GAMES Webinar 2024 – 322期(研究生成长论坛：”人机协同“ 硕博阶段科研分享) ｜Talk+Panel形式

【GAMES Webinar 2024-322期】(可视化专题-研究生成长论坛：“人机协同”硕博阶段科研分享｜Talk+Panel形式)

详细日程：2024年5月9号 20:00-22:00(北京时间)

20:00-20:12   Utilizing AI and Visualization Method to Mitigate Human Cognitive Bias in Decision-Making Processes(Qianyu Liu，ShanghaiTech University)

20:12-20:24   Improving Human-AI Collaboration in Traffic Signal Control: Multi-Agent Reinforcement Learning Understanding(Yutian Zhang，Sun Yat-sen University)

20:24-20:36   Competent but Rigid: Identifying the Gap in Empowering AI to Participate Equally in Group Decision-Making(Chengbo Zheng，Hong Kong University of Science and Technology)

20:36-20:48   Your Co-Workers Matter: Evaluating Collaborative Capabilities of Language Models in Blocks World(Guande Wu，New York University)

20:48-22:00     圆桌研讨

研讨嘉宾：Qianyu Liu、Yutian Zhang、Chengbo Zheng、Guande Wu

报告嘉宾：Qianyu Liu(ShanghaiTech University)

报告时间：2024年5月9号星期四20:00-20:12(北京时间)

报告题目：Utilizing AI and Visualization Method to Mitigate Human Cognitive Bias in Decision-Making Processes

讲者简介：

Qianyu Liu(刘芊渝) is a 2nd year master student at ViSeer Lab (智能交互与可视分析实验室) , School of Information Science and Technology, ShanghaiTech University, under the guidance of Prof. Quan Li. Her researches focus encompassing the domains of decision-making analysis, probing into human cognitive biases, and the development of interactive systems attuned to the mitigation of biases. Concurrently, her hold a keen interest in the exploration of user-centric design paradigms aimed at enhancing emotional experience perception. Website: https://qianyu-l.github.io

报告摘要：

This presentation introduces BiasEye, an innovative system designed to enhance impartial candidate assessment by addressing cognitive biases. By integrating machine learning and advanced data visualization techniques, BiasEye empowers reviewers to identify and mitigate biases in real-time. The talk will delve into the system’s design, model selection, and user study experiments, offering valuable insights into the potential of human-AI collaboration in addressing bias in decision-making processes.

报告嘉宾：Yutian Zhang(Sun Yat-sen University)

报告时间：2024年5月9号星期四20:12-20:24(北京时间)

报告题目：Improving Human-AI Collaboration in Traffic Signal Control: Multi-Agent Reinforcement Learning Understanding

讲者简介：

Yutian Zhang is currently working towards the M.S. degree in the School of Intelligent Systems Engineering at Sun Yat-sen University. He received a B.S. degree in transportation engineering from Sun Yat-Sen University. His research interests include interpretable machine learning, visual analytics and transportation big data. Email: zhangyt85@mail2.sysu.edu.cn

报告摘要：

Traffic congestion hinders global city development. Intelligent Traffic Signal Control (TSC), especially using reinforcement learning (RL), shows promise. However, current evaluation methods for RL-based TSC are limited, hindering practical implementation. Effective TSC requires coordination across intersections, challenging existing visual analysis solutions. To address this, we propose MARLens, a visual analytics approach for Multi-Agent RL (MARL)-based TSC. MARLens offers versatile exploration of model features, revealing decision-making processes and agent interactions. Multiple visualization views and a traffic simulation module aid in quick identification of critical states. Validation through case studies, expert interviews, and user studies demonstrates MARLens’ effectiveness in enhancing understanding and informing traffic management strategies.

报告嘉宾：Chengbo Zheng(Hong Kong University of Science and Technology)

报告时间：2024年5月9号星期四20:24-20:36(北京时间)

报告题目：Competent but Rigid: Identifying the Gap in Empowering AI to Participate Equally in Group Decision-Making

讲者简介：

Chengbo Zheng is a PhD candidate at the Hong Kong University of Science and Technology, supervised by Prof. Xiaojuan Ma. His research interests include human-AI collaboration, AI in education, and data visualization. His work has been published in ACM CHI, IEEE VIS, and IEEE PacificVic. He has also served as a reviewer for ACM CHI, ACM UIST, ACM CSCW, and IEEE VIS. Homepage: https://www.cbzheng.link

报告摘要：

Existing research on human-AI collaborative decision-making focuses mainly on the interaction between AI and individual decision-makers. There is a limited understanding of how AI may perform in group decision-making. This paper presents a wizard-of-oz study in which two participants and an AI form a committee to rank three English essays. One novelty of our study is that we adopt a speculative design by endowing AI equal power to humans in group decision-making. We enable the AI to discuss and vote equally with other human members. We find that although the voice of AI is considered valuable, AI still plays a secondary role in the group because it cannot fully follow the dynamics of the discussion and make progressive contributions. Moreover, the divergent opinions of our participants regarding an “equal AI” shed light on the possible future of human-AI relations.

报告嘉宾：Guande Wu(New York University)

报告时间：2024年5月9号星期四20:36-20:48(北京时间)

报告题目：Your Co-Workers Matter: Evaluating Collaborative Capabilities of Language Models in Blocks World

讲者简介：

Guande (吴冠德) is a Ph.D. candidate in Computer Science at New York University, where he is advised by Prof. Claudio Silva. His research focuses on human-AI collaboration and visual analytics, with a special interest in integrating the user-state model into large language models and other deep-learning models. Guande has published his work in top-tier venues such as CVPR, IEEE VIS, and ACM CHI, and he has received an honorable mention at IEEE VIS. He has also served as a reviewer for AAAI, TPAMI, ACM UIST, ISMAR and TCSVT. Homepage: https://www.gdwu.xyz

报告摘要：

Language agents that interact with the world on their own have great potential for automating digital tasks. While large language model  (LLM) agents have made progress in understanding and executing tasks such as textual games and webpage control, many real-world tasks also require collaboration with humans or other LLMs in equal roles, which involves intent understanding, task coordination, and communication. To test LLM’s ability to collaborate, we design a blocks-world environment, where two agents, each having unique goals and skills, build a target structure together. To complete the goals, they can act in the world and communicate in natural language. Under this environment, we design increasingly challenging settings to evaluate different collaboration perspectives, from independent to more complex, dependent tasks. We further adopt chain-of-thought prompts that include intermediate reasoning steps to model the partner’s state and identify and correct execution errors. Both human-machine and machine-machine experiments show that LLM agents have strong grounding capacities, and our approach significantly improves the evaluation metric.

主持人简介：

李权，上海科技大学信息科学与技术学院助理教授（终身教授序列）、研究员、博士生导师，从事人工智能及可视分析、可解释性机器学习以及人机交互技术的研究。他博士毕业于香港科技大学计算机科学与工程学系。任中国图象图形学学会可视化与可视分析专委会委员，IEEE VIS Paper程序委员会委员、ChinaVis论文国际程序委员会委员、IEEE VIS, EuroVis, PacificVis, ChinaVis, ACM CHI/CSCW及TVCG等顶级学术会议期刊审稿人，他曾任美国佐治亚理工学院计算机科学与工程学院的访问研究员、微众银行人工智能部资深研究员及网易游戏资深研究员。他的学术成果发表在IEEE VIS, EuroVis, IEEE PacificVis, ACM CHI, CSCW, IUI, CGF, TVCG等可视化及人机交互顶级期刊和会议。主持国家自然科学基金面上项目。更多信息见https://faculty.sist.shanghaitech.edu.cn/liquan/

石楚涵，东南大学计算机科学与工程学院副教授，2023年获得香港科技大学计算机专业博士学位。研究方向包括数据可视化、可视分析、人机交互及其在自然科学、精准医疗等领域的应用，在相关领域的IEEE TVCG、ACM CHI、ACM CSCW等国际顶级期刊和会议发表论文10余篇。曾担任ACM CSCW、PacificVis VisNotes等国际权威会议的程序委员会委员，以及ACM CHI、ACM UIST等会议审稿人。更多信息见：https://shichuhan.github.io/

GAMES主页的“使用教程”中有 “如何观看GAMES Webinar直播？”及“如何加入GAMES微信群？”的信息；
GAMES主页的“资源分享”有往届的直播讲座的视频及PPT等。
观看直播的链接：https://live.bilibili.com/h5/24617282