Deqian Kong
Contact: deqiankong AT ucla.edu
Contact: deqiankong AT ucla.edu
University of California, Los Angeles, 2019 -
I'm a PhD candidate in Department of Statistics and Data Science, UCLA, advised by Prof. Ying Nian Wu.
Currently, I am particularly interested in generative models and representation learning. My research aims to (1) design efficient and generalizable top-down generative models that enable explicit abstraction and compression, and (2) learn representations that support reasoning and decision making through explicit gradient-based planning algorithms.
Selected Publications/Preprints
Place Cells as Proximity-Preserving Embeddings: From Multi-Scale Random Walk to Straight-Forward Path Planning
M. Zhao*, D. Xu*, D. Kong*, W.-H. Zhang, Y. N. Wu
Latent Adaptive Planner for Dynamic Manipulation
D. Noh*, D. Kong*, M. Zhao, A. Lizarraga, J. Xie, Y. N. Wu^, D. Hong^
Latent Thought Models with Variational Bayes Inference-Time Computation
D. Kong*, M. Zhao*, D. Xu*, B. Pang, S. Wang, E. Honig, Z. Si, C. Li, J. Xie^, S. Xie^, Y. N. Wu^
International Conference on Machine Learning (ICML 2025) | Blog |
Latent Space Energy-based Neural ODEs
S. Cheng*, D. Kong*, J. Xie, K. Lee, Y. N. Wu^, Y. Yang^
Latent Plan Transformer for Trajectory Abstraction: Planning as Latent Space Inference
D. Kong*, D. Xu*, M. Zhao*, B. Pang, J. Xie, A. Lizarraga, Y. Huang, S. Xie*, Y. N. Wu
Neural Information Processing Systems (NeurIPS) 2024 | Project Page
Molecule Design by Latent Prompt Transformer
D. Kong*, Y. Huang*, J. Xie*, E. Honig*, M. Xu, S. Xue, P. Lin, S. Zhou, S. Zhong, N. Zheng, Y. N. Wu
Neural Information Processing Systems (NeurIPS) 2024 (Spotlight) | Project Page | An extended verision to previous NeurIPS 2023 AI for Science Workshop Paper.
Long-term social interaction context: The key to egocentric addressee detection
D. Kong, F. Khan, X. Zhang, P. Singhal, Y. N. Wu
ICASSP 2024 (Work done during the internship at Amazon AGI)
Molecule Design by Latent Space Energy-based Modeling and Gradual Distribution Shifting
D. Kong*, B. Pang*, T. Han, Y. N. Wu
Diverse and Faithful Knowledge-grounded Dialogue Generation via Sequential Posterior Inference
Y. Xu*, D. Kong*, D. Xu, Z. Ji, B. Pang, P. Fung, Y. N. Wu
International Conference on Machine Learning (ICML 2023) | Code
A statistical approach to topological entanglement: Boltzmann machine representation of higher-order irreducible correlation
S. Feng, D. Kong, N. Trivedi
Unsupervised Meta-learning via Latent Space Energy-based Model of Symbol Vector Coupling
D. Kong*, B. Pang* and Y. N. Wu
YouRefIt: Embodied Reference Understanding with Language and Gesture
Y. Chen, Q. Li, D. Kong, Y. L. Kei, T. Gao, Y. Zhu, S.-C. Zhu, S. Huang
International Conference on Computer Vision (ICCV) 2021 (Oral)