Papers by Cheng Wan
Knowledge-enhanced Multimodal ECG Representation Learning with Arbitrary-Lead Inputs (2025.findings-emnlp)
Copied to clipboard
| Challenge: | Current methods for multimodal representation learning for electrocardiograms often result in suboptimal alignment of ECG signals with their corresponding text reports. |
| Approach: | They propose a framework to learn ECG representations by aligning ECG signals with paired free-text reports. |
| Outcome: | The proposed framework outperforms existing methods in zero-shot classification and linear probing tasks using 12 leads. |
Gated Differentiable Working Memory for Long-Context Language Modeling (2026.acl-long)
Copied to clipboard
Lingrui Mei, Shenghua Liu, Yiwei Wang, Yuyao Ge, Baolong Bi, Jiayu Yao, Jun Wan, Ziling Yin, Jiafeng Guo, Xueqi Cheng
| Challenge: | Long contexts break transformers, attention scores dilute, model cannot adapt to novel patterns at inference time. |
| Approach: | They propose a framework that gates the memory consolidation process by estimating Contextual Utility . they propose GDWM to maintain a form of working memory with constant contexts . |
| Outcome: | The proposed framework achieves comparable or superior performance on sparse-information tasks with 4 fewer gradient steps compared to uniform baselines. |
Atoxia: Red-teaming Large Language Models with Target Toxic Answers (2025.findings-naacl)
Copied to clipboard
| Challenge: | Large language models (LLMs) are still vulnerable to generation safety vulnerabilities. |
| Approach: | They propose a method that A**tacks LLMs with target "toxi" given a particular harmful answer, the method generates a user query and a misleading answer opening to examine the internal defects of a given LLM. |
| Outcome: | The proposed method detects safety risks in open-source models and state-of-the-art models such as GPT-4o. |
ToolBeHonest: A Multi-level Hallucination Diagnostic Benchmark for Tool-Augmented Large Language Models (2024.emnlp-main)
Copied to clipboard
Yuxiang Zhang, Jing Chen, Junjie Wang, Yaxin Liu, Cheng Yang, Chufan Shi, Xinyu Zhu, Zihao Lin, Hanwen Wan, Yujiu Yang, Tetsuya Sakai, Tian Feng, Hayato Yamana
| Challenge: | Currently, tool-augmented large language models (LLMs) only achieve total scores of 45.3 and 37.0, respectively, on a scale of 100. |
| Approach: | They propose a multi-level diagnostic process to assess the LLM's hallucinations through two perspectives: depth and breadth. |
| Outcome: | The proposed diagnostic process assesses the hallucinations of large language models through two perspectives: depth and breadth. |
Reformulating Domain Adaptation of Large Language Models as Adapt-Retrieve-Revise: A Case Study on Chinese Legal Domain (2024.findings-acl)
Copied to clipboard
| Challenge: | Recent large language models like GPT-4 have demonstrated astonishing zero-shot capabilities in general domain tasks, but they often generate content with hallucinations in specific domains such as Chinese law. |
| Approach: | They propose a framework for adapting large language models (LLMs) to Chinese legal domains by reformulating generation as an adapt-retrieve-revise process. |
| Outcome: | The proposed framework outperforms existing models in the Chinese legal domain by +33.6 points in the zero-shot setting. |
DAEA: Enhancing Entity Alignment in Real-World Knowledge Graphs Through Multi-Source Domain Adaptation (2025.coling-main)
Copied to clipboard
| Challenge: | Entity Alignment (EA) is a critical task in Knowledge Graph (KG) integration. |
| Approach: | They propose a novel approach that leverages the data characteristics of synthetic benchmarks to improve performance in real-world datasets. |
| Outcome: | The proposed approach outperforms state-of-the-art models on real-world datasets and achieves a 29.94% improvement in Hits@1 on DOREMUS and 5.64% improvement on AGROLD. |
PaCoRe: Learning to Scale Test-Time Compute with Parallel Coordinated Reasoning (2026.acl-long)
Copied to clipboard
Jingcheng Hu, Yinmin Zhang, Shijie Shang, Xiaobo Yang, Yue Peng, Zhewei Huang, Hebin Zhou, Xin Wu, Jie Cheng, Fanqi Wan, Xiangwen Kong, Chengyuan Yao, Kaiwen Yan, Ailin Huang, Hongyu Zhou, Qi Han, Zheng Ge, Xiangyu Zhang, Heung-Yeung Shum
| Challenge: | Parallel Coordinated Reasoning (PaCoRe) overcomes a central limitation of contemporary language models: their inability to scale test-time compute (TTC) far beyond sequential reasoning under a fixed context window. |
| Approach: | They propose a training-and-inference framework to overcome a central limitation of language models: their inability to scale test-time compute (TTC) under a fixed context window. |
| Outcome: | The proposed model scales to multi-million-token effective TTC without exceeding context limits. |
MasRouter: Learning to Route LLMs for Multi-Agent Systems (2025.acl-long)
Copied to clipboard
| Challenge: | Multi-agent systems (MAS) powered by Large Language Models (LLMs) have been demonstrated to push the boundaries of LLM capabilities, yet they often face significant costs and challenges in dynamic LLM selection. |
| Approach: | They propose a multi-agent system routing solution that integrates all components of MAS into a unified routing framework. |
| Outcome: | The proposed solution is high-performing, cost-effective, and efficient . it reduces overhead by up to 52.07 compared to current methods on HumanEval . |
Seeking Diverse Reasoning Logic: Controlled Equation Expression Generation for Solving Math Word Problems (2022.aacl-short)
Copied to clipboard
| Challenge: | Existing methods to solve Math Word Problems rely on human annotation . empirical results suggest that our method universally improves the performance on single-unknown and multiple-un unknown benchmarks. |
| Approach: | They propose a controlled equation generation solver by leveraging a set of control codes to guide the model to consider certain reasoning logic and decode the corresponding equations expressions transformed from the human reference. |
| Outcome: | The proposed method improves performance on single-unknown and multiple-un unknown benchmarks with 13.2% accuracy on the challenging multiple-unequal datasets. |
Rapidly Developing High-quality Instruction Data and Evaluation Benchmark for Large Language Models with Minimal Human Effort: A Case Study on Japanese (2024.lrec-main)
Copied to clipboard
| Challenge: | Recent advances in large language models (LLMs) have aimed to refine their capacity to accurately follow human instructions and navigate intricate scenarios. |
| Approach: | They propose a method that uses a set of instructions to translate English into Japanese and then generates Japanese instruction data using GPT-4. |
| Outcome: | The proposed method outperforms Japanese-Alpaca models in the evaluation benchmarks without human references. |
What Language Do Non-English-Centric Large Language Models Think in? (2025.findings-acl)
Copied to clipboard
Chengzhi Zhong, Qianying Liu, Fei Cheng, Junfeng Jiang, Zhen Wan, Chenhui Chu, Yugo Murawaki, Sadao Kurohashi
| Challenge: | Despite their robust performance in English, these models often exhibit reduced proficiency in non-English languages, and their outputs may reflect an inherent bias toward English-centric perspectives. |
| Approach: | They categorize non-English-centric large language models into two groups: CPMs and BLMs, which are pre-trained on a balanced mix of multiple languages from scratch. |
| Outcome: | The proposed models exhibit a pronounced internal preference for English tokens when projected into the vocabulary space. |
Relation Extraction with Weighted Contrastive Pre-training on Distant Supervision (2023.findings-eacl)
Copied to clipboard
| Challenge: | Existing methods ignore the intrinsic noise of distant supervision during the pre-training stage. |
| Approach: | They propose a weighted contrastive learning method that explicitly reduces noise . they leverage supervised data to estimate reliability and reduce noise compared to non-weighted baselines . |
| Outcome: | The proposed method reduces the noise of distant supervision and estimates reliability of pre-training instances. |
AutoPKG: An Automated Framework for Dynamic E-commerce Product-Attribute Knowledge Graph Construction (2026.findings-acl)
Copied to clipboard
Pollawat Hongwimol, Haoning Shang, Chutong Wang, Zhichao Wan, Yi Gao, Yuanming Li, Lin Gui, Wenhao Sun, Cheng Yu
| Challenge: | Product attribute extraction in e-commerce is bottlenecked by ontologies that are inconsistent, incomplete, and costly to maintain. |
| Approach: | They propose a multi-agent Large Language Model framework that constructs a Product-attribute Knowledge Graph from multimodal product content. |
| Outcome: | The proposed framework achieves 0.953 WKE for product types, 0.724 WKEs for attribute keys, and 0.531 edge-level accuracy for value assertions after canonicalization on a large real-world marketplace catalog dataset from Lazada (Alibaba). |
a1: Steep Test-time Scaling Law via Environment Augmented Generation (2026.findings-acl)
Copied to clipboard
| Challenge: | Large Language Models (LLMs) have made remarkable advances in reasoning, yet continue to struggle with hallucinations, logical errors, and inability to self-correct during complex multi-step tasks. |
| Approach: | They propose a framework that enhances LLM reasoning through real-time environmental feedback validating each reasoning step, dynamic branch exploration for investigating alternative solution paths when faced with errors, and experience-based learning from successful reasoning trajectories. |
| Outcome: | The proposed model outperforms comparable models by 24.4 percentage points across benchmarks while outperforming comparable models. |
Rescue Implicit and Long-tail Cases: Nearest Neighbor Relation Extraction (2022.emnlp-main)
Copied to clipboard
| Challenge: | Existing RE models are incapable of handling implicit expressions and long-tail relation types due to language complexity and data sparsity. |
| Approach: | They propose a method to enhance relation extraction using k nearest neighbors (kNN-RE) kNN is a nearest-neighbor search tool that allows the model to consult training relations at test time . |
| Outcome: | The proposed model outperforms the best model to date on ACE05, SciERC, and Wiki80 datasets and outperformed the best on i2b2 and Wik80 dataset. |
SpeechIQ: Speech-Agentic Intelligence Quotient Across Cognitive Levels in Voice Understanding by Large Language Models (2025.acl-long)
Copied to clipboard
Zhen Wan, Chao-Han Huck Yang, Yahan Yu, Jinchuan Tian, Sheng Li, Ke Hu, Zhehuai Chen, Shinji Watanabe, Fei Cheng, Chenhui Chu, Sadao Kurohashi
| Challenge: | SIQ quantifies voice understanding abilities and provides unified comparisons between cascaded methods and end-to-end models. |
| Approach: | They propose a human cognition-inspired evaluation pipeline for voice understanding large language models (LLM_Voice) that quantifies voice understanding abilities and provides unified comparisons between cascaded methods and end-to-end models. |
| Outcome: | The proposed framework quantifies voice understanding abilities and provides unified comparisons between cascaded methods and end-to-end models, identifies annotation errors in existing benchmarks, and detects hallucinations in LLM_Voice. |
Causal Tree Extraction from Medical Case Reports: A Novel Task for Experts-like Text Comprehension (2025.emnlp-main)
Copied to clipboard
| Challenge: | Existing methods to extract causal relationships from medical case reports are insufficient for capturing causal relationships of an entire case. |
| Approach: | They propose a task that generates a causal tree with the primary disease as the root and extracts causal relationships from a medical case report. |
| Outcome: | The proposed method outperforms the baseline method by 20.2 points in the human evaluation and introduces evaluation metrics that reflect clinician preferences. |
AgentOCR: Reimagining Agent History via Optical Self-Compression (2026.acl-long)
Copied to clipboard
| Challenge: | Recent advances in large language models (LLMs) have enabled agentic systems trained with reinforcement learning over multi-turn interaction, but practical deployment is bottlenecked by rapidly growing textual histories that inflate token and memory costs. |
| Approach: | They propose a framework that represents the accumulated observation-action history as a compact rendered image. |
| Outcome: | The proposed framework preserves over 95% of text-based agent performance while significantly reducing token consumption (>50%), yielding consistent token and memory efficiency. |
Correcting Language Model Bias for Text Classification in True Zero-Shot Learning (2024.lrec-main)
Copied to clipboard
| Challenge: | Experimental results show that pre-trained language models outperform standard prompt learning in zero-shot settings. |
| Approach: | They propose a pipeline for annotating and filtering examples from unlabeled examples . they propose 'model bias validation' method that utilizes unlabed examples as validation set . |
| Outcome: | The proposed approach outperforms standard prompt learning on six text classification tasks. |
GPT-RE: In-context Learning for Relation Extraction using Large Language Models (2023.emnlp-main)
Copied to clipboard
| Challenge: | Existing approaches to in-context learning (ICL) are lacking in relation extraction (RE) . emergence of large language models (LLMs) such as GPT-3 represents a significant advancement in natural language processing. |
| Approach: | They propose to incorporate task-aware representations into demonstration retrieval and enrich the demonstrations with gold label-induced reasoning logic. |
| Outcome: | The proposed model achieves SOTA and competitive performances on the Semeval and SciERC datasets. |
Robustness Testing of Language Understanding in Task-Oriented Dialog (2021.acl-long)
Copied to clipboard
Jiexi Liu, Ryuichi Takanobu, Jiaxin Wen, Dazhen Wan, Hongguang Li, Weiran Nie, Cheng Li, Wei Peng, Minlie Huang
| Challenge: | a lack of systematic studies on the robustness of language understanding models in task-oriented dialog systems is limiting . authors propose a model-agnostic toolkit LAUG to approximate natural language perturbations . |
| Approach: | They propose a model-agnostic toolkit LAUG to approximate natural language perturbations for testing the robustness of language understanding models in task-oriented dialog systems. |
| Outcome: | The proposed toolkit reveals critical robustness issues in state-of-the-art models. |
Rethinking the Video Sampling and Reasoning Strategies for Temporal Sentence Grounding (2022.findings-emnlp)
Copied to clipboard
Jiahao Zhu, Daizong Liu, Pan Zhou, Xing Di, Yu Cheng, Song Yang, Wenzheng Xu, Zichuan Xu, Yao Wan, Lichao Sun, Zeyu Xiong
| Challenge: | Existing methods for temporal sentence grounding ignore two crucial issues . 1) Boundary-bias: the video downsampling process may lose these two frames . 2) Reasoning-biases: such incorrect new boundary frames lead to the reasoning bias . |
| Approach: | They propose a siamese sampling mechanism to generate additional contextual frames . they use a reasoning strategy to learn the inter-relationship among these frames a . |
| Outcome: | Extensive experiments demonstrate the effectiveness of a new siamese sampling network on three challenging datasets. |
Local Byte Fusion for Neural Machine Translation (2023.acl-long)
Copied to clipboard
| Challenge: | Existing NLP models rely on a pre-built subword tokenizer to tokenize a sentence . this can be rigid and subwords from low-resource languages are under-represented . |
| Approach: | They propose a method for byte-based machine translation that aggregates local semantic information. |
| Outcome: | The proposed method improves on multilingual translation and cross-lingual transfer . it is parameter-efficient and performs competitively to subword models, it is shown . |
AceGPT, Localizing Large Language Models in Arabic (2024.naacl-long)
Copied to clipboard
Huang Huang, Fei Yu, Jianqing Zhu, Xuening Sun, Hao Cheng, Song Dingjie, Zhihong Chen, Mosen Alharthi, Bang An, Juncai He, Ziche Liu, Junying Chen, Jianquan Li, Benyou Wang, Lian Zhang, Ruoyu Sun, Xiang Wan, Haizhou Li, Jinchao Xu
| Challenge: | Significant concerns emerge when addressing cultural sensitivity and local values. |
| Approach: | They propose a localized Large Language Model (LLM) specifically for Arabic, a language imbued with unique cultural characteristics inadequately addressed by current mainstream models. |
| Outcome: | The proposed model sets the state-of-the-art standard for open Arabic LLMs across various benchmarks. |
LEDOM: Reverse Language Model (2026.acl-long)
Copied to clipboard
Xunjian Yin, Sitao Cheng, Yuxi Xie, Xinyu Hu, Li Lin, Xinyi Wang, Liangming Pan, William Yang Wang, Xiaojun Wan
| Challenge: | Autoregressive language models are trained exclusively left-to-right, yet they are limited in their ability to factorize text. |
| Approach: | They propose a purely reverse autoregressive language model that factorizes text as a product of left-to-right conditionals. |
| Outcome: | The proposed model can be used to score forward outputs using reverse posterior estimates. |
RASAT: Integrating Relational Structures into Pretrained Seq2Seq Model for Text-to-SQL (2022.emnlp-main)
Copied to clipboard
Jiexing Qi, Jingyao Tang, Ziwei He, Xiangpeng Wan, Yu Cheng, Chenghu Zhou, Xinbing Wang, Quanshi Zhang, Zhouhan Lin
| Challenge: | Experimental results show RASAT can leverage a variety of relational structures while inheriting the pretrained parameters from the T5 model. |
| Approach: | They propose a Transformer seq2seq architecture augmented with relation-aware self-attention that leverages relational structures while inheriting pretrained parameters from the T5 model. |
| Outcome: | The proposed model can leverage relational structures while inheriting pretrained parameters from the T5 model effectively. |
Leveraging High-Resource English Corpora for Cross-lingual Domain Adaptation in Low-Resource Japanese Medicine via Continued Pre-training (2025.findings-emnlp)
Copied to clipboard
Kazuma Kobayashi, Zhen Wan, Fei Cheng, Yuma Tsuta, Xin Zhao, Junfeng Jiang, Jiahao Huang, Zhiyi Huang, Yusuke Oda, Rio Yokota, Yuki Arase, Daisuke Kawahara, Akiko Aizawa, Sadao Kurohashi
| Challenge: | low-resource language corpora in professional domains like medicine hinder cross-lingual domain adaptation of pre-trained large language models. |
| Approach: | They examine how linguistic features affect performance on a Japanese–English medical knowledge benchmark. |
| Outcome: | The proposed model can leverage English-language resources in medical domains while ensuring sufficient coverage of language-specific expressions in a target language. |
LAMB: A Training-Free Method to Enhance the Long-Context Understanding of SSMs via Attention-Guided Token Filtering (2025.acl-short)
Copied to clipboard
Zhifan Ye, Zheng Wang, Kejing Xia, Jihoon Hong, Leshu Li, Lexington Whalen, Cheng Wan, Yonggan Fu, Yingyan Celine Lin, Souvik Kundu
| Challenge: | Recent work attributes performance degradation to an exponential decay in hidden-state memory. |
| Approach: | They propose a token filtering strategy that is training-free and attention-guided . they propose 'LAMB' to preserve critical tokens during inference . |
| Outcome: | The proposed token filtering improves long-context performance by 30.35% over state-of-the-art methods on benchmarks. |