活动通知：GAMES Webinar346期-Advanced Interactive AI Systems（10月31日）

报告题目

SLInterpreter: An Exploratory and Iterative 

Human-AI Collaborative System 

for GNN-based Synthetic Lethal Prediction

报告嘉宾

姜浩然

上海科技大学

报告时间

2024年10月31号 晚上8:00-8:20（北京时间）

报告方式

GAMES 直播间: 

https://live.bilibili.com/h5/24617282

报告摘要

Synthetic Lethal (SL) relationships, although rare among the vast array of gene combinations, hold substantial promise for targeted cancer therapy. Despite advancements in AI model accuracy, there remains a persistent need among domain experts for interpretive paths and mechanism explorations that better harmonize with domain-specific knowledge, particularly due to the significant costs involved in experimentation. To address this gap, we propose an iterative Human-AI collaborative framework comprising two key components: 1) Human-Engaged Knowledge Graph Refinement based on Metapath Strategies, which leverages insights from interpretive paths and domain expertise to refine the knowledge graph through metapath strategies with appropriate granularity. 2) Cross-Granularity SL Interpretation Enhancement and Mechanism Analysis, which aids domain experts in organizing and comparing prediction results and interpretive paths across different granularities, thereby uncovering new SL relationships, enhancing result interpretation, and elucidating potential mechanisms inferred by Graph Neural Network (GNN) models. These components cyclically optimize model predictions and mechanism explorations, thereby enhancing expert involvement and intervention to build trust. This framework, facilitated by SLInterpreter, ensures that newly generated interpretive paths increasingly align with domain knowledge and adhere more closely to real-world biological principles through iterative Human-AI collaboration. Subsequently, we evaluate the efficacy of the framework through a case study and expert interviews.

嘉宾简介

Haoran Jiang(姜浩然) 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 Unvieristy, under the guidance of Prof. Quan Li. (李权). His research primarily focuses on user participation and cognitive debiasing in human-AI collaboration processes, which includes: Enhancing the interpretability of biomedical models and eliminating cognitive biases in human decision-making activities.

个人主页

https://jianghr-shanghaitech.github.io/jianghr.github.io/

报告题目

Designing Scaffolding Strategies 

for Conversational Agents in Dialog Task 

of Neurocognitive Disorders Screening

报告嘉宾

胡佳雄

香港科技大学

报告时间

2024年10月31号 晚上8:20-8:40（北京时间）

报告方式

GAMES 直播间: 

https://live.bilibili.com/h5/24617282

报告摘要

Regular screening is critical for individuals at risk of neurocognitive disorders (NCDs) to receive early intervention. Conversational agents (CAs) have been adopted to administer dialog-based NCD screening tests for their scalability compared to human-administered tests. However, unique communication skills are required for CAs during NCD screening, e.g., clinicians often apply scaffolding to ensure subjects’ understanding of and engagement in screening tests. Based on scaffolding theories and analysis of clinicians’ practices from human-administered test recordings, we designed a scaffolding framework for the CA. In an exploratory wizard-of-Oz study, the CA empowered by ChatGPT administered tasks in the Grocery Shopping Dialog Task with 15 participants (10 diagnosed with NCDs). Clinical experts verified the quality of the CA’s scaffolding and we explored its effects on task understanding of the participants. Moreover, we proposed implications for the future design of CAs that enable scaffolding for scalable NCD screening.

嘉宾简介

I’m Jiaxiong HU 胡佳雄. Presently, I serve as a Post-doctoral Fellow at the Hong Kong University of Science and Technology (HKUST), under the guidance of Professor Xiaojuan Ma. In June 2022, I obtained my Ph.D. degree from Tsinghua University (THU). My dissertation, titled The Conversational User Interface Design Method Based on Emotional Design Theory, was awarded the Outstanding Thesis Award and was supervised by Professor Yingqing Xu.My scholarly pursuits primarily revolve around enhancing human-computer conversational interactions by employing emotional design theory and leveraging state-of-the-art artificial intelligence technologies. My prior research encompasses a diverse range of subjects, including aging, customer service, accessibility, mental health, art, and creativity. My works have been presented at top conferences such as CHI, CSCW, and IEEE Transactions. Meanwhile, I have held the position of Associate Chair for CSCW since 2023.

个人主页

https://www.notion.so/jiaxiong/Jiaxiong-Hu-fdd7394b52f74ae78ab438aa93164c3d

报告题目

Enhancing Medical Data Labeling 

and Diagnostic Skills through Machine Learning 

and Visual Analytics

报告嘉宾

欧阳阳

上海科技大学

报告时间

2024年10月31号 晚上8:40-9:00（北京时间）

报告方式

GAMES 直播间: 

https://live.bilibili.com/h5/24617282

报告摘要

Machine learning models have great potential to enhance our ability to analyze complex medical data, from extracting insights in medical texts to integrating multimodal information like patient records and images. However, these models require high-quality labeled data, which is often time-consuming to generate. Without accurate labels, even advanced models struggle to produce reliable results, making the labeling process a critical challenge. To address this, we developed KMTLabeler, a visual analytics tool that streamlines the medical text labeling process. By combining visual cluster analysis, active learning, and task-specific embeddings, KMTLabeler allows domain experts to efficiently label texts with high accuracy. It continuously integrates expert feedback, refining the labeling process and significantly reducing the time and effort required, as demonstrated in case studies. In medical education, models also play a key role, especially in helping novice physicians navigate multimodal data. We created DiagnosisAssistant, a visual analytics system designed for Simulation-based Medical Education (SBME). By leveraging historical medical records, DiagnosisAssistant visualizes relationships between data modalities, offers diagnostic hints, and enables comparative analysis, improving learners' diagnostic skills. Together, KMTLabeler and DiagnosisAssistant showcase how combining machine learning with visual analytics can enhance both medical text labeling and education, streamlining workflows and improving the learning experience for future healthcare professionals.

嘉宾简介

Yang Ouyang is a second-year Ph.D. student at the ViSeer Lab (智能交互与可视分析实验室) in the School of Information Science and Technology, ShanghaiTech University, under the supervision of Professor Quan Li (李权). His research focuses on empowering deeper insights, intuitive analysis, and effective storytelling in complex data landscapes by integrating visual analytics with AI advancements. His research goal is to support a more profound understanding of complex data through scalable visual interfaces and interpretable techniques, while fostering improved human-AI collaboration, interaction, and communication through AI-driven visualizations.

个人主页

https://cat-ouyang.github.io/

主持人简介

李权

上海科技大学

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, UIST, 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/

GAMES Website: http://games-cn.org