The International Conference on learning representatives, the top conference in the field of deep learning, will officially open online on April 25.
Due to the impact of the epidemic, this is the first International Ai Academic Summit held in Africa, but unfortunately it has become the first online virtual Academic Summit, and all accepted papers have to be pre recorded and displayed in video. Although the opportunity of face-to-face communication with the tycoons has been reduced, it is also a good choice to be able to sit at home and listen to the talk of the tycoons.
ICLR 2020 has received 2594 papers and 687 papers, including 48 orals, 108 spotlights and 531 posters; the admission rate is 26.5%, slightly lower than last year's 31.4%.
ICLR 2020: outstanding performance of Chinese
ICLR 2020 has received 687 papers from 2566 authors. Among them, there are 15 authors selected for more than 5 articles (including), 15 authors selected for 4 articles, 53 authors selected for 3 articles, 246 authors selected for 2 articles, and 2239 authors selected for only 1 work. See the figure below for details.
Among them, Sergey Levine, an associate professor from UC Berkeley, was selected as the top scholar, with 13 papers in total; in neurips 2019, he topped the list with 12 papers, while in ICML 2019, he also had 6 senior high schools. I have to admire that Sergey has such a high yield and has an unlimited future!
Zhu Jun, a professor in the computer department of Tsinghua University, and song Le, a lifelong associate professor at Georgia Institute of technology, respectively, have received seven papers, ranking second.
Professor Zhu Jun is a professor in the computer department of Tsinghua University. He is mainly engaged in machine learning, Bayesian statistics and other basic theories, efficient algorithms and related application research. He has published more than 100 academic papers in important international journals and conferences.
Song Le is a lifelong associate professor in the Department of computational science and engineering, Georgia Institute of technology, and vice director of the machine learning center. His main research fields include embedded methods of kernel function and deep learning, large-scale algorithms and efficient systems of machine learning, static and dynamic network analysis, etc. He has won many top international awards in machine learning, including the best paper awards in neuroips 2017, recsys 2016, aistats 2016, etc.
Cho Jui Hsieh, Professor of the University of California at Davis, Jimmy Ba, assistant professor of the University of Toronto, Wang Liwei, Professor of the school of information science and technology of Peking University, pushmeet Kohli, chief scientist of deepmind, and Tom Goldstein, associate professor of the computer department of the University of Maryland, were selected and ranked third respectively.
In terms of the number of authors of each paper, most of the papers received by ICLR 2020 have 3-4 authors, of which the most are 4 authors, 163 in total, 159 in total, and 15 in total for papers with 10 or more authors, and 15 in the most one.
According to the nationality of the author, there are 412 papers participated by Chinese, accounting for 60% of the total number of papers. Among them, 301 papers were written by Chinese, accounting for 73% and 44% of the total.
It can be seen that the Chinese have contributed a lot to the ICLR 2020. The word cloud map generated by the key words of the contribution highlights hot topics such as deep learning, reinforcement learning, representation learning, generation model, graphical neural network, etc.
The figure below shows the selected Chinese scholars with more than 3 articles.
Chinese stars in ICLR 2020
Let's take a detailed look at the number of Chinese scholars who have published more than 4 articles in ICLR 2020?
Zhu Jun
Professor Zhu Jun, who has 7 papers selected, is ranked first in ICLR 2020 Chinese contribution list with song Le. He is a professor in the computer department of Tsinghua University and deputy director of the State Key Laboratory of intelligent technology and systems. He received his bachelor's and doctor's degrees in computer science from Tsinghua University, and then worked as a postdoctoral fellow at Carnegie Mellon University. He returned to Tsinghua in 2011 to teach. In 2013, Zhu Jun was selected as "Ai's 10 to watch" by IEEE intelligent systems. His research fields are mainly machine learning, data mining, nonparametric Bayesian method, maximum interval learning, etc., and he has published more than 100 papers in the top international conferences and journals of machine learning, such as ICML, neuroips, jmlr, PAMI, etc.
Song Yue
Professor Song le of Georgia Institute of technology has 7 papers selected this time, and 5 papers selected in neurips 2019. He is an absolute high-yield author. Song Le is a lifelong associate professor in the Department of computational science and engineering, Georgia Institute of technology, and vice director of the machine learning center. He studied at South China University of technology, received his Ph.D. in machine learning from Sydney University and nicta. After finishing his doctoral degree at Carnegie Mellon University, he joined Google's machine learning department for research. His main research fields include kernel function and deep learning embedding method, large-scale algorithm and efficient system of machine learning, static and dynamic network analysis, etc. He has won many top international awards for machine learning, including nips' 17 Best Paper Award for machine learning and materials science seminar, recsys' 16 best paper award for deep learning and recommendation system seminar, etc.
Wang Liwei
Professor Wang Liwei of Peking University has 6 articles selected in ICLR 2020. Professor Wang Liwei received his bachelor's and master's degrees from Tsinghua University and his doctor's degree from the school of mathematics of Peking University. Since 2005, he has taught in the school of information, Peking University. His main research interests are machine learning theory, and he has published more than 60 papers in the top conference of machine learning, neurips, colt, ICML, jmlr and PAMI. On the Margin Explanation of Boosting Algorithms, published in 2008 at the COLT Conference on machine learning theory, is the first article of Chinese mainland scholars at the conference. He was selected as AI's 10 to watch in 2010 and is the first Asian scholar to win the award.
Feng Jiashi
Five papers were selected for ICLR 2020. Feng Jiashi is currently an assistant professor in the Department of electronic and computer engineering, National University of Singapore, and head of machine learning and Vision Laboratory. He holds a bachelor's degree in automation from University of science and technology of China and a doctor's degree in electronic and computer engineering from National University of Singapore. The current research direction is image recognition, deep learning and robust machine learning for big data. He has won the iccv'2015 task-cv best paper award. At present, he has published more than 60 papers in top conferences and journals in the fields of computer vision and machine learning.
Four selected Chinese scholars
Gu Quanquan has 4 selected papers this time, while he has 6 selected papers in neurips 2019.
Gu Quanquan is currently an assistant professor of computer science at UCLA and head of the statistical machine learning laboratory. He studied at Tsinghua University, received his Ph.D. in computer science from the University of Illinois at Urbana Champaign, and studied under Professor Han Jiawei, a leader in the field of data mining. He successively served as a postdoctoral fellow at Princeton University and an assistant professor at the University of Virginia. His research direction is statistical machine learning, focusing on the development and analysis of non convex optimization algorithms for machine learning. He has been awarded the "achievement award for outstanding young scholars" by the National Science Foundation of the United States.
Bo Li is an assistant professor of computer science at the University of Illinois at Urbana Champaign. She received her Ph.D. from Vanderbilt University in 2016 and was a graduate scholar of Symantec research laboratory. Her research interests are antagonistic deep learning, security, privacy and game theory. She developed and analyzed a scalable and robust learning framework for learning algorithms in an anti circumvention environment.
Zhou Mingyuan, assistant professor of statistics at the University of Texas at Austin, is also a core faculty member of the Department of statistics and data science at the school of natural sciences. He graduated from Nanjing University with a master's degree from the Chinese Academy of Sciences and a doctor's degree from Duke University. His research fields mainly include Bayesian statistics and machine learning.
Zhang Yang Wang is an assistant professor of computer science and engineering at Texas A & M University. He received his Ph.D. in electrical and computer engineering from the University of Illinois at Urbana Champaign. His research interests include machine learning, deep learning, computer vision, image and video processing optimization, etc.
Jiatao Gu is a research scientist at the Facebook artificial intelligence research center. Graduated from Tsinghua University, he is interested in the application of deep learning methods to natural language processing (NLP).
According to the statistics of aminer, ICLR 2020 receives 185 papers from Chinese scholars, 123 from academia and 62 from industry. In the high score papers, domestic enterprises such as HUAWEI, byte beating, Tencent and fast hand, domestic universities such as Tsinghua University, Nanjing University, Kwai Tai University, Xi'an Electronic and Science University and so on. We have combed out full score papers for you, and hope to see more and more Chinese teams get full score papers in the future! Huawei (oral paper) causal discovery with reinforcement learning Author: Shengyu Zhu, ignavier ng, Zhitang Chen organization: Huawei, University of Toronto key words: causal discovery; structural learning; reinforcement learning; directed acyclic graph link:
HTTPS: / / openreview. Net / PDF? Id = s1g2skstpb the problem to be solved in this paper is: given the neighborhood aggregation (first-order neighborhood aggregation) function of directed acyclic graph (DAG), according to the observed data set, through reinforcement learning, reverse search the original DAG structure of the graph.
Byte jumping, Nanjing University (oral paper) mirror general neural machine translation Author: Zaixiang Zheng, Hao Zhou, Shujian Huang and other institutions: Nanjing University, byte jumping Keywords: neural machine translation; generating network links:
HTTPS: / / openreview. Net / PDF? Id = hkxqrtnyph the researchers proposed a mirror generated machine translation model (mgnmt), which can better use non parallel corpus to improve the effect of NMT. Mg-nmt uses a generative approach to optimize translators and language models in both directions, so that it can be improved from two perspectives. Experiments show the effectiveness of this method.
Tsinghua University, sparse coding with gate learned ISTA Author: Kailun Wu, Yiwen Guo, Zhang Li, Changshui Zhang organization: Tsinghua University, byte jumping Keywords: sparse coding; deep learning; convergence analysis link: https://openreview.net/pdf? Id = bygpo2vkph Peking University a theory of usable information under computational Constraints Author: Yun Xu, Shengjia Zhao, Jiang song, Russell Stewart, Stefano ermon institutions: Peking University, Stanford University and other links:
Https://openreview.net/pdf? Id=r1eBeyHFDH, in addition, Kwai's collaboration with Tencent was also selected for Oral. Kwai Fu, Tencent, Watch the Unobserved: A Simple Approach to Parallelizing Monte Carlo Tree Tree Author:: Unobserved:, Simple, Simple, Simple, and Simple:
Reviewed and edited by: