iQIYI Tech Salon Session 3 (Beijing): AI Technology Practices and Applications

Despite a strong 40 km/h wind in Beijing, the third iQIYI Technical Salon – "AI Technology Practice and Application" – was held last Saturday. Engineers enjoyed a clear blue sky and a series of in‑depth technical talks.

Talk 1: "AI in iQIYI’s Massive Video Library"

Leo, responsible for intelligent review and news video analysis at iQIYI, presented an overview of iQIYI’s AI landscape, how AI boosts content creation efficiency, enhances user viewing experience, and drives innovative video production.

Talk 2: "Natural Language Processing in Video Scenarios"

Gene explained how NLP enables machines to better understand entertainment‑related video and image content, providing intelligent services to users. He introduced the research topics and application scenarios of iQIYI’s NLP team.

Talk 3: "Exploration and Practice of AI‑Based Video Encoding"

Dr. Xing Huaifei from Baidu Cloud, an expert in video encoding algorithm optimization, shared his experience with AI‑driven content‑adaptive encoding and its application in Baidu’s video feed business. He emphasized the need for higher video quality and lower bandwidth costs in the booming short‑video market.

Talk 4: "iQIYI Traffic Anti‑Fraud Overview"

Lapatto, a specialist in traffic anti‑fraud, described the challenges of fraudulent traffic, the architecture of iQIYI’s anti‑fraud system, and methods for evaluating anti‑fraud effectiveness. The system ensures clean, transparent data for the industry.

Talk 5: "iQIYI Short‑Video Personalized Recommendation Algorithm Practice"

Tony, leading the short‑video recommendation research at iQIYI, discussed how massive short‑video platforms suffer from ineffective and low‑quality recommendations. iQIYI leverages big data and machine‑learning techniques to precisely distribute billions of short videos, continuously optimizing recall and ranking modules to improve product performance.

In summary, the technical sharing was inspiring and full of future possibilities. Readers are invited to leave comments about topics or speakers they would like to see in future salons.