Keynote Speaker I

Koichi Takeuchi
Okayama University, Japan 

Constructing a Japanese Predicate-Argument Database for Ex​tracting Similar Semantic Events and Their Elements 

Semantic role labeling (SRL) is a fundamental task in natural language understanding, aimed at identifying the relationships between predicates and their arguments. Semantic roles represent classifications of these relationships that go beyond surface-level grammatical functions. For example, in “He opened the door” and “The door opened,” the subjects differ syntactically (“he” vs. “the door”), but the semantic roles remain consistent: “he” is the Agent and “the door” is the Theme in both cases. While resources such as PropBank and FrameNet provide semantic role annotations for English, there is no widely accepted standard set of semantic roles, and annotated data for Japanese remains limited.To address this gap, we are building a lexicon and annotated dataset of semantic roles and conceptual frames for Japanese. In this talk, I will introduce the construction of a comprehensive Japanese predicate-argument database designed to support the extraction of semantically similar events and their elements. We have developed a dictionary of semantic roles and conceptual frames for approximately 10,000 Japanese predicates and created annotations for around 50,000 example sentences from NPCMJ (Ninjal Parsed Corpus of Modern Japanese), a syntactically annotated corpus. During the construction, we extended the base features proposed in PropBank and FrameNet to a Japanese lexical database in order to capture alternations in Japanese. By applying a hierarchical structure to conceptual frames—similar to the synset hierarchy in WordNet—we enable the organization of more specific predicates (e.g., evacuate) under abstract ones (e.g., move). In the presentation, I will show concreate examples of semantic roles and conceptual frame structures, and discuss their potential applications.

Koichi Takeuchi is an Associate Professor at the Graduate School of Environmental, Life, Natural Science and Technology, Okayama University. He received his PhD from Nara Institute of Science and Technology in 1998 and subsequently worked as an Assistant Professor.at National Institute of Informatics. From 2002 to 2003, he conducted research at INRIA Lorraine in France as an invited researcher. He joined Okayama University in 2003 and has been serving as an Associate Professor since 2021. His current research interests include automatic essay grading, analysis of predicate-argument structures, text mining, terminology extraction, and the application of language models in the medical field.

Keynote Speaker II

Boris Koldehofe
Technical University of Ilmenau, Germany

Analog Networking: An enabler for supporting energy-efficient data-driven applications?  

Nowadays, data-driven distributed applications heavily rely on efficiently transferring data between producers and consumers. Programmable and adaptive computer networks have the potential to significantly reduce the cost of data movements and even accelerate data preprocessing on the path from producers to consumers. Yet, current computer networks also rely on costly components, such as TCAM memory, to enable high performance, i.e., line-rate processing in the switches and routers of a communication network. In this talk, we focus on how novel analog components, such as memristors or optical elements, can be utilized to offer more energy-efficient operations and discuss the required changes in programming such networks. In the context of a novel programming model for network switches, called PCAM, we demonstrate the potential for future adaptive data-driven applications and challenges for future work.

Boris Koldehofe is a Full Professor at the Technical University of Ilmenau, leading the Distributed and Operating Systems Group at the department of Computer Science and Automation. He received a Ph.D. degree from Chalmers University of Technology, Gothenburg, Sweden, in 2005. Since then, he has worked in the field of distributed and network-centric computing systems at the EPFL (PostDoc), the University of Stuttgart, the Technical University of Darmstadt (Senior researcher and lecturer), and the University of Groningen (Full professor). He has a long-standing interest in event-based and stream processing systems, covering issues related to scalability, performance, mobility, reliability, and security. His current research focuses complementary on software-defined networks, adaptive communication middleware, distributed in-network computing, and energy-efficient computing. He has contributed to more than 150 scientific publications in major journals, e.g., the IEEE Transactions on Networking (ToN) and the IEEE Transactions on Parallel and Distributed Systems, and conferences, e.g., the ACM/USENIX Middleware and the ACM DEBS conferences. He has also served as a Tutorial Speaker for the ACM/USENIX Middleware, ACM DEBS, GI, and NetSys conferences.

Keynote Speaker III

Minh N. H. Nguyen 
The University of Danang – Vietnam-Korea University of Information and Communication Technology

Towards Knowledge Transfer and Collaborative Mechanisms across Distributed Models in Federated Learning 

Emerging cross-device artificial intelligence (AI) applications require a transition from conventional centralized learning systems towards large-scale distributed AI systems that can collaboratively perform complex learning tasks. In this regard, federated learning (FL) lays out a novel learning mechanism for building distributed machine learning systems for multiple clients to collaboratively train a generalized global model without sharing their private data. In this talk, we first cover various designs of FL to cope with different perspectives of FL systems for realizing robust personalized Federated Learning (FL) systems, efficient model aggregation methods for dealing with the consequences of non-i.i.d. properties of client's data, often referred to as statistical heterogeneity and small local data samples from the various data distributions. We developed novel approaches for knowledge transfer between the global model and local models regarding single-modal data as well as multimodal data. Utilizing a variety of recent techniques such as knowledge distillation, contrastive learning regularization, and class-based prototype representation opens a promising direction for transfer knowledge across models. Second, we then introduce the proposed Democratized learning (Dem-AI) systems which is a holistic framework for building self-regulating scalable distirbuted learning systems extending beyond traditional federated learning. This approach relies on hierarchical self-organization of personalized learning agents through group contribution and hierarchical generalization to enable both specialized and generalized learning processes. Moreover, we also share a brief introduction to the emerging multi-agent communication protocols that enable intelligent agentic collaboration and open fruitful research directions in robotics, autonomous systems, and medical agents. To this end, the talk introduce promising collaborative schemes for future AI agent collaboration across diverse domains, fostering more robust, scalable, and intelligent distributed learning systems.

Minh N. H. Nguyen received a BE degree in Computer Science and Engineering from Ho Chi Minh City University of Technology, Vietnam, in 2013 and a Ph.D. degree in Computer Science and Engineering from Kyung Hee University, South Korea, in 2020. He continued the research on Federated Learning and Democratized Learning with the PostDoc at Intelligent Networking lab, Kyung Hee University, South Korea till 2022. He is Deputy Head of Department of Science, Technology, and International Cooperation, and In charge of Research Program at Digital Science and Technology Institute, The University of Danang – Vietnam - Korea University of Information and Communication Technology, Vietnam. He received the best KHU Ph.D. thesis award in engineering in 2020. He received the best KHU Ph.D. thesis award in engineering in 2020. Among over 60 his publications, he had 20 publications in premier journals such as IEEE TNNLS, Information Fusion, ACM/IEEE ToN, IEEE TWC, IEEE TMC, IEEE TVT, IEEE IoT, IEEE CM, IEEE CIM, Neural Networks, EAAI, etc., and INFOCOM rank A* conference. He participated as Track Chair and Session Chair for International ATC and CITA Conferences. His research interests include federated learning, natural language processing, multimodal learning, and wireless communications. He delivered Federated Learning keynotes in EAI ADHOCNETS-2023, Artificial Intelligence & Machine Learning (AIM 2024).