NAVER Demonstrates AI Leadership with 51 Papers Accepted by International Academic Conferences So Far This Year
NAVER Demonstrates AI Leadership with 51 Papers Accepted by International Academic Conferences So Far This Year
NAVER Demonstrates AI Leadership with 51 Papers Accepted by International Academic Conferences So Far This Year
- Over 40% of new technologies published in the papers are used in NAVER services, such as CLOVA Note and X-eye, to enhance user experience
- NAVER, the leader in hyper-scale AI, to accelerate its AI R&D through active collaboration with academia
2021-08-02
NAVER Corporation announced on August 2 that the company has had 51 research papers on artificial intelligence (AI) accepted by top academic conferences around the world so far this year, including 43 in the first half alone, which matched its 2020 publication total. NAVER is establishing itself as a global AI leader beyond Korea by increasing its AI research performance every year.
Among the conferences that have accepted NAVER’s AI research papers are Conference on Computer Vision and Pattern Recognition (CVPR), Associated for Computational Linguistics (ACL), Associated for the Advancement of Artificial Intelligence (AAAI) and The International Conference on Learning Representations (ICLR).
NAVER’s AI research papers cover a variety of topics, including computer vision, natural language processing, deep learning, and voice & speech technology. In particular, 18 papers on NAVER’s research in voice & speech technology were accepted by prestigious conferences of the same theme, with nine accepted to the International Conference on Acoustics, Speech, & Signal Processing (ICASSP) and another nine at Interspeech. Also, NAVER will present eight papers at the International Conference of Computer Vision (ICCV) in October, proving its leading technology in the computer vision field.
In addition, NAVER has presented and plans to present papers at various academic conferences such as Conference on Human Factors in Computing Systems (CHI), North American Chapter of the Associated for Computational Linguistics (NAACL), International Conference on Machine Learning (ICML), and Knowledge Discovery and Data Mining (KDD).
Over 40% of new technologies featured in the papers accepted by conferences this year are already applied to NAVER products and services and contribute to enhanced user experiences. For example, the research on speaker recognition technology presented at ICASSP was used in CLOVA Note to enhance the accuracy of distinguishing participants by their voices. Also, the image classification model, RexNet, presented at CVPR and the optimization technique in deep learning, AdamP, presented at ICLR, are the key technologies that enabled X-eye 2.0, the pornography filtering system. In addition, AI research results have been applied to various services and solutions, such as CLOVA Dubbing, CLOVA Forecast and CLOVA CareCall.
This remarkable academic achievement is the result of NAVER's massive investment in AI technology, as well as active collaboration with research institutes both in Korea and overseas. NAVER has worked with the best universities such as Yonsei University, Korea University, Gwangju Institute of Science and Technology (GIST), Inha University, and Ulsan National Institute of Science and Technology (UNIST), on AI research. The company also established AI centers with about 100 researchers at Seoul National University and Korea Advanced Institute of Science and Technology (KAIST) each and closely works with them on next-generation AI research.
In addition, NAVER has expanded its global AI R&D network by establishing joint research centers with Hanoi University of Science and Technology (HUST) and Posts and Telecommunications Institute of Technology (PTIT) in Vietnam and signed a Memorandum of Understanding (MoU) with the University of Tübingen for research on ‘Trustworthy AI.’ As a result, 21 out of 51 papers accepted by conferences this year were a product of NAVER’s collaborative work with global research institutes.
NAVER is also working with top AI experts, including Professor Cho Kyung-hyun of New York University, Professor Andrew Zisserman of Oxford University, and Professor Jun-Yan Zhu of Carnegie Mellon University. A paper on speaker verification at the movies, which was presented at ICASSP this year, was the result of collaboration between NAVER Corporation and University of Oxford. This month, Professor Cho Kyung-hyun of New York University and researcher Kim Kyu-wan of NAVER AI LAB will present the results of their joint study at ACL 2021 on how to enhance the efficiency of the ‘Transformer’ models, which is GPT-3’s foundation technology.
NAVER plans to further accelerate its research focusing on hyper-scale AI technology. The company unveiled ‘HyperCLOVA’, the world’s first hyper-scale AI language model in May and started commercializing the new technology through NAVER services, such as query spelling corrections, featured collections creation in NAVER Shopping, and one-sentence summaries of customers' reviews. NAVER has developed HyperCLOVA Studio, a no-code AI platform, and kicked off its closed beta test internally.
“NAVER not only contributes to academia through AI research but also integrates findings into actual services that can enhance the user experience,” said Jung-woo Ha, head of NAVER AI LAB. “While the global competition in AI has heated up, NAVER, a leader in AI, will strive to secure competitiveness and strengthen the AI technology ecosystem in Korea.”
-END-