1新智元报道
作者:新智元编辑部
【新智元导读】我们统计了人工智能领域与智能最核心的四大顶会 AAAI、IJCAI、ICML 和 NIPS,以及作为视觉和语言代表的 CVPR 和 ACL 这 6 大学术会议在过去十年的最佳论文。我们重点关注了这些论文的第一作者及其发表时所属机构。过去十年,在 AAAI、ICLM、IJCAI 和 NIPS 这四大最最最核心的智能会议的最佳论文中,只有一篇的第一作者是中国人,来自北大。
为了一睹各大机构在AAAI、IJCAI、ICML和NIPS 人工智能四大顶级会议大会上获得最佳论文的情况。新智元手动对这些会议2007到2016年的最佳论文(Best Paper , Outstanding Paper )归属进行了统计和排名。其中,NIPS因为缺乏数据,只统计了2010年到2016年的数据。另外,2010年2012年,NIPS的最佳论文为最佳学生论文。特此说明。
话不多说,先看最终结果:
具体看:
下面,我们来具体看看每个会议的情况。
AAAI
2017
Label-Free Supervision of Neural Networks with Physics and Domain Knowledge
用物理和特定领域知识让神经网络进行无标签的监督学习
Russell Stewart and Stefano Ermon, Stanford University
2016
Bidirectional Search That Is Guaranteed to Meet in the Middle
双向搜索保证前向和后向搜索在“中间相遇”
Robert C. Holte, University of Alberta; et al.
2015
From Non-Negative to General Operator Cost Partitioning
从非负到一般算子的成本分配
Florian Pommerening, University of Basel; et al.
2014
Recovering from Selection Bias in Causal and Statistical Inference
因果和统计推理中的选择性偏差恢复
Elias Bareinboim, University of California Los Angeles; et al.
2013
HC-Search: Learning Heuristics and Cost Functions for Structured Prediction
HC搜索:为结构化推断学习启发式算法和成本函数
Janardhan Rao Doppa, Oregon State University; et al.
SMILe: Shuffled Multiple-Instance Learning
SMILe:打乱的多事例学习
Gary Doran & Soumya Ray, Case Western Reserve University
2012
Learning SVM Classifiers with Indefinite Kernels
通过不定核学习 SVM 分类器
Suicheng Gu & Yuhong Guo, Temple University
Document Summarization Based on Data Reconstruction
基于数据重构的文档摘要
陈纯, 卜佳俊, 何占盈, 王灿, 李平,蔡登,何晓飞,浙江大学
2011
Dynamic Resource Allocation in Conservation Planning
保护规划中的动态资源分配
Daniel Golovin, California Institute of Technology; et al.
Complexity of and Algorithms for Borda Manipulation
波达操纵的复杂性和算法
Jessica Davies, University of Toronto; et al.
2010
How Incomplete Is Your Semantic Web Reasoner? Systematic Analysis of the Completeness of Query Answer
你的语义网推理器有多不完备?对问答系统完备性的系统分析
Giorgos Stoilos, Oxford University; et al.
A Novel Transition Based Encoding Scheme for Planning as Satisfiability
一个基于转换的新的编码方案,用于可满足性问题的计划
Ruoyun Huang, Washington University in St. Louis; et al.
2008
How Good is Almost Perfect?
多好算得上是几乎完美?
Malte Helmert & Gabriele Röger, Albert-Ludwigs-Universität Freiburg
Optimal False-Name-Proof Voting Rules with Costly Voting
多好算得上是几乎完美?
Liad Wagman & Vincent Conitzer, Duke University
2007
PLOW: A Collaborative Task Learning Agent
PLOW:一个协作任务智能体
James Allen, Institute for Human and Machine Cognition; et al.
Thresholded Rewards: Acting Optimally in Timed, Zero-Sum Games
有阈值的奖励:在计时零和游戏中表现最佳
Colin McMillen & Manuela Veloso, Carnegie Mellon University
ICML
2016
Ensuring Rapid Mixing and Low Bias for Asynchronous Gibbs Sampling
确保异步吉布斯采样的快速混合和低偏差
Christopher De Sa, Stanford University; et al.
Pixel Recurrent Neural Networks
像素循环神经网络
Aaron Van den Oord, Google; et al.
Dueling Network Architectures for Deep Reinforcement Learning
深度强化学习的竞争网络架构
Ziyu Wang, Google; et al.
2015
A Nearly-Linear Time Framework for Graph-Structured Sparsity
图结构稀疏性的近似线性时间框架
Chinmay Hegde, Massachusetts Institute of Technology; et al.
Optimal and Adaptive Algorithms for Online Boosting
在线boosting的最优化和自适应算法
Alina Beygelzimer, Yahoo! Research; et al.
2014
Understanding the Limiting Factors of Topic Modeling via Posterior Contraction Analysis
通过后验收缩分析理解主题建模的限制因素
Jian Tang, Peking University; et al.
2013
Vanishing Component Analysis
消失成分分析
Roi Livni, The Hebrew University of Jerusalum; et al.
Fast Semidifferential-based Submodular Function Optimization
基于半微分的快速子模块函数优化
Rishabh Iyer, University of Washington; et al.
2012
Bayesian Posterior Sampling via Stochastic Gradient Fisher Scoring
通过随机梯度Fisher得分进行贝叶斯后验采样
Sungjin Ahn, University of California Irvine; et al.
2011
Computational Rationalization: The Inverse Equilibrium Problem
计算合理化:逆向平衡问题
Kevin Waugh, Carnegie Mellon University; et al.
2010
Hilbert Space Embeddings of Hidden Markov Models
隐马尔科夫模型的希尔伯特空间嵌入
Le Song, Carnegie Mellon University; et al.
2009
Structure preserving embedding
结构保持嵌入
Blake Shaw & Tony Jebara, Columbia University
2008
SVM Optimization: Inverse Dependence on Training Set Size
SVM优化:训练数据集大小的逆向依赖
Shai Shalev-Shwartz & Nathan Srebro, Toyota Technological Institute at Chicago
2007
Information-theoretic metric learning
信息理论度量学习
Jason V. Davis, University of Texas at Austin; et al.
IJCAI
2016年
论文:Hierarchical Finite State Controllers for Generalized Planning
用于泛化规划的层次有限状态控制器
作者:Javier Segovia, Universitat Pompeu Fabra 等
2015 年
论文:Bayesian Active Learning for Posterior Estimation
用于后验评估的贝叶斯主动学习
作者:Kirthevasan Kandasamy, Carnegie Mellon University 等
论文:Recursive Decomposition for Nonconvex Optimization
非凸优化问题的递推分解
作者:Abram L. Friesen & Pedro Domingos, University of Washington
2013 年
论文:Bayesian Optimization in High Dimensions via Random Embeddings
经由随机嵌入的高维度贝叶斯优化
Ziyu Wang, University of British Columbia 等.
论文:Flexibility and Decoupling in the Simple Temporal Problem
简单时间问题的灵活性和去耦
作者:Michel Wilson, Delft University of Technology 等.
2011 年
论文:Unweighted Coalitional Manipulation Under the Borda Rule is NP-Hard
博尔达计数下的无加权联合操纵是 NP 难题
作者:Nadja Betzler, Technische Universität Berlin 等.
论文:Nested Rollout Policy Adaptation for Monte Carlo Tree Search
对于蒙特卡洛树搜索的嵌套部署策略的适应
作者:Christopher D. Rosin, Parity Computing
论文:On the Decidability of Connectedness Constraints in 2D and 3D Euclidean Spaces
2D 与 3D 欧几里得空间中连接约束的判别
作者:Roman Kontchakov, Birkbeck College London 等.
2009年
论文:Consequence-Driven Reasoning for Horn SHIQ Ontologies
Horn SHIQ 本体论的结果驱动推导
作者:Yevgeny Kazakov, University of Oxford
论文:Learning Conditional Preference Networks with Queries
通过查询学习条件偏爱网络
作者:Frederic Koriche, Université Montpellier II
Bruno Zanuttini, Université de Caen Basse-Normandie
2007年
论文:Automated Heart Wall Motion Abnormality Detection From Ultrasound Images using Bayesian Networks
使用贝叶斯网络的超声图像,进行心脏壁异常活动自动检测
作者:Maleeha Qazi, Siemens Medical Solutions 等.
论文:Building Structure into Local Search for SAT
用于 SAT 本地搜索的架构搭建
作者:Duc Nghia Pham, Griffith University 等.
论文:Performance Analysis of Online Anticipatory Algorithms for Large Multistage Stochastic Integer Pro...
对大型、多级式随机整数程序的在线预期算法的性能分析
作者:Luc Mercier & Pascal Van Hentenryck, Brown University
NIPS
2015
Competitive Distribution Estimation: Why is Good-Turing Good
竞争分布估计:为什么 Good-Turing 好?
Alon Orlitsky · Ananda Suresh
Fast Convergence of Regularized Learning in Games
游戏中正则学习的快速收敛
Vasilis Syrgkanis · Alekh Agarwal · Haipeng Luo · Robert Schapire
2014 Outstanding Paper Award
Asymmetric LSH (ALSH) for Sublinear Time Maximum Inner Product Search (MIPS)
次线性时间中最大内积搜索(MIPS)的非对称LSH(ALSH)
Anshumali Shrivastava (Cornell University)
Ping Li (Rutgers University)
2013 Outstanding Paper Awards
A memory frontier for complex synapses
复杂突触的记忆边界
Subhaneil Lahiri (Stanford University)
Surya Ganguli (Stanford University)
Submodular Optimization with Submodular Cover and Submodular Knapsack Constraints
子模块覆盖和子模块背包约束的子模型优化
Rishabh Iyer (University of Washington, Seattle)
Jeff Bilmes (University of Washington, Seattle)
Scalable Influence Estimation in Continuous-Time Diffusion Networks
连续时间扩散网络中的可扩展影响估计
Nan Du (Georgia Tech)
Le Song (Georgia Tech)
Manuel Gomez-Rodriguez (MPI for Intelligent Systems)
Hongyuan Zha (Georgia Tech)
NIPS Classic Paper Award
Algorithms for non-negative matrix factorization (2000)
非负矩阵分解算法
Daniel D. Lee
H. Sebastian Seung
2012Outstanding Student Paper Awards
No voodoo here! Learning discrete graphical models via inverse covariance estimation
通过逆协方差估计学习离散图模型
Po-Ling Loh and Martin Wainwright
Discriminative Learning of Sum-Product Networks
和积网络的判别学习
Robert Gens and Pedro Domingos
2011 Outstanding Student Paper Awards
Efficient Inference in Fully Connected CRFs with Gaussian Edge Potentials
具有高斯边缘电位的完全连接CRF中的有效推理
Philipp Krähenbühl and Vladlen Koltun
Priors Over Continuous Time Processes
连续时间过程的优先级
Ardavan Saeedi and Alexandre Bouchard-Côte
Fast and Accurate K-means for Large Datasets
大型数据集的快速准确K-means
Michael Shindler, Alex Wong, and Adam Meyerson
2010 Outstanding Student Paper Awards:
Construction of dependent dirichlet Processes based on Poisson Processes
基于泊松过程的DDP构建
Dahua Lin,Eric Grimson and John Fisher
A Theory of Multiclass Boosting
一种 Multiclass Boosting 理论
Indraneel Mukherje, Robert E Schapire
除了上述四大会议,要谈人工智能也不能不看视觉和语言,这两个领域对应的我们选择了 CVPR 和 ACL 做统计。可以发现,近年来,华人在 CVPR 和 ACL 的表现有上升的趋势。
ACL(自然语言处理相关)
计算语言学协会(The Association for Computational Linguistics,ACL)是涉及自然语言和计算问题的国际专业学会,会员遍布世界各地。协会成立于 1962 年,最初名为机器翻译和计算语言学协会(AMTCL),1968 年成为 ACL。
ACL 主席由全体会员投票选举产生。主席除了要在 NLP 领域有突出成果和贡献,还需有卓越的领导才干。ACL 现任主席是百度副总裁、AI 技术平台体系(AIG)总负责人王海峰,他也是该学术组织 50 年历史上第一位华人主席。
微软亚洲研究院副院长周明博士是 ACL 候任主席。
ACL 的年度会议汇集了领域当年优秀成果,我们的 ACL 最佳论文从 2007 年到 2016 年。不过,2017 年的 ACL 最佳论文已出,也一并列在下面。
值得一提,清华大学智能技术与系统国家重点实验室博士生布凡(李航博士、朱小燕教授)的论文“String Re-writing Kernel”被评为 ACL 2012 大会最佳学生论文。
时间 | 论文 | 作者及机构 |
---|---|---|
2017 | 概率类型学:元音库的深度生成模型 Probabilistic Typology: Deep Generative Models of Vowel Inventories |
Ryan Cotterell & Jason Eisner, Johns Hopkins University |
从计算建模角度看韵律和语域在词语切分中的作用 The Role of Prosody and Speech Register in Word Segmentation: A Computational Modelling Perspective |
Bogdan Ludusan, EHESS/ENS/PSL/CNRS; et al. | |
2016 | 使用字符串度量学习找到非任意形式意义的系统性内核回归 Finding Non-Arbitrary Form-Meaning Systematicity Using String-Metric Learning for Kernel Regression |
E. Darío Gutiérrez, University of California Berkeley; et al. |
2015 | 改进机器翻译质量估算的评测方法 Improving Evaluation of Machine Translation Quality Estimation |
Yvette Graham, Trinity College Dublin |
学习动态特征选择用于快速序列预测 Learning Dynamic Feature Selection for Fast Sequential Prediction |
Emma Strubell, University of Massachusetts Amherst; et al. | |
2014 | 用于统计机器翻译的快速、稳健神经网络联合模型 Fast and Robust Neural Network Joint Models for Statistical Machine Translation |
Jacob Devlin, Raytheon BBN Technologies; et al. |
2013 | 从带有语句描述的视频中学习语言表征 Grounded Language Learning from Video Described with Sentences |
Haonan Yu & Jeffrey Mark Siskind, Purdue University |
2012 | 重写字符串的内核 String Re-writing Kernel |
布凡,李航,朱小燕,清华大学 |
Bayesian Symbol-Refined Tree Substitution Grammars for Syntactic Parsing |
Hiroyuki Shindo, NTT Communication Science Laboratories; et al. | |
2011 | 使用基于图的双语预测对词性进行无监督标 Unsupervised Part-of-Speech Tagging with Bilingual Graph-Based Projections记 |
Dipanjan Das, Carnegie Mellon University Slav Petrov, 谷歌 |
2010 | Beyond NomBank: A Study of Implicit Arguments for Nominal Predicates Matthew |
Gerber & Joyce Y. Chai, Michigan State University |
2009 | 将指令转化为行动的强化学 习Reinforcement Learning for Mapping Instructions to Actions |
S.R.K. Branavan, Massachusetts Institute of Technology; et al. |
K-Best A* Parsing | Adam Pauls & Dan Klein, University of California Berkeley | |
Concise Integer Linear Programming Formulations for Dependency Parsing |
André F.T. Martins, Instituto de Telecomunicações; et al. | |
2008 | 森林重新排名:分析非局部特征 Forest Reranking: Discriminative Parsing with Non-Local Features |
Liang Huang, University of Pennsylvania |
具有目标依赖语言模型的全新字符串到依赖机器翻译算法 A New String-to-Dependency Machine Translation Algorithm with a Target Dependency Language Model |
Libin Shen, BBN Technologies; et al. | |
2007 | 学习用 lambda 演算语义解析的同步语 法Learning synchronous grammars for semantic parsing with lambda calculus |
Yuk Wah Wong & Raymond J. Mooney, University of Texas at Austin |
CVPR(计算机视觉)
计算机视觉与模式识别会议(Internaltional Conference on Computer Vision and Pattern Recogintion,CVPR)是计算机视觉领域三大顶会之一,每年举办一次,主要内容正如名字表明的那样,是计算机视觉与模式识别技术。
新智元在对 CVPR 2017 进行会议导览时统计发现,今年 CVPR 的全部 81 位主席中,华人主席有 31 位。根据上月公布的最新消息,到 2021 年,CVPR 主办团队中有6位华人主席。
相比 ICCV 和 ECCV,近年来华人力量在 CVPR 逐渐上升,也使我们的统计别有一番价值。
同样,CVPR 2017 年最佳论文已出,也放在表格中,但不进入本文的统计。
时间 | 论文 | 作者及机构 |
---|---|---|
2017 | Densely Connected Convolutional Networks | 黄高,康奈尔大学; 刘壮,清华大学;ti al. |
通过对抗训练从合成和无监督图像中学习Learning from Simulated and Unsupervised Images through Adversarial Training | Ashish Shrivastava, Apple; et al. | |
2016 | 用于图像识别的深度残差网络 Deep Residual Learning for Image Recognition |
何恺明、张祥雨、任少卿、孙剑,微软研究院 |
2015 | 动态融合:实时非刚性场景的重建与跟踪 DynamicFusion: Reconstruction and Tracking of Non-rigid Scenes in Real-Time |
Richard A. Newcombe, University of Washington; et al. |
2014 | What Camera Motion Reveals About Shape with Unknown BRDF |
Manmohan Chandraker, NEC Labs America |
2013 | 在单机上对 100,000 个目标类别进行快速准确的检测 Fast, Accurate Detection of 100,000 Object Classes on a Single Machine |
Thomas Dean, Google; et al. |
2012 | 对动作生成结构进行非刚性因式分解的简单且无需先验的方法 A Simple Prior-free Method for Non-Rigid Structure-from-Motion Factorization |
Yuchao Dai, Northwestern Polytechnical University; et al. |
2011 | 使用单层图像等信息进行实时人体姿态识别 Real-time Human Pose Recognition in Parts from Single Depth Images |
Anders Eriksson & Anton va den Hendel, University of Adelaide |
2010 | 在数据残缺的情况下使用 L1 norm 对低秩矩阵进行稳定估算的高效计算方法 Efficient computation of robust low-rank matrix approximations in the presence of missing data using the L1 norm |
Jamie Shotton, Microsoft Research; et al. |
2009 | 使用暗通道先验单图像去雾 Single Image Haze Removal Using Dark Channel Prior |
何恺明,香港中文大学; 孙剑,微软亚洲研究院; 汤晓鸥 香港中文大学 |
2008 | 二阶平滑先验下的全局立体重建 Global Stereo Reconstruction under Second Order Smoothness Priors |
Oliver Woodford, University of Oxford; et al. |
使用高效子窗口搜索进行目标定位 Beyond Sliding Windows: Object Localization by Efficient Subwindow Search |
Chistoph H. Lampert, Max Planck Institut; et al. | |
2007 | 在移动车辆中进行动态三维场景分析 Dynamic 3D Scene Analysis from a Moving Vehicle |
Bastian Leibe, ETH Zurich; et al. |