NeuroAI, Machine Learning & Computational Imaging

Research

Research experience spanning NeuroAI, machine learning, and computational imaging.

NeuroAI & Machine Learning

Northwestern University · Evanston, IL

2023–Present

  • Formulated a mathematically tractable framework to model optimal population codes under strict metabolic resource constraints, utilizing energy-dependent dispersed Poisson noise models.
  • Solved complex constrained optimization problems (using Lagrangian methods) to predict how systems adapt and maintain robust signal transmission in noisy environments.
  • Applied concepts from information geometry and efficient coding to derive optimal information processing strategies for complex, noisy systems operating under strict resource constraints.
  • Key Publications: C12 (ICLR 2026), C14 (ICML 2026)

Computational Imaging & Inverse Problems

Northwestern University · Evanston, IL

2023–Present

  • Developed a novel computational framework to solve highly underdetermined inverse problems, reconstructing high-dimensional hyperspectral images from severely aberrated data.
  • Designed resource-efficient reconstruction algorithms utilizing Block Circulant with Circulant Blocks (BCCB) matrix inversion, significantly accelerating large-scale computational tasks.
  • Key Publications: C13 (CVPR 2026), J7 (Annual Review of Vision Science 2025)

Deep Learning & Signal Processing

University of California, Los Angeles (UCLA) · Los Angeles, CA

2020–2022

  • Implemented and trained broadband diffractive neural networks, optimizing for enhanced output signal quality and superior power efficiency.
  • Constructed wave-propagation-based neural networks to identify hidden objects and anomalies from highly degraded, single-pixel sensor data.
  • Key Publications: J3 (Nature Communications 2023), J0 (Light: Science & Applications 2022)

Computational Imaging & Hardware-Aware Computing

National Tsing Hua University · Hsinchu, Taiwan

2018–2022

  • Proposed a physics-aware computational imaging framework that models complex spatiotemporal relationships in high-frequency (terahertz) waves to reconstruct signals.
  • Developed tensor-based reconstruction algorithms (e.g., compressive sensing) optimized for FPGA deployment, bridging high-level algorithmic design with hardware-level efficiency.
  • Key Publications: J6, J1 (Optics Express), J4 (IEEE Signal Processing Magazine), J2 (IEEE OJCAS)