Zihan Liang
  • Home
  • CV
  • Research
  • Experiences
  • Demos
  • Study Notes

Publications

Two-Sided State-Space Models for Sequential Recommendation with Non-Random Multimodal Review Feedback

Ziwen Pan*, Zihan Liang*, Ruoxuan Xiong

Accepted at EMNLP 2026 Findings

  • TS-SSM jointly tracks evolving user preferences and item states, treating multimodal reviews—and how they are observed and expressed—as non-random feedback, while capturing cross-item spillovers and the asymmetric persistence of positive and negative reviews.
  • Across six Amazon datasets, TS-SSM outperforms strong sequential, multimodal, and debiasing baselines, achieving 14.8%–18.8% relative improvements in Recall@20 over the strongest sequential baseline.

DART: Mitigating Harm Drift in Difference-Aware LLMs via Distill-Audit-Repair Training

Ziwen Pan*, Zihan Liang*, Jad Kabbara, Ali Emami

Accepted at ACL 2026 Findings

Paper Code
  • Identified the problem of harm drift in safety-tuned LLMs—where fine-tuning improves decision accuracy but makes rationales more harmful—and introduced DART, a Distill-Audit-Repair pipeline for improving difference-awareness classification while preserving safer explanations.
  • On eight benchmarks, DART improved Llama-3-8B-Instruct accuracy from 39.0% to 68.8%, boosted equal-treatment accuracy from 11.3% to 72.6%, and reduced harm drift cases by 72.6%, showing that accuracy and safety can be improved together.

Learning Dynamic Representations and Policies from Multimodal Clinical Time-Series with Informative Missingness

Zihan Liang*, Ziwen Pan*, Ruoxuan Xiong

Accepted at ACL 2026 Findings

Paper Code
  • Proposed an MNAR-aware multimodal framework for clinical time-series that treats observation patterns as signal rather than noise, combining explicit missingness features, adaptive fusion of sparse clinical text, and action-conditioned latent dynamics to learn patient states for both prediction and offline decision-making.
  • Demonstrated strong results on MIMIC-IV and eICU: mortality prediction reached AUROC 0.876 on MIMIC-IV (+3.9% over GRU-D), while the learned treatment policy improved FQE by 20.3% over clinician behavior and delivered the largest gains for high-severity patients.

MambaDATG: Domain-Adaptive Tri-Plane-Gated Pre-training for 3D Abdominal Segmentation

Yaomin Shen*, Dongming Jiang*, Zihan Liang*, Yangbo Wei*, Wenkai Yang, Xiaoxin Sun, Zhen Huang, Suhua Wang, Qingsong Yao

Accepted at ICASSP 2026 Oral

Paper
  • Proposed MambaDATG, a tri-plane-gated selective SSM pre-training that fuses axial, coronal, and sagittal scans via voxel gating—capturing directional anisotropy with linear-time efficiency and zero inference overhead.
  • Introduced a domain-adaptive MIM stage adapting high-level layers on unlabeled CTs, improving small-organ segmentation (e.g., +1.8 Dice on BTCV) and revealing the tri-plane gate as the key performance driver.

Causal Representation Learning from Multimodal Clinical Records under Non-Random Modality Missingness

Zihan Liang*, Ziwen Pan*, Ruoxuan Xiong

Accepted at EMNLP 2025 Main

Paper Code
  • Proposed CRL-MMNAR, a causal multimodal framework that treats modality missing-not-at-random as signal—combining missingness-aware fusion (gated by observation patterns) with cross-modal reconstruction + contrastive learning, and a multitask predictor with a cross-fitted rectifier to correct observation-pattern bias.
  • Demonstrated consistent gains on MIMIC-IV and eICU, including AUC 0.8687→0.9824 for ICU admission (+13.1%) and 0.7989→0.8657 for readmission (+8.4%) on MIMIC-IV, and readmission AUC to 0.9294 (+13.8%) on eICU, with lower Brier scores indicating better calibration.

CareLab at #SMM4H-HeaRD 2025: Insomnia Detection and Food Safety Event Extraction with Domain-Aware Transformers

Zihan Liang*, Ziwen Pan*, Sumon Kanti Dey, Azra Ismail

Accepted at AAAI ICSWM 2025

Paper Code
  • Ranked 1st on SMM4H–HeaRD 2025 Task 5, developing a RoBERTa + GPT-4–augmented system with classweighted training and ensembling, achieving F1 = 0.958 and revealing the limits of rule-based span extraction in clinical and food-safety texts.

Dynamic Policy Design for Autonomous Taxi Adoption: A Hierarchical Game-Theoretic Framework

Zihan Liang, Ziwen Pan

Accepted at TRB Annual Meeting 2026

Paper
  • Formulated a three-tier dynamic Stackelberg game modeling interactions among government, platform, and workers in autonomous taxi adoption, deriving equilibrium policies that balance innovation incentives with social welfare.
  • An extended version is under review at Transportation Research Record.

Manuscripts

Adaptive Spatiotemporal Graph Neural Networks with Trend-Aware Prediction and Validation-Gated Calibration for Traffic Flow Forecasting

Zihan Liang*, Ziwen Pan*, Shuyang Yu | May 2025 - Present

Under Review at IEEE Transactions on Intelligent Transportation Systems

  • Proposed an adaptive spatiotemporal GNN that fuses distance- and correlation-based graphs with dual temporal–spatial encoders and a trend-aware head, achieving state-of-the-art accuracy on PeMS benchmarks.
  • Designed a validation-gated calibration pipeline with hour-conditional quantile mapping and AR(1) residual correction, applied only when improving MAPE/MAE to ensure reliable and interpretable post-processing.

Theses and Dissertations

Causal Representation Learning under Informative Missingness for Clinical Multimodal Prediction and Offline Decision-Making

Zihan Liang | Advised by Prof. Ruoxuan Xiong

Emory College of Arts and Sciences Honors Thesis (Highest Honors)

Paper Slides Poster
  • Proposed a unified causal representation learning framework that models cross-scale informative missingness (patient-level modality assignment and step-level monitoring intensity) as explicit variables, enabling both prediction and decision-making under MNAR clinical data.
  • Developed a two-stage MMNAR pipeline with missingness-aware fusion, cross-modal self-supervision, and a cross-fitted pattern-wise rectifier that corrects residual bias, achieving consistent AUROC gains under severe missingness and distribution shift.
  • Introduced MNAR-aware dynamic state learning with action-conditioned latent dynamics and process-aware text modeling, enabling reliable offline policy optimization (IQL) and significant improvements over clinician policies and strong RL baselines.

Collaborative Research Projects

Research Assistant at Collective Action & Research for Equity (CARE) Lab, Emory University

Sept. 2024 - Present

  • Analyzed nearly 1,000 user data points and developed an interactive dashboard for MakerGhat, an India-based organization, utilizing JavaScript, Python, and HTML for seamless data visualization.
  • Maintaining a WhatsApp Education Bot using Python and Twilio for MakerGhat, enabling over 15,000 teachers across India to document lesson plans and teaching activities, impacting 600,000+ students and supporting 10,000+ educational projects.
  • Developed an offline-first Android app for classroom session recording and metadata management, integrating Google Sheets validation, offline caching, and secure Drive synchronization.

Research Assistant at Polymath Jr REU Summer Program

Jun. 2025 - Aug. 2025

  • Developed and analyzed mathematical models using ordinary differential equations (ODEs) to study the community transmission of Clostridioides difficile (C. difficile).
  • Collaborated with a team to identify optimal intervention strategies for controlling the spread of C. difficile in community settings.
  • Assisted in modifying and implementing code in MATLAB and R to simulate disease dynamics and evaluate mitigation strategies.

* These authors contributed equally to this work.

 

Computational Biology and Bioinformatics Program, Duke University | zihan.liang@duke.edu
中文 | © 2026 Zihan Liang. All rights reserved.