\(|\)Morris\(\rangle\) = \( {\alpha} |\)AI\(\rangle\) + \( {\beta} |\)Science\(\rangle\)

Hey, my name is Morris (Yu-Chao) Huang (黃禹超 in Chinese). I have recently graduated with a Master’s in Physics under the guidance of Prof. Hsi-Sheng Goan at National Taiwan University. I am currently working as research assistant at National Center for Theoretical Sciences.

The equation \(|\)Morris\(\rangle\) = \( {\alpha} |\)AI\(\rangle\) + \( {\beta} |\)Science\(\rangle\) reflects my superposition of research passion at the intersection of artificial intelligence and science. My research interests lie in three main directions:

  1. Theoretical understanding of the computational and statistical properties of deep learning models.
  2. Developing methodologies with theoretical guarantees to ensure both optimality and practical applicability.
  3. Tackling interdisciplinary scientific problems from the machine learning perspective — physics-based modeling, quantum computing.

I am actively seeking PhD opportunities for Fall 2025.

Somewhere, something incredible is waiting to be known.
— Carl Sagan

Machine Learning Demo - Hopfield Networks

The retrieve pattern is ... :drum:
(Steps: 0/3000)

The Nobel Prize in Physics 2024 is awarded to John Hopfield and Geoffrey Hinton! (see here) The Hopfield network is inspired by the Ising model. Hopfield network acts as a dynamic energy system where neurons interact to reach stable, low-energy states, similar to particles finding equilibrium in a physical system. The energy function is given by \[ E = -\frac{1}{2} \sum_{i \neq j} W_{ij} s_i s_j - \sum_{i} b_i s_i, \] where neuron interactions mimic energy exchanges, guiding the network to "remember" stored patterns (see a nice blog post). :brain: :sparkles: Check out our paper on utilizing modern Hopfield networks for tabular learning [ICML'24].

Physics Demo - Two Stream Instability

Beam Width ( \(1 / V_b \) )

The phenomenon of the two-stream instability occurs where plasma flows are moving in opposite directions. This instability arises from the transfer of energy between particles in the plasma and EM wave, leading to the exponential growth of certain wave modes, described by the dispersion relation. \[ 1 = \dfrac{\omega_p^2/2}{(\omega+\omega_D)^2} + \dfrac{\omega_p^2/2}{(\omega-\omega_D)^2}. \] (See blog post for more detail :smile:)

selected publications

  1. EMNLP
    Two Tales of Persona in LLMs: A Survey of Role-Playing and Personalization
    Yu-Min Tseng*, Yu-Chao Huang*, Teng-Yun Hsiao*, Yu-Ching Hsu, and 3 more authors
    Findings of the Association for Computational Linguistics: EMNLP 2024, 2024
    * These authors contributed equally to this work
  2. ICML
    BiSHop: Bi-Directional Cellular Learning for Tabular Data with Generalized Sparse Modern Hopfield Model
    Chenwei Xu*, Yu-Chao Huang*, Jerry Yao-Chieh Hu*, Weijian Li, and 3 more authors
    International Conference on Machine Learning (ICML), 2024
    * These authors contributed equally to this work
  3. Under Review
    L2O-g†: Learning to Optimize Parameterized Quantum Circuits with Fubini-Study Metric Tensor
    Yu-Chao Huang, and Hsi-Sheng Goan
    arXiv preprint arXiv:2407.14761, 2024
  4. ArXiv
    Test-Time Training with Quantum Auto-Encoder: From Distribution Shift to Noisy Quantum Circuits
    Damien Jian*, Yu-Chao Huang*, and Hsi-Sheng Goan
    arXiv preprint arXiv:2411.06828, 2024
    * These authors contributed equally to this work