Huu Dat Do

Undergraduate Student at VinUniversity
Intern at CogAI4Sci @ NUS

prof_pic.jpg

I am an undergraduate student at VinUniversity, working closely with Professor Wray Buntine and Professor Laurent El Ghaoui. I’m working as research intern under Professor Dianbo Liu. I had the opportunity to intern at the Advanced Machine Intelligence Lab (AMI Lab) at KAIST under the supervision of Professor Tae Hyun Oh and collaborated with Professor Minsu Cho from POSTECH.

📩 Contact: 22dat.dh[at]vinuni.edu.vn
Google Scholar  



I’m driven by a simple question: How does intelligence emerge, and how can we reconstruct it computationally? Vision and language seem especially central—complex vision is widely considered a key catalyst for the Cambrian explosion, and only humans have many effective communicative language that none of other species have. More broadly, evolution (and learning) can be viewed as optimization under constraints. Concretely, my research focuses on:

  • Compositionality: How can systems learn hierarchical, compositional representations that connect low-level visual concepts to high-level abstractions, and further compose across modalities in a synchronous manner?
  • Grounding: How can we build agents that acquire language by being grounded in perception, communication, shared goals, and pluralistic values, rather than treating language as disembodied text?
  • Optimization: Can intelligence be understood as lossless compression that yields emergent sparse compositional structure, and can we formalize compositional sparsity to make learning and generalization more efficient?
  • Creativity: The ultimate capability of intelligence, emerging at the intersection of compositionality, grounding, optimization.

selected publications

  1. ICCV 2025
    VSC: Visual Search Compositional Text-to-Image Diffusion Model
    Do Huu Dat, Nam Hyeonu, Po Yuan Mao, and 1 more author
    In International Conference on Computer Vision, 2025
  2. IEEE/CVF WACV 2025
    HOPE: A Memory-Based and Composition-Aware Framework for Zero-Shot Learning with Hopfield Network and Soft Mixture of Experts
    Do Huu Dat, Po Yuan Mao, Tien Hoang Nguyen, and 2 more authors
    In Proceedings of the 2025 IEEE/CVF Winter Conference on Applications of Computer Vision, 2025
  3. NAACL 2025
    Discrete Diffusion Language Model for Long Text Summarization
    Do Huu Dat, Do Duc Anh, Anh Tuan Luu, and 1 more author
    In Annual Conference of the Nations of the Americas Chapter of the Association for Computational Linguistics, 2025