Yitu Technology CTO Yan Shuicheng is reported to have resigned and may join Southeast Asian e-commerce company Shopee

Yitu Technology CTO Yan Shuicheng has resigned and will join a Southeast Asian e-commerce company. He has now gone to Singapore to start building a team. However, Yitu Technology has not yet officially confirmed the authenticity of this news.

As one of the "Four AI Dragons", Yitu Technology is currently in a critical period of its sprint to the Science and Technology Innovation Board. It is shocking that Yan Shuicheng resigned at this time.

According to 36Kr, Yan Shuicheng will join the Southeast Asian e-commerce company Shopee. A Shopee insider told 36Kr: "Shopee has recently recruited a lot of talents in technical algorithms, and we can already find information about (Yan Shuicheng) in the internal system."

It is worth noting that Yan Shuicheng joined the National University of Singapore in 2008 and currently leads the Machine Learning and Computer Vision Laboratory and holds a tenured position at the university. Perhaps he has more long-term considerations in changing his club this time.

Yan Shuicheng studied mathematics at the School of Mathematical Sciences of Peking University as an undergraduate. Under the guidance of his advisor Professor Cheng Gansheng, in addition to studying computational mathematics and applied mathematics, he also "did something else" to study topics related to pattern recognition and machine learning. This experience laid the foundation for his future research in artificial intelligence.

In 2001, Yan Shuicheng entered Microsoft Research Asia for an internship, where he studied under Li Ziqing, an expert in facial recognition and intelligent video surveillance, and Zhang Hongjiang, the "founder" in the field of computer video retrieval research. He also worked with He Xiaofei, the former director of Didi Research Institute, to publish four papers with over a thousand citations each, which had a huge impact in the field. One of the researches worth mentioning is that they cleverly and simply connected Subspace Learning and Manifold Learning, two very popular fields at that time.

Later, Yan Shuicheng met Zhu Long and Lin Chenxi, the founders of Yitu Technology, at Microsoft Research Asia. It can be said that the seed of Professor Yan Shuicheng joining Yitu Technology was planted during this period.

In 2006, Yan Shuicheng did postdoctoral research at the University of Illinois at Urbana-Champaign (UIUC), studying under computer vision master Thomas Huang.

In December 2007, Professor Yan Shuicheng joined the National University of Singapore, beginning the most productive eight years of his academic career. In five years, his team has won more than ten championships or honorable mentions in the core competitions in the field of computer vision, namely the PASCAL Visual Object Class and ImageNet Large Scale Visual Recognition Challenges, as well as more than ten (best) student paper awards.

The core 1×1 convolution of the "Network in Network" (NIN) network structure proposed by Yan Shuicheng's team has been used as a standard module by almost all computer vision deep learning models in recent years. Later models such as GoogleNet and ResNet also borrowed its ideas. In addition, the "Purine" developed by his team is the world's first open source deep learning system that supports multiple machines and multiple GPUs.

As of now, according to Google Scholar, Yan Shuicheng's papers have a total citation of 71,063, an H-index of 96, and he has been selected as a "Thomson Reuters Highly Cited Scholar" five times. He is currently an IEEE Fellow, IAPR Fellow, and ACM Distinguished Scientist.

In 2015, Yan Shuicheng accepted Zhou Hongyi's invitation to serve as vice president, chief scientist, and dean of the Artificial Intelligence Research Institute of 360 Group. In the following three years, Yan Shuicheng led the AI ​​team he formed to apply AI technology to various products under 360.

More than a year ago, Professor Yan Shuicheng bid farewell to 360 and took up the position of CTO of Yitu Technology, mainly responsible for basic AI research. He led the team to further consolidate Yitu's technical advantages in the basic theories of artificial intelligence and original algorithms, and provided strong technical support for Yitu's commercialization scenarios.

As for why Yitu was chosen out of the vast sea of ​​AI, Professor Yan Shuicheng once mentioned in an interview with AI Gold Rush that it was mainly due to three considerations.

Firstly, he was familiar with Zhu Long and Lin Chenxi, and he admired their low-key and pragmatic approach, and their ability to complement each other.

He said that he had spent a long time in the same office with Yitu founder Leo (Zhu Long) at Microsoft Research Asia, and had also met co-founder Lin Chenxi at Microsoft Research Asia. He also knew Yitu's development process very well, so when Leo invited him, the decision to join Yitu was very simple and quick.

Secondly, I agree with Yitu’s three major development concepts.

In Professor Yan Shuicheng's opinion, the layouts of CV startups now have their own characteristics. From Yitu's perspective, he is more attracted by the integrated hardware and software solutions, multimodal technology accumulation, and technology-for-good product concepts.

Third, I want to know the secret behind its continued dominance.

Professor Yan Shuicheng told AI Gold Rush that he was very familiar with Yitu's development history, but Yitu's technical team had always maintained a certain mystery, which made him curious and drove him to want to get to know the team more deeply. He was also very willing to do something different with such a team.

During his tenure at Yitu, Yan Shuicheng participated in proposing the "."

He believes that to make an AI chip, one must first think clearly about its typical application scenarios, and also have a prediction of the development trends of chip design and future AI algorithms. At the same time, only when driven by high-efficiency AI models and high-efficiency AI chips can "AI" be transformed into "Affordable Intelligence", so that end users can afford it and use it well.

Yan Shuicheng also once mentioned that compared with giant enterprises, universities or start-ups are more advantageous in achieving a closed loop of business development. As for the reason for leaving Yitu Technology, neither he nor Yitu Technology has made any statement yet. In addition, Yitu Technology is currently facing considerable challenges in the listing stage.

Yitu started out with facial recognition, but domestic supervision of personal privacy data is becoming increasingly strict. On December 1, 2020, Tianjin passed the "Tianjin Social Credit Regulations", which stipulates that no organization or individual may illegally collect, aggregate, use, process, or transmit social credit information, or illegally buy, sell, provide, or disclose social credit information. The Civil Code promulgated on May 28, 2020 (implemented on January 1, 2021) proposed in the "Personal Rights Code" that the processing of personal information, including facial features, should follow the principles of legality, legitimacy, and necessity.

On the afternoon of November 20, 2020, the first-instance judgment was made in the Hangzhou Court. The verdict was: Wildlife World compensated the plaintiff for the loss of contractual benefits and transportation expenses, and deleted the facial feature information submitted by the plaintiff. Perhaps in the eyes of many people, this is just a breach of contract case, but the privacy issues involved in this matter determine the direction and boundaries of the application of facial recognition technology. Therefore, the significance of the case goes far beyond general breach of contract litigation.

Information disclosed in Yitu Technology's prospectus shows that Yitu has not yet turned a positive net profit during the reporting period, and its cumulative losses as of June 2020 were 7.22 billion yuan, of which the losses caused by changes in the fair value of preferred shares totaled 5.1 billion yuan in four years.

Business supervision, long-term losses, departure of key personnel...How will Yitu face the numerous challenges on the road to listing?

References:

https://36kr.com/p/1037216538930179

https://mp.weixin.qq.com/s/XBIlYLteovoo0LJAMV2lLQ