Kuaishou Showcases Multimedia Research at ACM MM2021 and Announces Strategic Collaboration with CCF‑MM Committee

From October 20‑24, the ACM International Conference on Multimedia 2021 (ACM MM2021) was held in Chengdu, where Kuaishou participated through papers, competitions, an exhibition booth, and presentations, highlighting its multimedia research and applications.

Kuaishou had two papers accepted: "Contrastive Learning for Cold‑Start Recommendation," co‑authored with NUS and Shandong University, which proposes a contrastive‑learning‑based objective and a simple yet effective cold‑start recommendation framework that outperforms baselines on four public datasets; and "Complementary Factorization towards Outfit Compatibility Modeling," also with Shandong University, introducing the OCM‑CF model that captures harmonious relationships among clothing items and achieves superior performance on real‑world data.

In the Grand Challenge track, Kuashou’s audio‑video network transmission team (Kwai2021) won first place with the DAP system—"Deadline and Priority‑aware Congestion Control for Delay‑sensitive Multimedia Streaming"—addressing the challenge of prioritizing and rate‑adjusting data transmission under limited bandwidth and highly variable channel conditions, a problem intensified by the COVID‑19‑driven surge in real‑time communication applications.

On October 21, Kuaishou signed a strategic cooperation agreement with the CCF‑MM Special Committee, committing to deep joint activities in academic events, research projects, talent cultivation, and data sharing, with senior representatives from both sides attending the signing ceremony.

Following the signing, an open “Multimedia Night” dinner gathered scholars from Tsinghua, the Chinese Academy of Sciences, and UESTC to discuss future multimedia trends, reflecting Kuaishou’s growing academic visibility and collaborative invitations.

As a conference sponsor, Kuaishou operated a technology exhibition booth themed “Kuaishou Future Media Technology Map” and “Kuaishou Tech Collaboration Ecosystem,” demonstrating applications such as ultra‑realistic beauty filters, automatic subtitles, AI‑driven voice‑overs, mixed‑reality effects, and large‑scale multimedia understanding that power content creation, recommendation, safety review, and distribution on the platform.

On October 22, Kuaishou visited the University of Electronic Science and Technology of China, where Dr. Wang Zhongyuan delivered a talk titled “AI Technology in Kuaishou’s Applications and Exploration,” showcasing the company’s in‑house computer‑vision, animation, speech, NLP, and multimodal AI technologies that support the short‑video ecosystem.

In conclusion, after a decade of sustained technical investment, Kuaishou has become a nationally influential short‑video platform serving billions daily, and the ACM MM2021 event provided a valuable opportunity to deepen industry‑academia cooperation and launch a new chapter of joint multimedia research.