讲座题目:《Data–Knowledge Environment and Knowledge Landmarks in Machine Learning》
主讲嘉宾:Witold Pedrycz教授
讲座时间:2025年4月26日(星期六)10:00-11:40
讲座地点:沙河校区二教203室
讲座主持:李玉龙
讲座摘要:The rapid progress in ML, exemplified by LLMs, relies heavily on data, yet the role of knowledge remains understudied. We propose KD-ML, a unified paradigm integrating data and knowledge in ML design. Key challenges include reconciling numeric data with symbolic knowledge, addressed via information granulation. We categorize knowledge as scientific (e.g., physics-informed ML) or commonsense (e.g., rule-based constraints), and explore representation schemes. KD-ML enhances models through: Knowledge-augmented regularization (constraining data-driven models), Granular expansions (bridging abstraction levels), Rule-guided learning (incorporating expert knowledge). This approach mitigates data blinding—overreliance on data—while improving interpretability. We also examine LLM-based knowledge acquisition for scalable KD-ML. By harmonizing data and knowledge, KD-ML fosters more robust, explainable, and efficient learning systems.
嘉宾简介:

Witold Pedrycz (IEEE终身会士),加拿大埃德蒙顿阿尔伯塔大学电气和计算机工程系的教授,加拿大皇家科学院院士,波兰科学院外籍院士,加拿大计算智能研究中心首席科学家,历任IFSA主席和北美模糊信息处理协会会长。Witold Pedrycz院士曾获得包括IEEE系统人类和控制论学会的诺伯特-维纳奖、IEEE加拿大计算机工程奖、欧洲软计算中心的卡贾斯特尔软计算奖、基拉姆奖、IEEE计算智能学会的模糊先锋奖,以及2019年IEEE系统、人类和控制论学会的功勋服务奖等在内的多项荣誉。Witold Pedrycz院士的主要研究方向包括计算智能、模糊建模和粒度计算、知识发现和数据挖掘、模糊控制、模式识别、基于知识的神经网络、关系计算和软件工程等。Witold Pedrycz院士现任《Information Sciences》主编、《WIREs Data Mining and Knowledge Discovery (Wiley)》主编、《Int. J. of Granular Computing (Springer)》和《J. of Data, Information and Management (Springer)》的联合主编以及IEEE Trans. SMC、IEEE Trans. Fuzzy Systems等多个国际知名期刊的编委。
本次讲座得到“中央财经大学2025年专题学术讲座”支持。
主办单位:管理科学与工程学院