Research Interests

Hi, I am a senior researcher in the Shanghai AI/ML group of Microsoft Research Asia.

I am interested in neural networks, both artificial ones and biological ones, as well as the principles underlying decision making. In particular, my ultimate research objective is to understand the computational mechanisms of the brain and create machines as intelligent as humans.

Research directions

• Computational neuroscience / cognitive science
• Spiking neural networks
• Deep reinforcement learning
• Brain-inspired artificial intelligence
• Embodied artificial intelligence
• Brain-computer interface

Education

University of Science and Technology of China

B.S. in Theoretical Physics (Supervisor: Dr. Jinlin Xie)
Hefei, China, Sep. 2012 - Jun. 2016

Okinawa Institute of Science and Technology

Ph.D. (Supervisor: Dr. Jun Tani and Dr. Kenji Doya)
Okinawa, Japan, Sep. 2016 - Sep. 2022

Selected Publications

See my Google Scholar for the full list.

• Synergizing habits and goals with variational Bayes
  D. Han, K. Doya, D. Li, J. Tani
  Nature Communications 15 (1), 4461, 2024
  Link

• Advancing Spiking Neural Networks for Sequential Modeling with Central Pattern Generators
  C. Lv, D. Han, Y. Wang, X. Zheng, X. Huang, D. Li
  NeurIPS, 2024 (Spotlight)
  Link

• Understanding Training-free Diffusion Guidance: Mechanisms and Limitations
  Y Shen, X Jiang, Y Wang, Y Yang, D Han, D Li
  NeurIPS, 2024
  Link

• Efficient and Effective Time-Series Forecasting with Spiking Neural Networks
  C. Lv, D. Han, Y. Wang, X. Zheng, X. Huang, D. Li
  ICML, 2024
  Link

• Addressing Signal Delay in Deep Reinforcement Learning
  W. Wang, D. Han, X. Luo, D. Li
  ICLR, 2023 (Spotlight)
  Link

• Toward Open-ended Embodied Tasks Solving
  W. Wang, D. Han, X. Luo, Y. Shen, C. Ling, B. Wang, D. Li
  Second Agent Learning in Open-Endedness Workshop, NeurIPS, 2023
  Link

• Variational oracle guiding for reinforcement learning
  D. Han, T. Kozuno, X. Luo, Z.Y. Chen, K. Doya, Y. Yang, D. Li
  ICLR, 2022
  Link

• Variational Recurrent Models for Solving Partially Observable Control Tasks
  D. Han, K. Doya, J. Tani
  ICLR, 2020
  Link

• Self-organization of action hierarchy and compositionality by reinforcement learning with recurrent neural networks
  D. Han, K. Doya, J. Tani
  Neural Networks 129, 149-162, 2020
  Link

• Lamina-specific neuronal properties promote robust, stable signal propagation in feedforward networks
  D. Han, E. De Schutter, S. Hong
  NeurIPS, 2020
  Link

• In situ relative self-dependent calibration of electron cyclotron emission imaging via shape matching
  D. Han, J. Xie, A. Hussain, B. Gao, C. Qu, W. Liao, X. Xu, F. Gao, H. Li, T. Lan
  Review of Scientific Instruments 89 (10), 2018
  Link