I am Thanh Tran, an AI Resident / AI Research Engineer at the FPT Software AI Center, working under the supervision of Dr. Van Nguyen and Prof. Truong-Son Hy. Previously, I earned my B.Sc. degree in Computer Science (Honors) from the VNU University of Engineering and Technology. I spent my sophomore year interning at VinBigData, where I worked on Automatic Speech Recognition.
My research interests lie in the integration of Graph Neural Networks (GNNs) and LLMs to solve diverse applications, thereby allowing pretrained LLMs to understand and make use of relational information. In parallel, I am also passionate about audio-visual learning.
My attached CV (last updated: 2025 Dec).
🔥 News
- 2025.05: 🎉 One paper is accepted at Interspeech 2025!
- 2025.03: 🎉 One paper is accepted at Machine Learning: Science and Technology!
- 2024.12: 🎉 One paper is accepted at ICASSP 2025!
- 2024.11: 🎉 One paper is accepted at KDD 2025!
- 2024.10: 🎉 One paper is accepted at Machine Learning in Structural Biology (MLSB) Workshop in NeurIPS 2024!
- 2024.08: 🎉 One paper is accepted at IEEE Transactions on Evolutionary Computation!
- 2023.08: I join FPT Software AI Center
as an AI Resident in Vietnam! - 2021.10: I join VinBigData
as a speech research intern in Vietnam!
📝 Publications
Effective Context Modeling Framework for Emotion Recognition in Conversations
Cuong Tran Van*, Thanh V. T. Tran*, Van Nguyen, Truong Son Hy
- We design a GNN framework modeling both multiscale and multivariate interactions among modalities and utterances within conversations.
- We address class imbalance with a re-weighting scheme in the loss function.
GROOT: Effective Design of Biological Sequences with Limited Experimental Data
Thanh V. T. Tran*, Nhat Khang Ngo*, Viet Anh Nguyen, Truong Son Hy
[Paper] [Code] [Poster] [Slides]
- We introduce a novel framework using graph-based smoothing to train a surrogate model, which is then used in the optimization process.
- We theoretically and empirically show that our technique can expand into extrapolation regions while keeping a reasonable distance from the training data.
- Our method can be applied on diverse tasks of different domains.
- RESOUND: Speech Reconstruction from Silent Videos via Acoustic-Semantic Decomposed Modeling, Long-Khanh Pham, Thanh V. T. Tran, Minh-Tan Pham, Van Nguyen | [Interspeech 2025] [Project page]
- LatentDE: Latent-based Directed Evolution for Protein Sequence Design, Thanh V. T. Tran, Nhat Khang Ngo, Viet Thanh Duy Nguyen, Truong Son Hy | [Machine Learning: Science and Technology] [Code]
- Protein Design by Directed Evolution guided by Large Language Models, Thanh V. T. Tran, Truong Son Hy | [IEEE Transactions on Evolutionary Computation] [Code]
🎖 Honors and Awards
- Merit Scholarship (6 semesters), VNU University of Engineering and Technology
- Second prize in ASR Task 1, VLSP 2022 (Certificate)
- Third prize in ASR Task 2, VLSP 2022 (Certificate)
- Third prize in Scientific Research Contest, VNU University of Engineering and Technology (Certificate)
📖 Educations
- 2019.08 - 2023.06, B.Sc. in Computer Science (Honors), VNU University of Engineering and Technology
💻 Industry Experience
- 2023.08 - now, AI Resident & AI Research Engineer, FPT Software AI Center, Hanoi, Vietnam.
- 2021.10 - 2022.11, Spoken Language Processing Department, VinBigData, Hanoi, Vietnam.