👋 A Brief Introduction

I am currently pursuing a Ph.D. degree at the Hong Kong University of Science and Technology (HKUST), now in my third year, supervised by @Prof. Guang Zhang from HKUST-GZ and @Dr. Zhong Li from Microsoft Research Asia (MSRA).

My research interests focus on Data-Centric AI, generative model theory, and interpretability. I am also interested in applied research that can generate practical value, such as Fintech and other domains.

My work has been published at venues including ACL 2023, ACL 2025, ICAIF, and ICASSP 2026, with additional manuscripts under review at ICLR and ACL Rolling Review (ARR). I also serve as a reviewer for leading conferences such as NeurIPS, ICLR, and AAAI.

🔥 News

• 2026.01:  🎉 Paper accepted at ICASSP 2026!
• 2025.10:  😊 Two papers received all positive reviews in the October ACL ARR cycle!
• 2025:  🎉🎉 Two papers accepted at ACL 2025 and ICAIF 2025!
• 2023:  🎉🎉 Paper accepted at ACL 2023!

📝 Selected Papers

• Yuanjian Xu and Guang Zhang. *HGAN-SDEs: Learning Neural Stochastic Differential Equations with Hermite-Guided Adversarial Training [ Paper] [ Code]. ICASSP 2026 [CCF B, CORE A]
  ✓ Key Contribution: We introduce HGAN-SDEs, leveraging Neural Hermite functions to construct an expressive yet lightweight discriminator for Neural SDEs, achieving reduced computational complexity and improved training stability with theoretical guarantees.
• Yuanjian Xu and Guang Zhang. *FinRipple: Aligning Large Language Models with Financial Market for Event Ripple Effect Awareness [ Paper] [ Code]. ACL 2025 [CCF A, CORE A*]
  ✓ Key Contribution: We propose FinRipple to align LLMs with financial markets by modeling event ripple effects, enabling better understanding of how financial events propagate and impact market dynamics.
• Yuanjian Xu and Guang Zhang. *LENS: Large Pre-trained Transformer for Exploring Financial Time Series Regularities [ Paper] [ Code]. ICAIF 2025 (Leading conference for AI in Finance)
  ✓ Key Contribution: We present LENS, a large-scale pre-trained Transformer that captures inherent regularities in financial time series, achieving superior performance in financial forecasting tasks.
• Yuanjian Xu and Zaiqing Nie. *Hard Sample Aware Prompt-Tuning [ Paper] [ Code]. ACL 2023 [CCF A, CORE A*]
  ✓ Key Contribution: We introduce a hard sample aware mechanism for prompt-tuning that dynamically adjusts learning focus on difficult samples, improving model performance on challenging instances.
• Jianing Hao†, Yuhe Wu†, Yuanjian Xu† and Guang Zhang. *BizCompass: Benchmarking the Reasoning Capabilities of LLMs in Business Knowledge and Applications. ACL ARR (Oct 2025) [CCF A, CORE A*] ✓ All Positive Reviews († Equal Contribution)
  🔑 Key Idea: We introduce BizCompass, a dual-axis benchmark connecting theoretical foundations (finance, economics, statistics, OR) with practical business applications (analyst, trader, consultant), revealing how theoretical knowledge translates into real-world business performance.
• Yuanjian Xu and Guang Zhang. *Rethinking Data Mixing from the Perspective of Large Language Model. ACL ARR (Oct 2025) [CCF A, CORE A*] ✓ All Positive Reviews
  🔑 Key Idea: We establish formal connections between gradient dynamics and domain distributions, and introduce DoGraph, a graph-constrained optimization framework for data mixing that clarifies how domain weighting influences LLM generalization.

📌 Note: Papers with ✓ All Positive Reviews badges are currently under review and have received positive feedback from all reviewers, indicating a high likelihood of acceptance.