ASEAN Young Scientists Connect 2026 Pitching Trans-GMS LLM at Dusit Thani Bangkok

Tue, 19 May 2026 09:00:00 +0000

Field Report from Bangkok

At ASEAN Young Scientists Connect (AYSC) 2026, hosted at Dusit Thani Bangkok, researchers and innovators from across Southeast Asia gathered to discuss the future of artificial intelligence, healthcare technologies, and regional scientific collaboration. Among the featured presentations was Trans-GMS LLM, a multilingual medical AI initiative designed for low-resource healthcare environments across the Greater Mekong Subregion.

The event created a collaborative platform for young scientists, engineers, and researchers from ASEAN countries to exchange ideas on trustworthy AI, multilingual technologies, and inclusive innovation strategies capable of addressing regional healthcare challenges.

Pitching Trans-GMS LLM in Three Minutes

One of the most challenging aspects of the session was translating a technically complex research project into a concise and compelling three-minute scientific pitch.

The presentation introduced Trans-GMS LLM, a multilingual large language model designed to support healthcare communication and medical intelligence across linguistically diverse Southeast Asian regions, including:

Thailand
Laos
Cambodia
Myanmar
Vietnam
Malaysia
Singapore
Indonesia
Philippines
Brunei
Timor-Leste
China (Greater Mekong Subregion contex, Yunnan & Guangxi)

Unlike globally dominant language models trained primarily on high-resource English datasets, Trans-GMS LLM focuses on underserved regional languages where healthcare communication remains difficult due to limited digital infrastructure and scarce annotated medical corpora.

Rethinking Medical AI for Southeast Asia

The project approaches multilingual healthcare AI as a regional accessibility challenge rather than a purely computational problem.

Building effective healthcare AI for Southeast Asia requires more than larger models. It requires culturally grounded language technologies capable of operating reliably across diverse linguistic and resource-constrained environments.

The discussion emphasized several major research directions:

Developing multilingual medical reasoning capabilities for ASEAN languages
Improving healthcare accessibility through localized AI systems
Supporting trustworthy AI deployment in low-resource environments
Enabling regional collaboration between ASEAN researchers and institutions

By positioning Southeast Asian languages as central design priorities rather than secondary adaptation targets, the project highlights the importance of inclusive AI development for emerging regions.

A Regional Conversation on Trustworthy AI

Beyond the technical presentation, the event reflected the growing momentum of scientific collaboration within ASEAN.

Researchers exchanged perspectives on:

Cross-border AI research collaboration
Responsible deployment of medical AI systems
Data accessibility and multilingual representation
Sustainable AI ecosystems for developing economies
The future of regional language technologies

The atmosphere emphasized a shared vision of building AI systems grounded in local realities while remaining globally relevant.

Why This Matters

Large language models are rapidly transforming healthcare and scientific research, yet many low-resource communities remain excluded due to language barriers and limited infrastructure.

Projects such as Trans-GMS LLM aim to reduce these gaps by exploring how multilingual foundation models can support:

Healthcare communication
Clinical knowledge access
Medical education
Decision-support systems
Regional healthcare collaboration within ASEAN

The initiative demonstrates how trustworthy and culturally aware AI systems can contribute to more equitable access to healthcare technologies across Southeast Asia.

Looking Forward

ASEAN Young Scientists Connect 2026 demonstrated that the future of AI in Southeast Asia will depend not only on technological innovation, but also on regional collaboration, interdisciplinary thinking, and inclusive research ecosystems.

Presenting Trans-GMS LLM at Dusit Thani Bangkok was more than a technical presentation — it was an opportunity to contribute to a broader regional conversation about the future of trustworthy, multilingual, and accessible AI for ASEAN communities.

I would like to sincerely thank my new friends in Group 4 at ASEAN Young Scientists Connect 2026 for the inspiring discussions, collaborative spirit, and thoughtful exchange of ideas throughout the event. Working together on regional AI and healthcare challenges made the experience both intellectually enriching and genuinely memorable. I truly appreciate the openness, creativity, and energy everyone brought to the group, which made our short time together feel impactful and meaningful. I am grateful to have shared this journey with such talented and forward-thinking peers across the region.

Overall, ASEAN Young Scientists Connect 2026 at Dusit Thani Bangkok was an inspiring and impactful experience that brought together diverse minds across Southeast Asia to exchange ideas on science, technology, and regional innovation. The event created a rare and valuable space for meaningful dialogue on how AI and emerging technologies can address real-world challenges, particularly in healthcare and low-resource settings. I am sincerely grateful to the organizers for their excellent coordination and for fostering such a vibrant and collaborative environment. It was an honor to be part of this gathering, and I leave with deep appreciation for the connections made, the insights shared, and the shared vision of advancing science for the benefit of the region.

GateKD: Confidence-Gated Closed-Loop Distillation for Robust Reasoning

Wed, 13 May 2026 00:00:00 +0000

GateKD (Confidence-Gated Closed-Loop Distillation) addresses a critical challenge in reasoning distillation: transferring reliable multi-step reasoning abilities from large language models into smaller student models without propagating hallucinated or noisy supervision. While conventional knowledge distillation methods treat the teacher model as a uniformly reliable oracle, real-world reasoning traces often contain uncertain intermediate steps, unstable rationales, and misleading attention patterns. GateKD reframes reasoning distillation as a confidence-aware closed-loop learning problem in which the teacher dynamically regulates how much information should be transferred during training.

The framework introduces three complementary confidence-gated mechanisms that collectively stabilize reasoning transfer. First, confidence-gated soft supervision selectively distills predictive signals only when the teacher exhibits high certainty. Second, gated hidden-state evolution aligns intermediate latent representations between teacher and student models while suppressing unreliable hidden dynamics. Third, reliability-filtered attention distillation transfers stable reasoning structures while filtering noisy or hallucinated attention maps that could otherwise corrupt student reasoning trajectories.

Together, these components establish a closed feedback loop in which teacher confidence continuously modulates supervision strength across the distillation pipeline. Unlike traditional open-loop distillation approaches, GateKD enables the student model to learn reasoning behaviors from selectively trusted intermediate states rather than blindly imitating all teacher outputs. Extensive experiments across commonsense, logical, and symbolic reasoning benchmarks demonstrate consistent improvements over strong open-loop baselines using T5 and Flan-T5 architectures of varying sizes.

Beyond empirical performance gains, GateKD highlights a broader paradigm shift toward confidence-aware reasoning supervision for trustworthy language models. The framework shows strong robustness under low-resource distillation settings, reduces hallucination transfer, and maintains stable reasoning fidelity even under noisy supervision conditions. These findings suggest that confidence-gated teacher-student interaction is a promising direction for building scalable, efficient, and reliable reasoning systems suitable for real-world deployment in resource-constrained environments.

Large Language Models | Teerapong Panboonyuen