Papers by Hui Wan
Safety Sidecar: Reflection-Driven Runtime Control for Safer Agents (2026.findings-acl)
Copied to clipboard
| Challenge: | Existing safety controls fail to provide runtime intervention or cross-architecture portability for autonomous LLM agents. |
| Approach: | They propose a model-agnostic, plug-and-play module to provide arbitrary agent safety control and auditability. |
| Outcome: | The proposed module improves the secure-solution rate by 2.9–11.2 percentage points . it adds only 3.2s to end-to-end latency and a negligible average cost of 5.37 10-4 per scenario . |
MultiDoc2Dial: Modeling Dialogues Grounded in Multiple Documents (2021.emnlp-main)
Copied to clipboard
| Challenge: | Existing work treats document-grounded dialogue modeling as a machine reading comprehension task based on a single document or passage. |
| Approach: | They propose a task and dataset for modeling goal-oriented dialogues grounded in multiple documents. |
| Outcome: | The proposed task and dataset address realistic scenarios where goal-oriented dialogues involve multiple topics and hence are grounded on different documents. |
Rewarding Smatch: Transition-Based AMR Parsing with Reinforcement Learning (P19-1)
Copied to clipboard
| Challenge: | Abstract meaning representations (AMRs) are labeled directed acyclic graphs that represent a non intersentential abstraction of natural language with broad-coverage semantic representations. |
| Approach: | They build upon a transition-based AMR parser that uses Stack-LSTMs and augment training with policy learning. |
| Outcome: | The proposed parser performs comparable to the best published parsers. |
INS: An Interactive Chinese News Synthesis System (N19-4)
Copied to clipboard
| Challenge: | In the last decade, news websites and apps become more popular, which can provide us an extremely large volume of news articles. |
| Approach: | They propose a system which automatically synthesizes news articles into a long overview article by interacting with users. |
| Outcome: | The proposed system can generate news overview articles automatically or by interacting with users. |
doc2dial: A Goal-Oriented Document-Grounded Dialogue Dataset (2020.emnlp-main)
Copied to clipboard
| Challenge: | doc2dial dataset is a goal-oriented document-grounded dialogue model . it is based on how the authors compose documents for guiding end users . |
| Approach: | They propose a dataset of goal-oriented dialogues grounded in documents . they use annotated conversations with an average of 14 turns to generate conversational utterances . |
| Outcome: | The proposed dataset includes over 4500 annotated conversations with an average of 14 turns grounded in over 450 documents from four domains. |
Jointly Learning to Align and Summarize for Neural Cross-Lingual Summarization (2020.acl-main)
Copied to clipboard
| Challenge: | Existing studies on cross-lingual summarization focus on pipeline methods and training end-to-end models. |
| Approach: | They propose to jointly learn to align and align to train a neural cross-lingual summarization model by using a large-scale corpus. |
| Outcome: | The proposed model outperforms competing models in most cases and can generate cross-lingual summaries without access to any cross-linguistic corpus. |
Fast and Light-Weight Answer Text Retrieval in Dialogue Systems (2022.naacl-industry)
Copied to clipboard
| Challenge: | a recent study shows that text retrieval can be used to find information relevant to user requests. |
| Approach: | They propose to use a corpus of text to search for relevant responses to user requests . they compare this approach to other methods that use intent detection . |
| Outcome: | a new approach can be used to search through a corpus of text to find relevant responses to user requests. |
Models See Hallucinations: Evaluating the Factuality in Video Captioning (2023.emnlp-main)
Copied to clipboard
| Challenge: | Existing evaluation metrics show little correlation with human factuality annotation. |
| Approach: | They propose a weakly-supervised, model-based factuality metric FactVC which outperforms previous metrics on factual evaluation of video captioning. |
| Outcome: | The proposed model outperforms previous metrics on factuality evaluation of video captioning. |
MEIT: Multimodal Electrocardiogram Instruction Tuning on Large Language Models for Report Generation (2025.findings-acl)
Copied to clipboard
Zhongwei Wan, Che Liu, Xin Wang, Chaofan Tao, Hui Shen, Jing Xiong, Rossella Arcucci, Huaxiu Yao, Mi Zhang
| Challenge: | Recent studies have focused on classifying cardiac conditions using ECG data but have overlooked ECG report generation, which is time-consuming and requires clinical expertise. |
| Approach: | They propose a Multimodal ECG Instruction Tuning framework that extends the capability of large language models (LLMs) for the task. |
| Outcome: | The proposed framework outperforms open-source LLMs and LLM backbones across two large-scale ECG datasets. |
Argus: Benchmarking and Enhancing Vision-Language Models for 3D Radiology Report Generation (2025.findings-acl)
Copied to clipboard
| Challenge: | Existing work on 3D radiograph report generation focuses on 2D images, but 3D medical images provide more comprehensive diagnostic information. |
| Approach: | They propose a comprehensive training recipe for building high-performing VLMs for 3DRRG using a publicly available 3D CT-report dataset. |
| Outcome: | The proposed model achieves superior performance across different model sizes and input 3D medical image resolutions. |
MEDA: Dynamic KV Cache Allocation for Efficient Multimodal Long-Context Inference (2025.naacl-long)
Copied to clipboard
| Challenge: | Long-context Multimodal Large Language Models (MLLMs) require substantial computational resources as their multimodal Key-Value (KV) cache grows with increasing input lengths, challenging memory and time efficiency. |
| Approach: | They propose a dynamic multimodal KV cache allocation strategy that dynamically allocating KV size based on attention entropy to better adapt to multimodal interactions. |
| Outcome: | The proposed model achieves up to 72% KV cache memory reduction and 2.82 faster decoding speeds while maintaining or enhancing performance on various multimodal tasks in a long context. |
Does Structure Matter? Encoding Documents for Machine Reading Comprehension (2021.naacl-main)
Copied to clipboard
| Challenge: | Existing Transformer-based models for machine reading comprehension treat documents as flat sequences. |
| Approach: | They propose a Transformer-based method that reads a document as tree slices and jointly trains and consults the modules at inference time. |
| Outcome: | The proposed method outperforms several baseline approaches on two datasets from varied domains. |
Video Paragraph Captioning as a Text Summarization Task (2021.acl-short)
Copied to clipboard
| Challenge: | Existing methods for video paragraph captioning use ground-truth event segments. |
| Approach: | They propose a video paragraph captioning task that generates coherent paragraphs without ground-truth event segments. |
| Outcome: | The proposed framework outperforms existing methods on two popular datasets. |
Semi-Structured Object Sequence Encoders (2023.findings-emnlp)
Copied to clipboard
Rudra Murthy, Riyaz Bhat, Chulaka Gunasekara, Siva Patel, Hui Wan, Tejas Dhamecha, Danish Contractor, Marina Danilevsky
| Challenge: | Semi-structured object sequences are often represented as a sequence of key-value pairs over time . authors propose a two-part approach that takes each key independently and encodes a representation of its values over time. |
| Approach: | They propose a two-part approach that first considers each key independently and encodes a representation of its values over time. |
| Outcome: | The proposed approach outperforms existing methods on multiple prediction tasks using real-world data. |
Explaining Neural Network Predictions on Sentence Pairs via Learning Word-Group Masks (2021.naacl-main)
Copied to clipboard
| Challenge: | Existing methods to explain neural network models are computationally inefficient for text inputs. |
| Approach: | They propose a method to implicitly detect word correlations by grouping correlated words from input text pairs together and measuring their contribution to corresponding NLP tasks. |
| Outcome: | The proposed method is evaluated with two different model architectures across four datasets. |
SVD-LLM V2: Optimizing Singular Value Truncation for Large Language Model Compression (2025.naacl-long)
Copied to clipboard
| Challenge: | Existing methods for compressing Large Language Models suffer from significant truncation losses. |
| Approach: | They propose a novel method that optimizes singular value truncation in SVD compression . they use dynamic compression ratio allocation to balance the large tuncation loss . |
| Outcome: | The proposed method outperforms current state-of-the-art methods on ten datasets and five models on various scales. |
Learning as Conversation: Dialogue Systems Reinforced for Information Acquisition (2022.naacl-main)
Copied to clipboard
| Challenge: | a novel AI-empowered chat bot for learning as conversation can be applied to various domains without in-domain dialogue data. |
| Approach: | They propose a novel task where a user does not read a passage but gains information and knowledge through conversation with a teacher bot. |
| Outcome: | The proposed system can be transferred to various domains without in-domain dialogue data and can carry out conversations both informative and attentive to users. |
GenTool: Enhancing Tool Generalization in Language Models through Zero-to-One and Weak-to-Strong Simulation (2025.findings-acl)
Copied to clipboard
Jie He, Jennifer Neville, Mengting Wan, Longqi Yang, Hui Liu, Xiaofeng Xu, Xia Song, Jeff Z. Pan, Pei Zhou
| Challenge: | Large Language Models (LLMs) can expand their capabilities by integrating external tools. |
| Approach: | They propose a training framework that prepares LLMs for diverse generalization challenges in tool utilization. |
| Outcome: | The proposed framework improves the tool-usage capabilities of LLMs by up to 8B parameters, surpassing GPT-4o. |