Lim Tong Ming, Professor, Centre for Business Incubation and Entrepreneurial Ventures (CBIEV), Tunku Abdul Rahman University of Management and Technology (TARUMT), Malaysia

Title: A localised SLM and ASR for Malay Bahasa Rojak corpus with an experimental demo.

Abstract： Language serves as the primary medium of interaction between intelligent systems and end users. However, effective communication often extends beyond formal language to include domain-specific terminology, colloquialisms, and slang that are deeply embedded within particular communities and industries. These linguistic variations are typically underrepresented in existing AI models due to the scarcity of high-quality training data, presenting significant challenges for low-resource languages and specialized application domains. This talk presents TAR UMT's journey in developing and fine-tuning Small Language Models (SLMs) tailored to the linguistic needs of specific industries and Bahasa Rojak, the code-mixed language widely used in everyday communication among Malaysians. We will showcase our end-to-end development process, including the collection and annotation of text and speech datasets, the adaptation and fine-tuning of pretrained language models, and the AI tools developed over the past several years to support these efforts. In addition, we will present a comparative evaluation of model performance before and after fine-tuning, highlighting the effectiveness of domain adaptation in improving language understanding and generation. Finally, we will discuss the key technical challenges encountered throughout the project and outline future research directions for advancing AI solutions for low-resource languages and specialized domains.

Bio: Prof. Lim has about 12 years of commercial software and close 27 years of academic experiences with two software companies and four universities respectively. Having many years of industry exposure, Prof Lim understands well the needs of the IT industry. His experience has allowed him to have deep understanding of industrial driven applied research projects. Prof Lim's research areas span from object technologies on software engineering to databases, peer to peer technologies, knowledge sharing to social media analytics and social influence maximization with more than 140 conference proceedings and journals published.

He has completed many grants where these grants are secured from FRGS, eScience, SKMM, MDec, Monash Internal grant, UTAR internal grant, Sunway Internal and Tunku Abdul Rahman University of Management and Technology (TARUMT) grant. More than 20 master and 5 PhD students under his supervision have graduated from Monash University, Universiti Tunku Abdul Rahman, Sunway University and TARUMT in the last 27 years. Currently Prof Lim is the Director for CBIEV and Professor in FOCS. He is a professional ACM member and am currently society affiliate member of IEEE association.

Jiang Yujian, Professor, Doctoral Supervisor and Master's Supervisor, School of Information and Communication Engineering, Communication University of China, Head of the Department of Performing Arts Engineering; Deputy Director of the Key Laboratory of Audiovisual Technology and Intelligent Control System, Ministry of Culture and Tourism.

Title: Music-driven Stage Lighting Effect Generation

Abstract： Music-driven Stage Lighting Effect Generation, will focus on music feature extraction, stage lighting effect parameter modeling, matching between stage visual rhythm and music, generative algorithm-assisted design, and experimental demonstrations. It will discuss the application potential of artificial intelligence in performing arts engineering, stage design, and digital art creation.

Bio: Prof. Jiang Yujian holds a Ph.D. in Engineering and is a professor at the School of Information and Communication Engineering, Communication University of China, where he serves as a doctoral supervisor and master's supervisor. He is currently Head of the Department of Performing Arts Engineering and Deputy Director of the Key Laboratory of Audiovisual Technology and Intelligent Control System, Ministry of Culture and Tourism. His research interests include artificial intelligence, performing arts AIGC, digital twins for performing arts, intelligent perception and human-computer interaction in performing arts, audiovisual technology and intelligent control, and big data analysis and mining.

He has led multiple research projects, including National Key R&D Program projects/topics and provincial- and ministerial-level projects. He has published more than 50 SCI/EI-indexed papers and has been granted eight national invention patents. His related technical achievements were applied to the sports presentation interactive experience for ice hockey at the Beijing 2022 Olympic Winter Games.

Yu Yong Poh, Associate Professor, Deputy Director, Centre for Business Incubation and Entrepreneurial Ventures (CBIEV), Tunku Abdul Rahman University of Management and Technology (TAR UMT), Malaysia

Title: Predictive Consumption: How Socio-Cultural Spending Patterns Drive Personalized Financial Recommendations

Abstract： Traditional financial systems and recommendation engines are typically built around standardized transaction data. However, when these platforms scale across diverse regions, they often face unexpected drops in accuracy and user engagement. This misalignment occurs because spending is not purely financial—it is deeply shaped by cultural habits, varying from community-driven group buying and seasonal festival shopping to regional attitudes toward saving, credit, and debt. This keynote explores how data intelligence can bridge the gap between human cultural dynamics and automated financial systems. We will discuss how modern recommendation systems can move beyond basic user history by integrating alternative behavioral data that reflects local socio-cultural realities. By leveraging context-aware data structures and predictive analytics, financial platforms can better understand the underlying patterns of regional consumer behavior. The presentation will outline practical frameworks for analyzing these diverse spending habits to deliver highly personalized financial services—such as localized micro-investments and adaptive credit options—while maintaining system fairness. Ultimately, this session highlights how recognizing cultural nuance within data models allows us to build smarter, more inclusive, and highly effective digital economies.

Bio: Associate Professor Ts. Dr. Yu Yong Poh is a distinguished professional in Data Science and Artificial Intelligence with over 15 years of experience. He currently serves as the Deputy Director of the Centre for Business Incubation and Entrepreneurial Ventures (CBIEV) at Tunku Abdul Rahman University of Management and Technology (TAR UMT), where he actively drives academic-industrial collaboration, entrepreneurial growth, commercialization, and consultancy services. Throughout his career, Dr. Yu has successfully led numerous Data and AI projects across diverse sectors, including banking, healthcare, manufacturing, and digital marketing.

A dedicated educator and trainer, he frequently conducts professional training programs in AI, Data Science, Machine Learning, Big Data Analytics, and programming. He holds a Bachelor’s and a Ph.D. in Engineering from the University of Malaya (UM) and has published extensively in ISI journals and conference proceedings. Dr. Yu is a certified Enterprise Data Scientist, a SAS Certified Professional and a registered Professional Technologist (Ts.).

Guixuan Zhang, Associate Professor, Beijing University of Posts and Telecommunications, China.

Title: Zero-Shot Text-to-Speech for Cross-Lingual Cinematic Dubbing

Abstract： Cross-lingual cinematic dubbing is vital to cross-cultural communication, yet traditional pipelines are costly and dependent on professional voice actors. This talk explores how zero-shot text-to-speech (TTS) technologies—capable of cloning a character's voice from a reference speech of a movie and transferring it to another language—are reshaping cross-lingual film dubbing. We review state-of-the-art zero-shot TTS architectures, examine the unique demands of cinematic dubbing such as emotional fidelity and character timbre consistency, and present practical pipelines integrating speech recognition, machine translation, and expressive synthesis.

Bio: Guixuan Zhang is an Associate Professor with Beijing University of Posts and Telecommunications, China. He received the Ph.D. degree in pattern recognition from the University of Chinese Academy of Sciences (UCAS) and Institute of Automation, Chinese Academy of Sciences (CASIA), Beijing, in 2017. He worked at CASIA as an assistant/associate professor from 2017 to 2024. His research interests include artificial intelligence and their applications in image and video analysis, domain multi-modal large language models, and 3-D human reconstruction.

He has been presiding over several research projects, including the National Key Research and Development Program of China. Dr. Zhang has served as PC member or reviewer for several leading international conferences, and guest editors for journals.