Scene Description: IJCAI 2019 is in progress. As the most eye-catching special guest report segment of the conference, among the 7 heavyweight guests this time, we are familiar with Professor Zhou Zhihua. He gave a thematic report on deep learning. Take this opportunity to watch the sharing of the great god at the conference, and to have a deep understanding of his footprints on the AI road, so as to restore his contribution and promotion to the development of AI in an all-round way.
Key words: Zhou Zhihua machine learning IJCAI
As a great god scholar in AI field, Professor Zhou Zhihua has devoted himself to scientific research, talent cultivation and book dissemination. His extraordinary academic contribution has won him numerous awards.
At the top meeting IJCAI 2019, Zhou Zhihua also attended as promised. What kind of sharing did he bring this time? What other interesting stories are there behind this star scientist?
What did Zhou Zhihua share on IJCAI
IJCAI 2019 is halfway through, and many bright spots have appeared. Professor Zhou Zhihua, as a distinguished guest, gave an Invited Talks on the 14th, which attracted the attention of the whole audience.
In the report starting at 2 pm, Zhou Zhihua published “Deep Learning: Why Deep and Is It Only Doable for Neutral Networks? “speech.
He first introduced the connection between deep learning and deep neural networks (DNNs), reviewed the development of deep neural networks, and pointed out the fact that the number of layers of networks is increasing.
However, while the number of layers of the deep neural network increases, some shortcomings are also exposed, such as easy over-fitting, complicated training process and increased computational cost.
This raises the question of whether depth learning is equivalent to DNN (whether depth models can only be constructed through differentiable functional structures)?
Subsequently, Professor Zhou Zhihua pointed out another scheme of in-depth learning, and introduced their research technology GCForest (Multi-Graded Cascade Forest), learning from DNN’s characteristics to express learning, and integrated learning, using the method of decision tree forest to achieve the effect of in-depth model.
It has two parts, Cascade Forest and Multi-grained. Through the introduction and comparison with other models, it claims that it is the first depth model that does not need neural network and BP algorithm to realize. At the same time, it lists that gcForest is superior to other models in actual use cases.
Then he also looked forward to the future of in-depth learning, pointing out the difficulties that may be faced by completely relying on the development of in-depth neural networks, and gcForest’s abandonment of neural networks would be an effective attempt.
However, he also stressed that the technology was still under exploration and there was still a long way to go before it was mature, and then the challenges faced by the model were concretized. The environment in which deep forest and DNNs are good at respectively and the possibility of future combination are also predicted.
In addition to DNN, in-depth learning now includes in-depth forest and other ways.
In addition to this special sharing of dry goods, he will also serve as the chairman of the IJCAI 2021 procedure committee, which is the first mainland scholar to win this honor.
He once said in an interview, “It’s not my cow, it’s a Chinese herd cow. This reflects the recognition of the strength of Chinese scholars. I hope everyone will work together to further enhance the contribution and influence of Chinese scholars and produce excellent research results. 」
How did such a great god reach the peak of machine learning?
Local AI Master: Interest Is the Best Teacher
In 1995, when Zhou Zhihua was a junior, he turned to “Machine Learning: A Way to Artificial Intelligence” in the library, which made him come into contact with machine learning for the first time. The following year, 21-year-old Zhou Zhihua graduated from Nanjing University and chose to stay for further study.
Logo Building of Gulou Campus of Nanjing University
After a period of exploration, he was attracted by machine learning and data mining in the field of artificial intelligence. Driven by his interest and unremitting efforts, his AI road has made a rapid leap forward.
Both the master’s degree and the doctor’s degree took two years to graduate ahead of schedule. He stayed in Nantah to teach in 2001 and was promoted to associate professor in the second year. A year later, he was specially appointed as a professor. He was subsequently qualified as a doctoral tutor and was selected as a Changjiang Scholar in the Ministry of Education in 2006. …
When asked about this general resume and the topic of talent, Zhou Zhihua said, “Interest is the best teacher.” He also quoted Marx to express the importance of interest “when labor becomes the first need of people, not the means of making a living”.
The “native scholar” who completed his studies in Nantah has also been unanimously recognized by the international academic community.
From 2012 to 2016, Zhou Zhihua successively became “Fellow” of ——ACM, AAAI, AAAS, IEEE and IAPR, the five major international societies related to artificial intelligence.
He became the first Chinese in the “grand slam” of AI international society. In 2017, Zhou Zhihua was also elected a foreign academician of the European Academy of Sciences.
In 2015, Professor Zhou Zhihua was elected to AAAI Fellow
In addition, he has also served as the chairman of dozens of international academic conferences. The International Conference on Artificial Intelligence (AAAI 2019) held some time ago also served as the chairman of the Procedure Committee.
Over the years, he has published more than 200 papers in first-class international journals and top international conferences and has won more than 20 invention patents.
In a statistic released recently by Acemap team of Jiaotong University, Professor Zhou Zhihua ranked fifth among Chinese scholars in the recent ten years in the ranking of AI scholars’ paper influence, with 4452 references.
AI Education in Zhou Zhihua and China
In addition to his academic achievements, Professor Zhou Zhihua is also bent on training AI talents in China.
In 2004, under his leadership, lam da (learning and mining from data) research institute of Nanjing university was set up to mainly study machine learning and data mining.
After more than ten years of efforts, LAMDA has become a leading research group in China. Students who are interested in machine learning call it “holy land”. Every year, a special number of students enter for the examination.
There is also a letter from teacher Zhihua to the students who applied for the examination, which helps the candidates analyze whether they are suitable for LAMDA from the teacher’s point of view.
Last year, driven by Zhou Zhihua and others, Nantah set up the Artificial Intelligence Institute, which opened the AI undergraduate education, with Zhou Zhihua himself as the dean. This has undoubtedly brought a shot in the arm when AI talents and education are scarce.
In the face of all kinds of voices from the outside world and the industry, his determination to become an independent court is also very strong. In a media interview, he said, “instead of tinkering with the existing framework, it is better to build from scratch according to the characteristics of artificial intelligence.”
Nantah’s College of Artificial Intelligence opened its first batch of undergraduate enrollment in September of 18 years. Zhou Zhihua will also give lectures in person. When talking about the aims of Nantah’s Institute of Artificial Intelligence, he revealed that he had been studying the undergraduate education plan for a long time.
He said that the college is not to form training courses such as “artificial intelligence 120 days from introduction to mastery”, but to cultivate high-level talents with innovative ability in AI field and capable of solving key technical problems.
The most classic reading in the field: watermelon book
Actually speaking of Zhou Zhihua, many people are more familiar with the book Machine Learning. He used humorous and popular language to penetrate the main points and difficulties of machine learning. In this book, watermelon is also used as an example to make theoretical knowledge fresh, which is also called watermelon book.
Watermelon book has been published many times since it was published in 2016, and it is also often active in the best-selling lists of major e-commerce platforms, opening the door to machine learning for countless scholars.
He once revealed two purposes of this book in a conversation:
The first is to draw the overall outline for beginners to understand the field of machine learning and know what is probably in this field so that they can further study according to their needs and interests.
The other is to sort out the knowledge system after facilitating readers to read a variety of materials.
Interestingly, in the 2017 “Jingdong Literature Prize” award, watermelon book was listed as the TOP 5 of the year’s new works award.
The beginning of the book is like this, might as well feel Zhou Zhihua’s writing skills:
In the evening, the streets were wet after a slight rain, and the warm west wind was blowing. Looking up at the sunset sky, well, tomorrow will be another fine day.
Walking to the fruit stand, I picked a green watermelon with curled root and raised voice. I was full of expectation for the cool feeling of thin skin, thick flesh and sweet flesh. I thought happily that I had worked hard this semester, and the basic concepts were clear, and the arithmetic homework was easy to come by. The result of this course must not be bad!
Together, we look forward to the study of the great god Gengniu
As a big bull in the field of artificial intelligence, Zhou Zhihua is undoubtedly a leader in the industry. However, he has said on many occasions that his greatest pleasure lies in scientific research and his greatest wish is that his research team can do better research and have more influential work in the world.
When asked how he chose machine learning in an era when AI was not valued, his answer was simple but powerful: “stick to your own judgment and do important things.” 」
I wish Mr. Zhou Zhihua can realize his wish and make more and better achievements.