郝建业老师华为诺亚方舟实验室2:00 PM - 3:00 PM, July 13, 2023

报告摘要：本报告首先会介绍深度强化学习背景与基础，然后从如何学的好、学的快、学的稳三方面介绍深度强化学习所面临的挑战及相应的解决方案，同时介绍深度强化学习在自动驾驶、5G网络优化、供应链物流优化、EDA芯片设计等场景的应用。

讲者简介：郝建业博士，天津大学智算学部副教授，华为诺亚方舟决策推理实验室主任。主要研究方向为深度强化学习、多智能体系统。发表人工智能领域国际会议和期刊论文100余篇，专著2部。 主持国家科技部2030重大项目课题、国防科技创新重点项目课题及企业横向课题等项目10余项，研究成果荣获国际会议最佳论文奖3次，NeurIPS20-22大会竞赛冠军4次。相关成果在工业基础软件智能化、自动驾驶、游戏AI、广告及推荐、5G优化、物流调度等领域落地应用。

范长俊副教授国防科技大学7:30 PM - 8:30 PM, July 12, 2023

报告摘要：现实世界中大量问题的解决需要设计和求解算法。传统上通常是人类学习如何设计算法，而随着人工智能技术的发展，我们也可以通过机器学习方法去学习如何设计算法，如组合优化问题求解算法，这类方法在求解效率、求解灵活性等方面具有强大优势。本次报告主讲人将介绍机器学习求解组合优化的问题，特别是图上的组合优化问题，包括最小节点覆盖、网络瓦解、自旋玻璃基态求解问题的最新进展，并进一步探讨该方向的其他一些有意思的研究点。

讲者简介：范长俊，国防科技大学副教授，硕士生导师，加州大学洛杉矶分校(UCLA)访问学者，中国指控学会信息融合专委会副总干事，湖湘青年英才。主持与参与10余项国家和军队级重大项目。以第一及通讯作者在Nature Machine Intelligence、Nature Communications及AAAI等顶级期刊和会议发表论文20余篇。研究成果多次应用于部队重大活动。先后获吴文俊人工智能科技进步一等奖、世界人工智能大会青年优秀论文奖、中国指挥与控制学会优博论文奖和国防科技大学青年创新一等奖。

范喆楠高级研究员华为加拿大研究所10:00 AM - 11:00 AM, Jun 21, 2023

报告摘要：Operations Research (OR) has long been a bastion of analytical decision-making, drawing from mathematics, statistics, and computer science. As digitization reshapes our world, Artificial Intelligence (AI) is becoming a critical ingredient in OR. This presentation, "Artificial Intelligence for Operations Research", seeks to shed light on AI's groundbreaking potential in transforming the established processes in OR. In this talk, our objective is to unveil the multiple ways in which AI can augment and enhance the steps in the OR process, from initial data gathering to the final stages of optimization and interpretation. A noteworthy aspect of this integration is the revolutionary role of large language models like GPT-4 in the OR process. These models, by translating complex problem descriptions into mathematical models, are creating paradigm shifts in the way we approach problem-solving in OR. The exploration continues as we investigate the power of AI to enhance model optimization. This part of the discussion brings to the forefront the application of advanced AI methodologies such as deep learning-based optimization algorithms and graph neural networks, which facilitate efficient navigation through intricate search spaces. By integrating AI into OR, we open up new possibilities for efficient and effective solutions to the world's most challenging problems.

讲者简介：Zhenan Fan serves as a Senior Research Scientist at Huawei Vancouver Research Center, having previously earned his PhD from the University of British Columbia and his Bachelor's degree from the University of Toronto. His research interests include mathematical optimization, learning-based optimization methods, and large language model applications. He has published several papers in esteemed conferences such as ICML, ICDE, and ICLR, and in prominent journals including Foundations and Trends in Optimization and IEEE Transactions on Signal Processing. Furthermore, he contributes to the scholarly community by serving as a reviewer in these venues.

陈磊高级研究员华为诺亚方舟实验室10:00 AM - 11:00 AM, Mar 24, 2023

报告摘要：在集成电路设计中，逻辑综合（logic synthesis）是所设计数字电路的高抽象级描述，经过布尔函数化简、优化后，转换到的逻辑门级别的电路连线网表的过程。逻辑综合是EDA设计流程前端的“根技术”，通常包括翻译、逻辑优化和工艺映射三个阶段。逻辑优化和工艺映射都是NP-hard的难题。更快更有效的逻辑综合方法是关于芯片性能（更小，更快，更节能）的重中之重。本次分享将简要介绍我们在逻辑综合底层新算子的研发工作，以及一些AI对现有综合算子的改进探索工作。

讲者简介：陈磊，华为诺亚方舟实验室研究员，计算机博士，毕业于香港浸会大学，研究领域包括EDA逻辑综合优化、时间序列预测、数据库等。先后负责逻辑综合、供应链需求预测等多个项目的研发和落地，当前主要专注于逻辑综合的关键算法研发。在KDD、SIGMOD、ICDE、VLDB、DAC、TKDE等多个顶会顶刊发表论文10余篇，申请专利10+，曾担任国际会议MDM2019、MDM2021工业Track主席。

许思源主任工程师华为诺亚方舟实验室11:00 AM - 12:00 AM, Mar 3, 2023

报告摘要：芯片布局起着承上（逻辑综合）启下（布线）的作用，是现代超大规模集成电路设计物理设计流程中的一步，显著影响芯片设计的迭代周期。芯片布局问题可以描述为：给定标准单元和宏单元的大小和连接关系，并给出约束条件(如没有元件重叠)和优化目标(时序、拥塞、功耗、面积等)，基于特定策略确定每个单元的摆放位置。由于芯片布局被认为是NP-Complete的问题，因此如何快速生成高质量的芯片布局成为芯片制造过程中的重要环节。本次报告将介绍一些基于深度学习的芯片布局建模的方案，主要包括以下三方面内容：（1）时序、拥塞、线长等指标预测的模型建模；（2）基于强化学习的宏单元布局；（3）基于图神经网络的单元布局。

讲者简介：许思源，2019年毕业于美国德克萨斯大学达拉斯分校，获计算机工程博士学位。毕业后就职于美国Mathworks公司任高级工程师，从事基于FPGAs深度学习的研发项目。现任华为诺亚方舟实验室AI4EDA项目组主任工程师，目前研究方向为芯片物理实现的建模与优化，仿真效率加速等。

伯晓晨研究员、孙昱博士军事科学院军事医学研究院10:00 AM - 11:00 AM, Feb 10, 2023

报告摘要：自上世纪60年代发现DNA双螺旋结构至今，基因组的谜团被不断揭示。近年来，伴随高通量测序技术的发展，细胞内的复杂关系诸如调控网络、DNA损伤修复、染色质空间互作等逐渐走入人们视野。作为生命过程的关键调控者，这些复杂关系同时也是重大疾病发生发展的幕后关键。因此，建模并解析基因组复杂关系是理解生命的现实亟需。本次报告将介绍我们近年来开发的基于深度学习建模基因组复杂关系的方法及应用，主要包括以下三方面内容：1）基于变分图自编码器的转录因子调控网络预测算法；2）基于图神经网络的DNA双链断裂预测方法；3）生物知识启发的可解释神经网络计算框架。

讲者简介：伯晓晨，毕业于国防科技大学，先后获得自动控制专业学士、模式识别与智能系统专业硕士、控制科学与工程专业博士学位，现任军事科学院军事医学研究院的研究员、博士生导师。目前研究方向为生物大数据智能处理与分析。作为负责人和主要研究者先后承担精准医疗国家重点研发专项、国家重大新药创制、自然基金重大研究计划、863计划等20余项，在Nature、Nature Microbiology、Molecular Cell、Nucleic Acids Research等SCI期刊发表论文100余篇，相关成果先后入选“2015年中国科技十大进展”，军队科技进步一等奖、国家科技进步二等奖、北京市科学技术奖一等奖等。入选国家科技部 “中青年科技领军人才”、中组部“万人计划”科技领军人才、获得中国科协“求是杰出青年奖”、获得国防科技卓越青年基金支持、入选军队学科拔尖人才，兼任军委科技委生物交叉科学技术领域首席科学家、中国自动化学会智能健康与生物信息专委会副主任委员、中国药理学会网络药理学专委会常委。
孙昱，本硕毕业于战略支援部队信息工程大学，获得信息工程专业学士、智能信息处理专业硕士学位，后于军事科学院获得生物医学工程专业博士学位，博士期间在Nature Machine Intelligence发表SCI论文1篇，Genome Biology二次修回论文1篇。现任军事科学院军事医学研究院助理研究员，参与撰写2022年《人工智能系列白皮书——人工智能与药物发现》的可解释部分章节，主要研究方向为基于深度学习的生命组学数据分析。

张青富教授香港城市大学10:00 AM - 11:00 AM, Dec 2, 2022

报告摘要：Over the last ten years, decomposition based multiobjective evolutionary algorithms (MOEA/D) have become a major methodology in the field of evolutionary computation. These algorithms uses decomposition techniques from traditional multiobjective optimization to decompose a multiobjective optimization problem into a set of subtasks and then solve them in a collaborative manner. In this talk, I will explain the basic ideas behind MOEA/D and some recent developments. I will also outline some possible research issues in multiobjective evolutionary computation.

讲者简介：张青富，香港城市大学计算机科学系计算智能讲座教授，IEEE Fellow. 他曾任英国Essex大学计算机与电子工程学院教授，西安电子科技大学长江讲座教授，他的主要研究领域包括计算智能，多目标优化与决策，启发式算法及其应用，他所提出的MOEA/D 算法是进化计算领域最常用的算法框架。

汪云海教授山东大学16:00 PM - 17:00 PM, Nov 15, 2022

报告摘要：图布局是图数据可视化分析中必不可少的一步，主要探索如何将关系数据通过节点连接图等形式高效准确地展示。近几十年里，学者们提出了大量的图布局方法（如弹簧-电荷模型、应力模型和最大熵模型等），但各种方法对应不同的物理模型，缺乏统一的计算框架，使得应用和定制它们非常困难。为解决该问题，我们提出了一个通用的布局计算框架Taurus，将大多数现有的布局方法表示为同一个数学表达，采用统一的增强随机梯度下降优化器进行求解，并提出了平衡应力模型(BSM)，进一步提升了图布局质量。在此基础上，我们进一步提出了基于t分布的力引导布局方法，修正了力引导布局方法在短程力的缺陷，使得生成高质量布局的同时比当前方法能快100倍。我们开源了该框架（https://taurusgl.github.io/)，可支持用户快速定制满足应用需求的图布局方法。

讲者简介：汪云海，山东大学计算机科学与技术学院教授，教育部青年长江学者，山东省杰青。主要从事数据可视化、图形学与人机交互领域的研究，在ACM TOG、IEEE TVCG、ACM SIGGAPH (Asia)、ACM CHI、IEEE VIS、IEEE VR、ICCV、NeurIPS等权威期刊和会议发表论 60余篇，获得ACM CHI 2021 最佳论文提名奖、IEEE VIS 2021最佳论文提名奖等奖项。曾获山东省自然科学一等奖、山东省教学成果奖一等奖等，担任Computer Graphics Forum、 Frontiers in Computer Science、CCCF专题等期刊编委。

印卧涛教授阿里达摩院10:00 AM - 11:00 AM, Oct 26, 2022

报告摘要：While mixed-integer linear programming (MILP) is NP-hard in general, practical MILP has received 100--fold speedup in the past twenty years. Still, many classes of MILPs quickly become unsolvable as their sizes increase, motivating researchers to seek new techniques to accelerate MILP solution. With deep learning, they have obtained strong empirical results, and many results were obtained by applying graph neural networks (GNNs) to making decisions in various stages of MILP solution processes. This work discovers a fundamental limitation: there exist feasible and infeasible MILPs that no GNNs can distinguish, indicating their lacking the power to express general MILPs. Then, we show two positive results: 1. GNNs don't have this limitation on linear programs (LPs); 2. by restricting the unfoldable MILPs or, alternatively, by adding random features, there exist GNNs that can reliably predict MILP feasibility, optimal objective values, and optimal solutions up to prescribed precision. We conducted small-scale numerical experiments to validate our theoretical findings. Co-authors: Ziang Chen (Duke Math), Jialin Liu (Alibaba DAMO), Xinshang Wang (Alibaba DAMO), Jianfeng Lu (Duke Math).

讲者简介：Dr. Wotao Yin obtained his PhD in OR from Columbia U. He is the director of the Decision Intelligence Lab of Alibaba DAMO Academy. Before that, he was a full professor in the Department of Mathematics at UCLA. His research interests include computational optimization and its applications in image processing, machine learning, and other data science problems. He invented fast algorithms for sparse optimization and large-scale distributed optimization. He is been among the top 1% highly-cited mathematicians since 2018. He received an NSF CAREER Award 2006, Sloan Research Award 2007, Morningside Gold Medal 2016, DAMO Award 2021, and Egon Balas Award 2021. He discovered the first method that can be used to decompose three monotonic operators without increasing the dimensionality of the variable, known by academics as Davis-Yin Splitting.

李希君博士华为诺亚实验室10:30 AM - 11:30 AM, Jul 22, 2022

报告摘要：本次报告首先总结了bengio等人提出的学习优化三大范式，并据此阐述了三个经典的机器学习解决组合优化问题的算法；分析了2020~2021年美国NSF对于学习优化布局投资；最后分享了诺亚、华为云在学习优化求解器上的一些工作，总结了学习优化的趋势与挑战。

讲者简介：Xijun is a senior researcher of HUAWEI Noah’s Ark Lab. Before that, he has received M.Sc. degree from Shanghai Jiao Tong University, P.R. China, in 2018. And now he is working towards his Ph.D. degree in the University of Science and Technology of China (HUAWEI-USTC Joint Ph.D. Program) under the supervision of Prof. Jie Wang. He has published 1o papers on top peer-reviewed conferences and journals (such as SIGMOD, KDD, ICDE, DAC, CIKM, ICDCS, TCYB, etc.) and 10+ patents (first author in 6 patents) with Noah's Ark Lab. He has also been committee member of MDM 2021 Industry Track and reviewer of multiple top journals and conferences (TKDE, AAAI, ICDE, CIKM, etc.). His research interests focus on Learning to Optimization (L2O) and Machine Learning for Computer System (ML4CS). And now he is deeply involved in the development of HUAWEI OptVerse Solver (both MIP and LP).

陈磊博士华为诺亚实验室10:00 AM - 11:00 AM, Jun 30, 2022

报告摘要：在集成电路设计中，逻辑综合（logic synthesis）是所设计数字电路的高抽象级描述，经过布尔函数化简、优化后，转换到的逻辑门级别的电路连线网表的过程。逻辑综合是EDA芯片设计中的承上启下的关键环节，通常包括翻译、逻辑优化和工艺映射三个阶段。逻辑优化和工艺映射都是NP-hard的难题。更快更有效的逻辑综合方法是关于芯片性能（更小，更快，更节能）的重中之重。这次分享将简要介绍下逻辑综合的问题背景、应用场景以及我们在逻辑综合问题上的一些探索工作。

讲者简介：陈磊博士，华为诺亚方舟实验室研究员。研究领域包括EDA逻辑综合优化、时间序列预测等，先后负责逻辑综合优化、供应链需求预测等多个项目的研发和落地。在KDD、SIGMOD、ICDE、VLDB、TKDE等多个的顶会顶刊发表论文10余篇，担任TKDE、ICDE等A类期刊、会议审稿人，曾担任MDM2019、MDM2021工业track chair。

刘同亮教授悉尼大学14:00 PM - 15:00 PM, May 25, 2022

报告摘要：Label noise is ubiquitous in the era of big data. Deep learning algorithms can easily overfit the noise and thus cannot generalize well without properly handling the noise. In this talk, we will introduce the typical approaches to deal with label noise, i.e., extracting confident examples (whose labels are likely to be correct) and modelling label noise. The former one helps get rid of the incorrect labels while the latter one helps build statistically consistent classifiers. We will illustrate the intuitions of the state-of-the-art methods. We hope that the participants will roughly know how to learn with noisy labels via the talk.

讲者简介：刘同亮教授现任澳大利亚悉尼大学人工智能中心主任。主要从事可信机器学习及其在交叉领域的研究工作，特别是标签含噪学习、对抗学习、迁移学习和统计深度学习理论。目前已经在ICML, NeurIPS, ICLR, CVPR, ICCV, ECCV, AAAI, IJCAI, KDD, IEEE T-PAMI, T-NNLS, T-IP等会议/期刊发表论文100余篇。担任包括ICML, NeurIPS, ICLR, UAI, AAAI, IJCAI，KDD在内的多个顶级会议的领域主席、 TMLR 和MLJ编辑。获得澳大利亚研究委员会的优秀青年科研奖、悉尼大学工学院青年教职杰出研究奖、澳大利亚工程和计算机科学领域早期成就者奖等。

李怡康博士商汤科技14:00 PM - 15:00 PM, May 19, 2022

报告摘要：随着自动驾驶技术的不断发展，测试的重要性不断凸显，无论是技术研发，还是未来自动驾驶落地所需的技术验证与检验，都离不开高效、高质量的测试环节。然而，目前自动驾驶测试依然存在效率低、成本高、测试危险性高、测试数据有限、缺乏统一标准等诸多问题。因此，在本次分享中，我们将从自动驾驶未来安全落地的需求出发，分别从商汤科技和上海人工智能实验室所做的平台建设和基础技术研发，介绍我们目前正在搭建的数据驱动的自动驾驶仿真测试体系，并着重介绍在高一致性仿真引擎研发方面的一些基础研究工作进展。

讲者简介：李怡康博士，商汤科技研发副总监，上海人工智能实验室青年科学家/PI，主要负责无人驾驶产品开发及前瞻技术研究工作；本科毕业于清华大学电子工程系，博士毕业于香港中文大学多媒体实验室，师从王晓刚教授和汤晓鸥教授，主要研究方向为深度学习，计算机视觉，场景理解等，在CVPR/ECCV/ICCV/NeurIPS等国际顶级学术会议发表多篇论文。曾获得国家奖学金、“微软学者”奖学金及香港政府博士生奖学金等荣誉；曾在微软亚洲研究院、Facebook Reality Lab、商汤研究院等研究机构从事深度学习及计算机视觉相关的研究。

秦涛博士MSRA15:00 PM - 16:00 PM, May 7, 2022

报告摘要：In this talk, I will introduce two projects related to game playing and testing. (1) Project Suphx: the World-Best Mahjong AI. Mahjong is a popular multi-player imperfect-information game worldwide but very challenging for AI research due to its complex playing/scoring rules and rich hidden information. We built an AI for Mahjong, named Suphx, based on deep reinforcement learning with some newly introduced techniques including global reward prediction, oracle guiding, and run-time policy adaptation. Suphx has demonstrated stronger performance than most top human players in terms of stable rank and is rated above 99.99% of all the officially ranked human players in the Tenhou platform. This is the first time that a computer program achieves 10 DAN and outperforms most top human players in Mahjong. (2) Project Mariana: Pixel based AI for Automated Game Testing. AI based game testing is important to reduce development cost and improve gamer experiences. Early attempts reply on game internal information for game space exploration and thus require deep integration with games, which is inconvenient for practical applications.  We propose to use only screenshots/pixels as input for automated game testing and build a general game testing agent, Inspector, that can be easily applied to different games without deep integration with games. In addition to covering all game space for testing, our agent also tries to take human-like behaviors to interact with key objects in a game, since some bugs usually happen in player-object interactions. Inspector is based on purely pixel inputs and comprises three key modules: game space explorer, key object detector, and human-like object investigator. Game space explorer aims to explore the whole game space by using a curiosity-based reward function with pixel inputs. Key object detector aims to detect key objects in a game, based on a small number of labeled screenshots. Human-like object investigator aims to mimic human behaviors for investigating key objects via imitation learning. We conduct experiments on two popular video games: Shooter Game and Action RPG Game. Experiment results demonstrate the effectiveness of Inspector in exploring game space, detecting key objects, and investigating objects. Moreover, Inspector successfully discovers two potential bugs in those two games.

讲者简介：Dr. Tao Qin (秦涛) is a Senior Principal Researcher and managing the Deep and Reinforcement Learning group at Microsoft Research Asia. His research interests include deep learning (with applications to machine translation, healthcare, speech synthesis and recognition, music understanding and composition), reinforcement learning (with applications to games and real-world problems), game theory and multi-agent systems (with applications to cloud computing, online and mobile advertising), and information retrieval and computational advertising. He got his PhD degree and Bachelor degree both from Tsinghua University. He is a senior member of ACM and IEEE, and an Adjunct Professor (PhD advisor) in the University of Science and Technology of China. His team helped Microsoft achieve human parity in Chinese-English machine translation in 2018, won the first place for 8 translation tasks in WMT 2019, and built the world-best Mahjong AI, named Suphx, which achieved 10 DAN on the Tenhou platform in mid 2019.

邵斌博士、王童博士MSRA14:00 PM - 15:30 PM, Apr 21, 2022

报告摘要：COVID-19 that has been caused by SARS-CoV-2 has become a global concern and caused severe damage all over the world. In this talk, we will introduce our studies on the infectious mechanism of the Spike protein of SARS-CoV-2. By molecular dynamics simulations, we proposed a “wedge” model of N-terminal domain of Spike protein for the key role in regulating the binding affinity of SARS-CoV-2 and explained the high infectious ability of Omicron variant from a dynamic view. The second part of this talk is on drug design. We proposed Intermolecular Graph Transformer (IGT) to make accurate drug-target interaction prediction. IGT outperforms state-of-the-art algorithms and exhibits the best generalization ability. High throughput virtual screening by IGT against SARS-CoV-2 has been made and several promising drug candidates are selected.

讲者简介：
Dr. Bin Shao is a senior principal research manager at Microsoft Research Asia (Beijing, China), leading the Computational Biology Group. He joined Microsoft after receiving his Ph.D. degree from Fudan University in July 2010. His research interests include computational biology, computational chemistry, molecular dynamics, machine learning, and parallel graph processing. The results of his research have appeared in top conferences and leading journals. Microsoft Graph Engine, which was built by Bin's team, is serving many Microsoft products and services, such as Microsoft Satori knowledge graph, Bing search, MSN, Xbox, and Cognitive Services.
Dr. Tong Wang is the Senior Researcher at Microsoft Research Asia. He received his Ph.D in Tsinghua University and held a visiting PhD experience at Harvard University. His research focuses on the algorithm design and applications on molecular dynamics, computer-aided drug discovery and protein structure prediction. Dr. Wang published a series of peer-reviewed papers on Nature Machine Intelligence, Bioinformatics, Advanced Theory and Simulations and BMC Bioinformatics and holds several China patents. He is also the reviewers of Nature Communications, Bioinformatics, RECOMB and so on.

秦志伟博士Lyft10:30 AM - 11:30 AM, Apr 15, 2022

报告摘要：随着智能手机的日益普及和功能强大，共享出行平台已经成为人们实现实时、个性化出行的可靠选择。这些平台通过连接司机和乘客，使空闲的车座被有效利用，来满足不断增多的实时出行需求。此次报告会以主题概览形式，来系统得讲解共享出行中的核心决策问题，和对领域内关于使用强化学习解决这些问题的工作进行综述性介绍。报告也会对这些议题上的挑战和机会做一个总结。

讲者简介：秦志伟（Tony）博士现任Lyft网约车实验室Principal Scientist，致力于网约车交易市场中核心策略优化的研究。他从美国哥伦比亚大学获得运筹学博土，曾在滴滴Al Labs带领决策智能团队和在沃尔玛全球电子商务任职研究科学家，近几年主要聚焦在强化学习及其在运筹优化，智能交通，在线营销上的应用。他在机器学习，数据挖掘和运筹优化的顶级会议和期刊发表近30篇论文，并是多个会议和期刊（NeurIPS, ICML, KDD, AAAl, TR-C, Transportation Science等）的评审专家。他带领团队获得了INFORMS 2019年Daniel H. Wagner Prize（运筹学杰出应用奖），并入选 NeurIPS 2018 Best Demo Awards。Tony拥有10多项美国专利，涵盖智慧交通，供应链管理，和推荐系统。

罗世凯博士字节跳动10:30 AM - 11:30 AM, Mar 26, 2022

报告摘要：a new product with an old one in pharmaceutical, technological, and traditional industries. Major challenges arise in online experiments of two-sided marketplace platforms (e.g., Uber) where there is only one unit that receives a sequence of treatments over time, i.e. switchback design. In those experiments, the treatment at a given time impacts current outcome as well as future outcomes. We propose a novel potential outcome framework based on a temporal varying coefficient decision process model to capture the dynamic treatment effects in temporal dependent experiments. We further characterize the average treatment effect by decomposing it as the sum of direct effect (DE) and indirect effect (IE). We develop estimation and inference procedures for both DE and IE. We also introduce a reinforcement learning framework for carrying A/B testing in these experiments, while characterizing the long-term treatment effects. We conduct extensive simulations to investigate the finite sample performance of the proposed estimation and inference procedures. We examine how our models can help improve policy evaluation for various dispatching and dispositioning policies in DiDi.

讲者简介：罗世凯博士在2011年本科毕业于南开大学数学学院。2016年博士毕业于美国北卡州立大学，主要研究因果推断和强化学习。论文发表在Journal of American Statistical Association, Journal of Machine Learning Research, Electronic Journal of Statistics, Statistical Methodology, NeuroImage, CIKM等。先后就业于Quantlab Financial，滴滴，腾讯，字节跳动。2016-2018在Quantlab Financial利用深度强化学习从事高中低频交易策略开发。2018-2020年主要利用因果推断和强化学习做核心运营策略的优化和评估。2021年加入字节，主要负责UG的基础算法。

唐睿明博士华为诺亚实验室 10:00 AM - 11:00 AM, Mar 18, 2022

报告摘要：推荐系统在人们的日常生活中随处可见，是不可或缺的一部分。深度学习，在大数据时代，做为主流的机器学习模型之一，在图像和语音识别领域取得了突破性的进展。将深度学习技术应用于推荐系统，是一个非常有趣，且值得研究的课题。在这次演讲中，主讲人将向大家介绍诺亚推荐与搜索实验室三个重要方向的研究工作：精准、自适应和收益最大化。

讲者简介：唐睿明是华为诺亚推荐与搜索实验室主任。他于2009年在中国东北大学获得学士学位，并在2014年从新加坡国立大学计算机专业获得博士学位。2014年底，他加入华为诺亚方舟实验室。他的研究方向包括机器学习、推荐系统、深度学习、强化学习、AutoML、图神经网络等。在推荐系统领域，他的多篇论文发表于国际顶级会议和期刊，如TOIS，TKDE，WWW，IJCAI，AAAI，RecSys，SIGIR，KDD，CIKM，WSDM等。他还担任一些顶级国际会议和期刊的审稿人工作，如TKDE，IPM，SIGIR，KDD，CIKM，WSDM，AAAI等。他的主要成果之一DeepFM发表于IJCAI 2017，至今获得1200+引用量。

周敏博士华为诺亚实验室 16:30 PM - 17:30 PM, Feb 25, 2022

报告摘要：图数据、序列数据广泛存在我们的生活中。我们将以网络拓扑补全为背景，介绍我们基于双曲空间的图表征学习上的一些探索和成果以及讨论未来的研究方向。

讲者简介：周敏，现任华为诺亚方舟实验室研究员。中国科学技术大学自动化系本科，新加坡国立大学工业系统工程与管理系博士。2017年加入华为诺亚方舟实验室ICT智能化研究团队，从事智能运维相关算法研究与应用，主要研究方向为序列数据预测分析，图数据挖掘与分析，研究和应用成果申请专利多项，并在KDD, ICDE, Automatic, QREI等会议和期刊发表论文多篇。

董振华博士华为诺亚实验室 15:30 PM - 16:30 PM, Feb 25, 2022

报告摘要：本次分享首先回顾推荐系统30年的发展历史，着重讲解典型的推荐模型和系统架构，然后通过对当前推荐中偏置问题的介绍，引出因果启发的推荐模型，包括因果干预、反事实学习等技术，重点介绍产品上实用、学术上领先的因果纠偏方法，并在讲解的过程中普及因果推断的基本概念和方法。因果启发的推荐系统是新兴研究领域，值得业界同仁的关注，最后讨论未来值得探究的研究方向。

讲者简介：董振华，现任华为诺亚方舟实验室技术专家，本科毕业于天津大学，博士毕业于南开大学，明尼苏达大学 GroupLens Lab 访问学者。研究方向为推荐系统、信息检索、因果推理、反事实学习。现任华为诺亚方舟实验室技术专家，负责推荐系统、机器学习前沿技术的研究与落地，负责华为多个产品推荐系统的构建，包括：华为浏览器、华为应用市场、广告系统、负一屏信息与服务直达、金融产品推荐系统等。申请专利 20余项，在 TKDE、SIGIR、RecSys、WWW、AAAI、CIKM 等期刊、会议发表学术论文近40篇，担任 SIGKDD、WSDM、WWW、AAAI、WebConference、TOIS等学术会议和期刊的程序委员，《奇点临近》译者。

袁明轩博士华为诺亚实验室 14:30 PM - 15:30 PM, Feb 25, 2022

报告摘要：优化算法在各种实际场景与产品中有着广泛的应用。近些年，如何将机器学习算法与优化算法结合成为研究热点。这次分享将简要介绍我们在一些工业问题上的探索工作，并通过一个EDA场景工作的详细介绍，给出我们关于学习优化技术研究的一些初步观点。

讲者简介：袁明轩，华为诺亚方舟实验室高级研究员。研究领域包括调度与优化算法、时空数据分析等。负责诺亚方舟实验室-企业智慧团队和AI4EDA团队工作，带领团队完成了供应链排产引擎、数字化装箱、装柜与路径优化等系列模型、ATPG-SAT算法、逻辑综合子图并行/算子优化和电信大数据时空数据分析引擎、用户离网预测与维系挽留等模型的研发与上线工作。担任TKDE、KDD等A类期刊与会议审稿人，完成专利30+，发表论文70余篇。

夏应策博士微软亚洲研究院 10:00 AM - 11:00 AM, Feb 20, 2022

报告摘要：Drug discovery is an important task with big social and business impact. It usually takes tens of years and billions of dollars to discover a new drug. Therefore, using artificial intelligence to speed up drug discovery is an attractive direction. A main challenge is the lack of data. We propose series of techniques, including pre-training on compounds/proteins, dual learning for drug design, and advanced models taking 3D information into consideration. We achieved promising improvement over existing methods and look forward to more collaboration.

讲者简介：Yingce Xia is a senior researcher at machine learning group, Microsoft Research Asia. He received his BS and Ph.D degrees from University of Science and Technology of China in 2013 and 2018, supervised by Dr. Tie-Yan Liu and Prof. Nenghai Yu. His research areas include machine learning, drug discovery and natural language processing. His works, dual learning and deliberation learning, have been widely used in natural language processing and computer vision. He published around 50 papers in top conferences like ICML, NeurIPS, ICLR, CVPR, AAAI, etc. In 2019, his team won four champions on WMT’19 translation competition (English-to-German, German-to-English, German-to-French and French-to-German), and the techniques have been transferred to Microsoft translation system.

朱建伟博士微软亚洲研究院 15:00 PM - 16:00 PM, Jan 20, 2022

报告摘要：蛋白质结构预测是从蛋白质序列出发获得蛋白质三维结构过程。近半个世纪以来，该问题一直被誉为“21世纪的生物物理学”最重要的课题之一。在过去50年里，获取蛋白质空间结构的方法以冷冻电镜、X-ray、NMR等实验技术为主，单个蛋白质的解析成本高达数月及数百万人民币。直至AlphaFold2的出现，使得通过AI对蛋白质结构进行预测取得重大突破。AlphaFold2 凭借其接近实验精度的成绩取得CASP14蛋白质空间结构预测比赛的榜首，这一成就也被Nature誉为“前所未有的进步”。在这次演讲中，我们将先讲述蛋白质结构预测背景知识，然后介绍几个过去几年代表性工作，最后介绍一下AlphaFold2基本原理。

讲者简介：朱建伟，微软亚洲研究院主管研究员。研究方向：生物信息学、计算生物学和机器学习。于2019年博士毕业于中国科学院计算技术研究所前瞻研究实验室。担任NeurIPS、Bioinformatics等期刊及会议审稿人。在Nature Communications、Bioinformatics等顶级SCI期刊发表论文10多篇。

姬水旺教授美国农工大学 10:00 AM - 11:00 AM, Dec 27, 2021

报告摘要：While classic physics is concerned with objects in the macro world, quantum physics and chemistry deal with particles at the atomic scale. In this talk, I will present our work on using deep learning and graph computing to accelerate scientific discoveries in the quantum field. Specifically, we consider representation learning of 3D molecular graphs in which each atom is associated with a spatial position in 3D. This is an under-explored area of research, and a principled neural message passing framework is currently lacking. We propose the spherical message passing as a novel and powerful scheme for 3D molecular learning. Based on 3D information representation in quantum physics, we develop the SphereNet for 3D molecular learning. In addition to predictive tasks, we consider the problem of generating 3D molecular geometries. We propose G-SphereNet as a novel autoregressive flow model for generating 3D molecular geometries.

讲者简介：Dr. Shuiwang Ji is currently a Professor and Presidential Impact Fellow in the Department of Computer Science & Engineering, Texas A&M University, directing the Data Integration, Visualization, and Exploration (DIVE) Laboratory. He received the Ph.D. degree in Computer Science from Arizona State University in 2010, advised by Prof. Jieping Ye. His research interests include artificial intelligence, machine learning, and quantum information science. Dr. Ji received the National Science Foundation CAREER Award in 2014. Currently, he serves as an Associate Editor for IEEE Transactions on Pattern Analysis and Machine Intelligence (TPAMI), ACM Transactions on Knowledge Discovery from Data (TKDD), and ACM Computing Surveys (CSUR). He regularly serves as an Area Chair or equivalent roles for AAAI Conference on Artificial Intelligence (AAAI), International Conference on Learning Representations (ICLR), International Conference on Machine Learning (ICML), International Joint Conference on Artificial Intelligence (IJCAI), ACM SIGKDD Conference on Knowledge Discovery and Data Mining (KDD), and Annual Conference on Neural Information Processing Systems (NeurIPS). Dr. Ji is a Fellow of AIMBE, a Distinguished Member of ACM, and a Senior Member of IEEE.

吴国斌博士滴滴 20:00 PM - 21:00 PM, Dec 23, 2021

报告摘要：分享人工智能和自动驾驶领域的最新研究进展，并分享从学生转变为职场人12年的底层思考，以及践行技术公益梦想的经历和感悟。

讲者简介：吴国斌，博士，PMP，现任滴滴出行科技合作总监，大数据分析与应用技术国家工程实验室副主任，滴滴企业科协秘书长，滴滴AI赋能社会共创平台秘书长。担任CCF-滴滴大数据联合实验室学术委员会委员，CCF青年计算机科技论坛2018-2019学术秘书和企业学术委员会委员。2003年和2008年分别获得中国科技大学工学学士学位和工学博士学位。加入滴滴前，任职微软亚洲研究院，主要工作集中在科研合作，人才培养和学术交流等方面。2014年和2016年所负责的科研项目分别获得微软CEO颁发的Ability Award和微软Hackathon最具影响力奖。

王亚珅博士中电科 20:00 PM - 21:00 PM, Dec 22, 2021

报告摘要：以微博检索为代表的短文本信息检索，在社交媒体等应用场景中有着重要应用价值。结构化语义常识信息能够使人类先验知识数字化、可计算化，已经在诸多自然语言处理任务中得到应用。同时提高相关任务的可解释性。在这次演讲中，主讲人将向大家介绍语义常识驱动的短文本信息检索相关探索以及讨论未来的研究方向。

讲者简介：王亚珅，博士，高级工程师，现任职于中国电子科技集团公司电子科学研究院，目前任社会安全风险感知与防控大数据应用国家工程研究中心认知智能研究室主任、公安部全国立体化社会治安防控体系部级系统建设组组长、公安部全国****态势监测预警平台建设组组长，中国科学技术大学、上海交通大学、北京工业大学研究生企业导师，长期从事自然语言处理、数据挖掘、知识图谱、社交网络分析等方向研究。主持国家自然科学基金青年科学基金、中国博士后科学基金会面上资助、中国电科集团新一代人工智能专项行动计划项目等。现任中国中文信息学会语言与知识计算专委会委员，中国图象图形学学会情感计算与理解专业委员会委员，中国人工智能学会青年工作委员会成员、会员，《无人系统技术》期刊青年编委。近五年，以第一作者（含通讯作者）身份发表TKDE、TKDD、ACL、WWW、IJCAI等顶级会议/期刊论文20余篇，受理发明专利20项，出版学术专著1部。常年担任AAAI(CCF A类)、ECML-PKDD(CCF B类)等人工智能/机器学习顶级会议的领域主席或PC Member。

童行行教授University of Illinois at Urbana-Champaign 11:00 AM - 12:00 AM, Dec 16, 2021

报告摘要：Network (i.e., graph) mining plays a pivotal role in many high-impact application domains. State-of-the-art offers a wealth of sophisticated theories and algorithms, primarily focusing on answering who or what type question. On the other hand, the why or how question of network mining has not been well studied. For example, how can we ensure network mining is fair? How do mining results relate to the input graph topology? Why does the mining algorithm `think’ a transaction looks suspicious? In this talk, I will present our work on addressing individual fairness on graph mining. We develop an upper bound to characterize the cost. Toward the end of my talk, I will also introduce some other recent work on addressing the why & how question of network mining, and share my thoughts about the future work.

讲者简介：Hanghang Tong is currently an associate professor at Department of Computer Science at University of Illinois at Urbana-Champaign. Before that he was an associate professor at Arizona State University. He received his M.Sc. and Ph.D. degrees from Carnegie Mellon University in 2008 and 2009, both in Machine Learning. His research interest is in large scale data mining for graphs and multimedia. He has received several awards, including AMiner “Most Influential Scholar Award”, honorable mention in data mining (2020), IEEE ICDM Tao Li award (2019), SDM/IBM Early Career Data Mining Research award (2018), NSF CAREER award (2017), ICDM 10-Year Highest Impact Paper award (2015), six best paper awards and several best paper finalists. He has published over 200 refereed articles. He is the Editor-in-Chief of SIGKDD Explorations (ACM) and an associate editor of ACM Computing Surveys (ACM). He is a distinguished member of ACM and a fellow of IEEE.

汤继良教授Michigan State University 10:00 AM - 11:00 AM, Dec 9, 2021

报告摘要：In the past few decades, artificial intelligence (AI) technology has experienced swift developments, changing everyone’s daily life and profoundly altering the course of human society. However, recent research and AI applications indicate that AI can cause unintentional harm to humans. In this talk, I will first present the good, the bad, and the ugly of AI with real-world examples. To avoid the adverse effects that AI could bring to people, so people can fully trust and live in harmony with AI technologies, we have to tame the bad and the ugly of AI. Consequently, trustworthy AI has recently garnered increased attention. Trustworthy AI is a large and complex subject, involving various dimensions. I will focus on the six most crucial dimensions in achieving trustworthy AI: Safety & Robustness, Nondiscrimination & Fairness, Explainability, Privacy, Accountability & Auditability, and Environmental Well-being. For each dimension, I will briefly introduce the recent related technologies.

讲者简介：Jiliang Tang is an associate professor (assistant professor, 2016-2021) in the computer science and engineering department at Michigan State University. Before that, he was a research scientist in Yahoo Research. He got his PhD from Arizona State University in 2015 under Dr. Huan Liu and MS and BE from Beijing Institute of Technology in 2010 and 2008, respectively. His research interests include social computing, data mining and machine learning and their applications in education and biology. He was the recipient of 2021 IEEE ICDM Tao Li Award, 2021 IEEE Big Data Security Junior Research Award, 2020 ACM SIGKDD Rising Star Award, 2020 Distinguished Withrow Research Award, 2019 NSF Career Award, and 8 best paper awards (or runner-ups) including WSDM2018 and KDD2016. His dissertation won the 2015 KDD Best Dissertation runner up and Dean's Dissertation Award. He serves as conference organizers (e.g., KDD, SIGIR, WSDM and SDM) and journal editors (e.g., TKDD and TKDE). He has published his research in highly ranked journals and top conference proceedings, which have received tens of thousands of citations with h-index 67 (Google Scholar) and extensive media coverage.

唐建教授HEC Montreal 10:00 AM - 11:00 AM, Dec 7, 2021

报告摘要：Drug discovery is a very long and expensive process, taking on average more than 10 years and costing $2.5B to develop a new drug. Artificial intelligence has the potential to significantly accelerate the process of drug discovery by extracting evidence from a huge amount of biomedical data and hence revolutionizes the entire pharmaceutical industry. In particular, graph representation learning and geometric deep learning--a fast growing topic in the machine learning and data mining community focusing on deep learning for graph-structured and 3D data---has seen great opportunities for drug discovery as many data in the domain are represented as graphs or 3D structures (e.g. molecules, proteins, biomedical knowledge graphs). In this talk, I will introduce our recent progress on geometric deep learning for drug discovery and also a newly released open-source machine learning platform for drug discovery, called TorchDrug.

讲者简介：Jian Tang is an assistant professor at Mila-Quebec AI Institute and also at Computer Science Department and Business School of University of Montreal. He is a Canada CIFAR AI Research Chair. His main research interests are graph representation learning, GNNs, geometric deep learning, deep generative models, knowledge graphs and drug discovery. During his PhD, he was awarded with the best paper in ICML2014; in 2016, he was nominated for the best paper award in the top data mining conference World Wide Web (WWW); in 2020, he is awarded with Amazon and Tencent Faculty Research Award. He is one of the most representative researchers in the growing field of graph representation learning and has published a set of representative works in this field such as LINE and RotatE. His work LINE on node representation learning has been widely recognized and is the most cited paper at the WWW conference between 2015 and 2019.

邰骋、汤林鹏博士墨奇科技 14:30 PM - 16:00 PM, Oct 9, 2021

报告摘要：传统数据库和大数据系统主要处理SQL结构化数据和半结构化数据，并在此基础上驱动了信息产业的革命。而近年来，以图像、视频、音频、文本、时间序列为代表的复杂数据不断增长，深度神经网络为代表的AI算法能够对这类数据进行初步的处理并提取价值，并促成了近十年来AI发展的浪潮。但是，我们仍然缺乏一种能够统一处理这类复杂数据的数据库，这大大提高了AI从研发到落地的成本。在本次演讲中，我们提出一种新型的AI知识数据库，支持从复杂数据中提出以图和向量为代表的中间表示，并支持对于这类中间表示的高效存储和检索。我们会讨论这个新型数据库的一些关键设计要点，以及相关的开放问题。作为一个行业应用案例，通过多尺度的图像特征表示，小样本无监督学习框架和高效、异构的比对算法，我们打造了世界上首个十亿级别的精准可靠、无需标注的指纹数据库。同时，我们还将图像识别和密码学结合，构建了保护隐私的生物识别算法。利用这一新型的AI数据库，我们希望加快各行各业AI从研发到落地的速度，并促进AI算法的进一步发展。

讲者简介：邰骋，墨奇科技CEO及联合创始人。普林斯顿大学应用数学博士，师从知名的应用数学领域科学家、中国科学院院士鄂维南。主要研究方向为大规模非结构化数据处理算法和系统，在SIAM、JMLR、ICML等国际期刊会议发表多篇论文，是国内人工智能、应用数学方面的领军学者。2016年创立墨奇科技，致力于数学框架的构建及产业化应用。汤林鹏，墨奇科技 CTO 及联合创始人。普林斯顿大学计算机科学博士，师从知名计算机科学家、中国工程院外籍院士李凯。在KDD、AISTATS、OSDI、ATC等顶级机器学习和系统会议上发表多篇论文，获得 KDDCup 数据挖掘比赛第一名、CIKM 最佳学生论文等奖项。2016年联合创立墨奇科技，负责算法研发、系统设计和开发以及软硬件研发。