Papers by Yanjun Gao
Simple Yet Effective: An Information-Theoretic Approach to Multi-LLM Uncertainty Quantification (2025.emnlp-main)
Copied to clipboard
| Challenge: | Prior work on calibration and uncertainty quantification focuses on individual models, overlooking the potential of model diversity. |
| Approach: | They propose a method that uses Jensen-Shannon Divergence to identify and aggregate well-calibrated subsets of large language models (LLMs) to improve calibration. |
| Outcome: | The proposed method improves accuracy on binary prediction tasks compared to single-model and naive ensemble baselines. |
Hierarchical Annotation for Building A Suite of Clinical Natural Language Processing Tasks: Progress Note Understanding (2022.lrec-1)
Copied to clipboard
Yanjun Gao, Dmitriy Dligach, Timothy Miller, Samuel Tesch, Ryan Laffin, Matthew M. Churpek, Majid Afshar
| Challenge: | Existing corpus and annotations focus on textual features and relation prediction, but there are no structured corpus models for clinical diagnostic thinking. |
| Approach: | They propose a hierarchical annotation schema with three stages to address clinical diagnostic thinking. |
| Outcome: | The proposed model is based on a large collection of publicly available daily progress notes. |
Large Language Models with Temporal Reasoning for Longitudinal Clinical Summarization and Prediction (2025.findings-emnlp)
Copied to clipboard
Maya Kruse, Shiyue Hu, Nicholas Derby, Yifu Wu, Samantha Stonbraker, Bingsheng Yao, Dakuo Wang, Elizabeth M. Goldberg, Yanjun Gao
| Challenge: | Recent advances in large language models have shown potential in clinical text summarization, but their ability to handle long patient trajectories with multi-modal data spread across time remains underexplored. |
| Approach: | They evaluate open-source large language models, their Retrieval Augmented Generation variants and chain-of-thought prompting on long-context clinical summarization and prediction. |
| Outcome: | The proposed models can synthesize structured and unstructured EHR data while reasoning over temporal coherence. |
ABCD: A Graph Framework to Convert Complex Sentences to a Covering Set of Simple Sentences (2021.acl-long)
Copied to clipboard
| Challenge: | Existing work relies on rule-based methods dependent on parsing to identify atomic sentences. |
| Approach: | They propose a task to decompose complex sentences into simple ones . they propose atomic clauses as atomic sentences, and a graph edit task to predict edits . |
| Outcome: | The proposed model performs better than baselines on MinWiki and DeSSE. |
Summarizing Patients’ Problems from Hospital Progress Notes Using Pre-trained Sequence-to-Sequence Models (2022.coling-1)
Copied to clipboard
| Challenge: | Problem list summarization requires a model to understand, abstract, and generate clinical documentation. |
| Approach: | They propose a task that summarises patients' main problems from daily progress notes using input from the provider's progress notes during hospitalization. |
| Outcome: | The proposed model outperforms two state-of-the-art seq2seq transformer architectures in summarizing patients' main problems from daily progress notes in the medical information mart for Intensive Care (MIMIC)-III. |
When Raw Data Prevails: Are Large Language Model Embeddings Effective in Numerical Data Representation for Medical Machine Learning Applications? (2024.findings-emnlp)
Copied to clipboard
Yanjun Gao, Skatje Myers, Shan Chen, Dmitriy Dligach, Timothy Miller, Danielle Bitterman, Matthew Churpek, Majid Afshar
| Challenge: | Numerical data is pivotal for medical questions and answers, but tabular data is not fully integrated into LLMs. |
| Approach: | They examine the effectiveness of vector representations from last hidden states of LLMs for medical diagnostics and prognostics using electronic health record data. |
| Outcome: | The proposed representations outperform those using raw numerical EHR data in medical diagnostics and prognostics. |
Learning to Maximize Mutual Information for Chain-of-Thought Distillation (2024.findings-acl)
Copied to clipboard
| Challenge: | Knowledge distillation is a technique of transferring knowledge from large, complex models to smaller ones. |
| Approach: | They propose a method utilizing chain-of-thought distillation to transfer knowledge from large, complex models to smaller ones by maximizing mutual information of the representation features of the two tasks. |
| Outcome: | The proposed method outperforms the state-of-the-art knowledge distillation method on four datasets. |
QiMeng-Attention: SOTA Attention Operator is generated by SOTA Attention Algorithm (2025.findings-acl)
Copied to clipboard
Qirui Zhou, Shaohui Peng, Weiqiang Xiong, Haixin Chen, Yuanbo Wen, Haochen Li, Ling Li, Qi Guo, Yongwei Zhao, Ke Gao, Ruizhi Chen, Yanjun Wu, Zhao Chen, Yunji Chen
| Challenge: | Existing LLMs cannot comprehend the complex data flow and computation process of the attention operator and utilize low-level primitive to exploit GPU performance. |
| Approach: | They propose an LLM-friendly Thinking Language (LLM-TL) that can decouple the generation of high-level optimization logic and low-level implementation on GPU and enhance LLMs’ understanding of attention operator. |
| Outcome: | The proposed method outshines existing LLMs on A100, RTX8000, and T4 GPUs, achieving a speed-up of up to 35.16. |
LogosKG: Hardware-Optimized Scalable and Interpretable Knowledge Graph Retrieval (2026.acl-long)
Copied to clipboard
He Cheng, Yifu Wu, Saksham Khatwani, Maya Kruse, Dmitriy Dligach, Timothy A. Miller, Majid Afshar, Yanjun Gao
| Challenge: | Existing systems struggle to balance efficiency, scalability, and interpretability. |
| Approach: | They propose a hardware-aligned framework that enables scalable and interpretable k-hop retrieval on large KGs. |
| Outcome: | The proposed framework scales to billion-edge graphs without loss of retrieval fidelity. |
Anchored Answers: Unravelling Positional Bias in GPT-2’s Multiple-Choice Questions (2025.findings-acl)
Copied to clipboard
| Challenge: | Large Language Models (LLMs) exhibit a positional bias, particularly an even worse “anchored bias” in the GPT-2 family, where they consistently favour the first choice ‘A’ in MCQs. |
| Approach: | They propose to use the “logit lens” method to trace and modify the internal modules within GPT-2 models responsible for this bias. |
| Outcome: | The proposed approach mitigates the positional bias and improves the accuracy of the GPT-2 model across multiple datasets. |
PyrEval: An Automated Method for Summary Content Analysis (L18-1)
Copied to clipboard
| Challenge: | PyrEval automates manual summarization evaluation of abstractive summarizing systems . extractive summaries that select complete sentences have shifted in recent years . |
| Approach: | They propose a method for automatic summarization evaluation that automates the manual pyramid method by using pre-trained vectors and a greedy algorithm to evaluate the pyramid content. |
| Outcome: | The proposed method can be applied to human and machine summaries with no retraining and in excellent time. |