first among Korean companies to present double-digit papers at CVPR. The team’s achievement includes one oral presentation, 14 papers and three workshop papers. A total number of papers submitted by NAVER and its affiliates reached 22 at CVPR, including papers presented by NAVER LABS Europe and WEBTOON.
Established in 1983, CVPR is the most influential conference in vision AI and computer engineering*. The Institute of Electrical and Electronics Engineers (IEEE) and the Computer Vision Foundation (CVF) cohost the annual academic conference. This year, the event took place online and offline at the Ernest L. Morial Convention Center in New Orleans, US, from June 19–24.
* CVPR ranks fourth among Google Scholar’s Top Publications in Science. The publication also tops in Engineering & Computer Science category on Google Scholar.
NAVER presented on fundamental technologies in computer vision and other innovations that will support application of those technologies into NAVER services. One of the studies identifies a novel strategy to manage incorrect data labels in continual learning*. The study is expected to help enhance service quality in the real world where many data labels are inaccurate. NAVER also presented a study on a new method to increase minority samples through long-tail data training that can improve various services, including its e-commerce platform NAVER Shopping**.
* ‘Online Continual Learning on a Contaminated Data Stream with Blurry Task Boundaries’ — Bang Ji-hwan (NAVER CLOVA), Koh Hyun-seo (GIST & NAVER AI Lab intern), Park Seul-ki (SNU & NAVER AI Lab intern), Song Hwan-jun (NAVER AI Lab), Ha Jung-Woo (NAVER AI Lab) and Choi Jong-hyun (Yonsei University) (arXiv:2203.15355)
** ‘The Majority Can Help The Minority: Context-rich Minority Oversampling for Long-tailed Classification’ — Park Seul-ki (SNU & NAVER AI Lab intern), Hong Young-kyu (NAVER AI Lab), Heo Byeong-ho (NAVER AI Lab), Yun Sang-doo (NAVER AI Lab) and Choi Jin-young (SNU) (arXiv:2112.00412)
NAVER’s industry-university collaboration on AI showed impressive achievements with the presentation of three papers from Seoul National University (SNU) – NAVER Hyperscale AI Center, including a study to develop AI with fairness-aware learning*. NAVER established joint AI R&D centers with SNU and the Korea Advanced Institute of Science and Technology (KAIST) in 2021, and has been working closely with academia in the hyperscale AI field while actively collaborating with institutes to foster outstanding talents in AI through industry-academia internships.
* ‘Learning Fair Classifiers with Partially Annotated Group Labels’ — Jung Sang-won (SNU & NAVER AI Lab intern), Chun Sang-hyuk (NAVER AI Lab) and Moon Tae-sup (SNU) (arXiv:2111.14581)
NAVER participated in CVPR as a gold sponsor and operated a booth to recruit AI talents during the event. Various NAVER teams, including NAVER CLOVA, WEBTOON, NAVER LABS and NAVER LABS Europe, shared their most recent findings in AI and actively engaged with visitors to recruit talents for Team NAVER. Hundreds of computer vision researchers from all over the world visited the booth, showing high interest in NAVER’s AI technology.
NAVER also hosted two networking events ‘NAVER Night’ on June 20 and June 23. NAVER introduced the company and affiliates, and NAVER’s researchers interacted with around 200 researchers who attended NAVER’s event.
NAVER CLOVA has been proving its AI leadership as it has presented 60 papers at global top-tier AI conferences, such as the Association for the Advancement of Artificial Intelligence (AAAI); Conference on Human Factors in Computing Systems (CHI); International Conference on Acoustics, Speech and Signal Processing (ICASSP); International Conference on Learning Representations (ICLR); International Conference on Machine learning (ICML) and Interspeech in the first half of 2022. Based on its outstanding achievement, the team is becoming one of the global AI leaders after earning the title of ‘No.1 AI expert in Korea’.
“Following the 12 papers at ICLS in January 2022, the presentation of double-digit papers at CVPR means that NAVER CLOVA’s capability in computer vision and AI is top notch. It is very rare to achieve this level,” said Ha Jung-woo, Head of NAVER AI LAB. “NAVER will continue to provide innovative values to users based on the technology developed from our continuous investment in advanced AI research.”
END