Huang struck again.
Recently, Nvidia announced the creation of a new research division, Gear, a general-purpose embodied agent research lab. Surprisingly, the leaders of the lab are two Chinese post-90s PhDs, Jim Fan and Yuke Zhu.
Fan Linxi and Zhu Yuke have many labels, but the most distinctive is one point: Li Feifei's students. During their Ph.D. studies at Stanford University, they became Li Feifei's **.
Li Feifei, 48, known as the "Godmother of AI", is an iconic figure in global AI research and has long dominated artificial intelligence research at Stanford University. And the student corps she brought out is becoming the mainstay of the global AI community.
Huang Jenxun found two post-90s generations
According to the data, the gear lab established by Nvidia is committed to the research and development of embodied intelligence, which is an area that Huang has long been concerned about, as he once said, "The next wave of artificial intelligence is embodied intelligence", and the importance is self-evident. To this end, he specially found two post-90s generations to take the helm.
Fan Linxi was fascinated by AI as early as high school. When deep learning was still unpopular, Stanford's AI lab was included in his dream destination. At first, Fan Linxi studied computer science at Columbia University, and later came to Stanford University to study for a Ph.D. under the supervision of Professor Li Feifei. During this time, he became OpenAI's first intern.
Zhu Yuke also has a resume of a scholar: Zhejiang University and Simon Fraser University in Canada both have a double bachelor's degree, master's and doctoral degree at Stanford University. After graduating, he worked as an assistant professor at the University of Texas at Austin and was a senior research scientist at NVIDIA.
Earlier as a member of the Stanford Vision and Learning Lab, Zhu studied under Professor Feifei Li and Professor Silvio S**arese. Li Feifei's team developed surreal, a robot training framework, to speed up the learning process, and Zhu Yuke and Fan Linxi are part of the project. In addition, Zhu Yuke, as a co-first author, won the Best ** Award at the International Robotics Conference ICRA 2019.
Going around in circles, now the two are working on robots at NVIDIA. "In the future, every moving machine will be autonomous, and robots and simulated agents will be as ubiquitous as iPhones. Fan Linxi said that Gear Lab is committed to building AI with universal capabilities, which can learn how to act skillfully in multiple worlds, both virtual and real.
Nvidia is also quite bold, equipped with the world's richest embodied intelligence lab. Fan has revealed that the team has enough funds to solve the robot base model, game base model and generative simulation all at once. He believes that 2024 will be the year of robots, the year of game AI, and the year of simulation.
And about the past and future of artificial intelligence, they can't avoid their common teacher Li Feifei.
Who is Li Feifei?
Many years ago, Kai-Fu Lee was asked who would influence the direction of science and technology in the next 25 years, and he thought about it and said the name Li Feifei.
Born in Beijing and raised in Chengdu, Li Feifei studied at the well-known Chengdu No. 7 Middle School. In 1992, at the age of 16, Li Feifei went to the United States with his parents, and later completed a degree in physics at Princeton while helping out in the laundry.
In 2000, Li Feifei studied for a Ph.D. in artificial intelligence and neurobiology at the California Institute of Technology. After graduating, she came to Stanford University as an assistant professor and joined the Stanford Artificial Intelligence Lab.
Li Feifei is a physics fan. "Physics has taught me one thing, to maintain a passion for asking big questions and seeking a North Star. In the step-by-step research, Li Feifei found her interest in AI and shifted her research focus to the field of computer vision, which was very unpopular at the time.
The first "North Star" she found was to teach computers to see.
At the beginning of the 21st century, computer image recognition models could recognize only four types of objects: cars, airplanes, leopards, and faces. Li's vision is that if there is a large enough annotated dataset, it will be possible to train a theoretically "omniscient" computer vision model.
As a result, Li Feifei launched the image dataset project ImageNet in 2007. At that time, the collation and labeling of datasets was dirty work, and the academic and industrial circles did not pay much attention to it. Her peers often suggested that she change her "useful direction", and she often worried about the funds, so much so that she joked that she would go back to open a laundry to subsidize the project funds.
It wasn't until two years later that ImageNet was officially launched, and Li Feifei marked this one with 15 million **The dataset of 20,000 items is free and public, and almost all AI visual recognition learning benefits from this. Today, ImageNet is one of the most well-known large-scale visual databases in the global AI industry-academia community.
Just like the Pandora's box of artificial intelligence has been opened, AI has begun to develop rapidly in image recognition, and the world has seen the commercial value of AI technology. It can be said that Li Feifei has made up the last piece of the puzzle for the explosion of deep learning algorithms.
Since then, the name Li Feifei has resounded in the AI world.
In 2016, Li Feifei joined Google and promoted the establishment of the Google AI China Center, and she once said at the Google China Developers Conference in Shanghai: "Artificial intelligence knows no borders, and the well-being of artificial intelligence also knows no borders." ”
Soon, however, Li Feifei returned to academia and continued to teach at Stanford University, where she led the Stanford AI Lab, the Stanford Vision and Learning Lab (SVL), and the Stanford Institute for Human-Centered Artificial Intelligence.
What's next for AI? Li Feifei found another "North Star" embodied intelligence.
In 2018, she and her team began to combine machine learning and robotic arms to step into the field of embodied intelligence. Currently, her team has developed BEH**ior, a simulated dataset that evaluates the performance of embodied intelligence, with more than 5,000 models of more than 5,000 objects from humans. This data can be used by agents to complete tests in any virtual environment.
In 2023, Li Feifei's team will announce a number of achievements in embodied intelligence one after another. Among them, the NOIR system decodes human EEG signals into a library of robotic skills that can complete tasks such as cooking sukiyaki, ironing clothes, grinding cheese, playing tic-tac-toe, and even petting a robot dog.
Li Feifei also started a VC. In 2023, she joined Radical Ventures, a Canadian venture capital firm, as a scientific partner, but at the same time continues to teach at Stanford University. It is a venture capital firm focused on early-stage projects in areas such as artificial intelligence, robotics, and quantum computing.
Caring about artificial intelligence, but also about human well-being. In his early years, Li Feifei went to ** to conduct research on Tibetan medicine and Tibetan medicine. Today, we continue to research the application of AI in the medical and education fields. Time magazine released the top 100 most influential people in the field of AI last year, and Li Feifei was named among them as a "thinker" in the AI industry.
She brings out an AI legion
Quietly, a group of top AI talents walked out behind Li Feifei.
At Stanford, Li Feifei is a professor in the Department of Computer Science and co-director of the Institute for Human-Centered Artificial Intelligence at Stanford University, with research interests in cognitively inspired artificial intelligence, machine learning, deep learning, computer vision, robotics learning, and artificial intelligence + healthcare, especially environmental intelligence systems for healthcare delivery. Over the years, many outstanding AI talents have been developed in the team.
For example, Wang Gang, the former head of Ali's automatic driving. Wang Gang graduated from Harbin Institute of Technology in 2005 with a bachelor's degree, and then joined the University of Illinois at Urbana-Champaign, where he studied with Feifei Li, computer scientist D**id Forsyth, and Derek Hoiem. At the age of 34, Wang Gang was already a tenured professor at Nanyang Technological University.
In 2017, in order to truly implement the deep learning algorithm technology achievements he had learned for many years, Wang Gang resolutely resigned, entered the industry from academia, became the chief scientist of Alibaba AI Lab, and later incubated and created Alibaba's autonomous driving business.
Surprisingly, in January 2022, Wang Gang resigned from Ali's identity and turned around to found his own cleaning robot company, Xinsheng Intelligence. Although the current project is very secretive, according to the shareholder information, it can be seen that it has attracted institutions such as Hillhouse, Lightspeed, and Pricewater Capital.
Lu Cewu, a professor at Shanghai Jiaotong University, is also one of Li Feifei's students. In September 2014, Lu Cewu, who works in the Department of Computer Science at the Hong Kong University of Science and Technology, applied for a postdoctoral researcher position at the Stanford Artificial Intelligence Laboratory. Lu Cewu was deeply impressed by Li Feifei, who said in his letter of recommendation that few of his peers could reach his academic level, so he invited him to join the laboratory.
In September 2016, Shanghai Jiao Tong University hired Lu Cewu as a professor, researcher and doctoral supervisor, and established its own research team. In the same year, he and fellow Stanford alumnus Wang Shiquan hit it off and founded Feixi Robotics, which is committed to the research and development of advanced humanoid robots and artificial intelligence technology. Founded 8 years ago, the company has stood behind GSR Ventures, Gaorong Capital, Zhen**, Shunwei Capital, Meituan, New Hope Group, Yunfeng and other well-known investors. In June 2022, Feixi Robotics completed the B+ round of financing, with a valuation of more than $1 billion.
In addition, many of Li Feifei's students have entered giants such as Tesla, Microsoft, NVIDIA, and Google to engage in AI-related research work. For example, one of the founding members of OpenAI.
1. Andrej Karpathy, former senior director of artificial intelligence at Tesla, Li Jia, former president of Google AI China Center, etc.
In addition, Guo Wenjing, the founder of PIKA, who was popular before, also studied for a doctorate in the AI laboratory of Stanford University, where Li Feifei worked, before dropping out of school and starting a business. The ImageNet launched by Li Feifei later evolved into the annual ImageNet Challenge, in which Sun Jian, former chief scientist of Megvii Technology, and Ren Shaoqing, vice president of intelligent driving R&D of Weilai, both participated in and won the championship.
Li Feifei's requirements for students are to understand both technology and scenarios. In other words, students here are not only researchers, but also product managers, developers, and entrepreneurs. This experience also laid the groundwork for students to choose entrepreneurship day by day.
Looking at it, AI has become the most eye-catching entrepreneurial track. From this wave of entrepreneurs, we see a significant feature: they are basically from prestigious schools and study under various schools. Instead of the stereotype, professors are now starting to step out of the lab and think about the combination of technology and industry, encouraging students to apply what they have learned and leading them to do something that will truly benefit human life. And Li Feifei's sect is becoming a force to be reckoned with.
AI is penetrating all business verticals and customer, consumer experiences, and is changing the basic structure of our social, economic, and political landscapes. At this year's CES exhibition, Li Feifei appeared to share that unlike the wave of AI craze a few years ago, she judged that AI technology is becoming the real driving force of the next digital revolution.
Humanity is breaking into a sci-fi and unknown page.