Papers by Pan Zhou

MathGenie: Generating Synthetic Data with Question Back-translation for Enhancing Mathematical Reasoning of LLMs (2024.acl-long)

Challenge:	Recent advances in large language models (LLMs) have expanded their scope to encompass multimodal functions.
Approach:	They propose a robust and adaptive speech large language model with dual encoders . they validate the model on universal speech benchmarks and apply it to specialized speech-question-answer datasets based on a CoT approach .
Outcome:	The proposed model achieves state-of-the-art performance across a range of speech tasks on the same model size.

Copied to clipboard

Zimu Lu, Aojun Zhou, Houxing Ren, Ke Wang, Weikang Shi, Junting Pan, Mingjie Zhan, Hongsheng Li

Annotations Are Not All You Need: A Cross-modal Knowledge Transfer Network for Unsupervised Temporal Sentence Grounding (2023.findings-emnlp)

Challenge:	Existing models have demonstrated outstanding capabilities in mathematical reasoning, but there is a performance gap between open-source models and closed-source ones.
Approach:	They propose a method for generating diverse and reliable math problems by leveraging the ground-truth solutions of the seed data.
Outcome:	The proposed model outperforms open-source models across five representative mathematical reasoning datasets.

Copied to clipboard

Xiang Fang, Daizong Liu, Wanlong Fang, Pan Zhou, Yu Cheng, Keke Tang, Kai Zou

Misalignment Attack on Text-to-Image Models via Text Embedding Optimization and Inversion (2025.findings-emnlp)

Challenge:	Existing work on temporal sentence grounding rely on expensive video-query paired annotations . despite this, there are no ground-truth annotations in the current work .
Approach:	They propose to use paired video-query and segment boundary annotations to generate temporal sentence grounding without training.
Outcome:	The proposed model outperforms existing unsupervised methods and beats supervised ones on two challenging datasets.

Copied to clipboard

Zhijie Du, Daizong Liu, Pan Zhou

SafeAgent: Safeguarding LLM Agents via an Automated Risk Simulator (2026.acl-long)

Challenge:	Text embedding is a key component of modern NLP models but also poses additional risks.
Approach:	They propose a framework that optimizes embeddings and inverts them to obtain misaligned prompts.
Outcome:	The proposed framework exploits the continuity and distribution characteristics of text embeddings to obtain misaligned prompts of discrete tokens.

Copied to clipboard

Xueyang Zhou, Weidong Wang, Lin Lu, Jiawen Shi, Guiyao Tie, Xu Yongtian, Lixing Chen, Pan Zhou, Neil Zhenqiang Gong, Lichao Sun

Challenge:	SafeAgent improves agent safety through fully automated synthetic data generation.
Approach:	They propose a framework that improves agent safety through fully automated synthetic data generation.
Outcome:	The proposed framework outperforms closed-source models on two safety benchmarks and one real-world task.

STARS: A Unified Framework for Singing Transcription, Alignment, and Refined Style Annotation (2025.findings-acl)

Copied to clipboard

Wenxiang Guo, Yu Zhang, Changhao Pan, Zhiyuan Zhu, Ruiqi Li, ZheTao Chen, Wenhao Xu, Fei Wu, Zhou Zhao

Speech & Audio in NLP

Challenge:	Existing automated singing annotation (ASA) methods tackle isolated aspects of the annotation pipeline.
Approach:	They propose a framework that addresses transcription, alignment, and refined style annotations.
Outcome:	The proposed framework delivers comprehensive multi-level annotations encompassing: (1) precise phoneme-audio alignment, (2) robust note transcription and temporal localization, (3) expressive vocal technique identification, and (4) global stylistic characterization including emotion and pace.

Aligning VLM Assistants with Personalized Situated Cognition (2025.acl-long)

Copied to clipboard

Yongqi Li, Shen Zhou, Xiaohu Li, Xin Miao, Jintao Wen, Mayi Xu, Jianhao Chen, Birong Pan, Hankun Kang, Yuanyuan Zhu, Ming Zhong, Tieyun Qian

SocAoG: Incremental Graph Parsing for Social Relation Inference in Dialogues (2021.acl-long)

Challenge:	Existing studies on vision-language models aligned with general human objectives have not been successful because people with diversified backgrounds have different cognition even in the same situation.
Approach:	They propose to characterize individuals based on the sociological concept of Role-Set and then evaluate their actions to see whether personalized alignment is achieved.
Outcome:	The proposed framework constructs a cognition-aware and action-based reward model for personalized alignment.

Copied to clipboard

Liang Qiu, Yuan Liang, Yizhou Zhao, Pan Lu, Baolin Peng, Zhou Yu, Ying Nian Wu, Song-Chun Zhu

Structured Data in NLP

Challenge:	Existing studies focus on identifying entities' relations from the semantics of dialogues-they utilize either the attention mechanism or a refined token graph to locate informative words.
Approach:	They propose a sequential structure prediction task to incrementally parse SocAoG for dynamic inference upon any incoming utterance.
Outcome:	Empirical results show that the proposed model infers social relations more accurately than the state-of-the-art methods.

SDiaReward: Modeling and Benchmarking Spoken Dialogue Rewards with Modality and Colloquialness (2026.acl-long)

Copied to clipboard

Jingyu Lu, Yuhan Wang, Fan Zhuo, Xize Cheng, Changhao Pan, Xueyi Pu, Yifu Chen, Chenyuhao Wen, Tianle Liang, Zhou Zhao

Speech & Audio in NLP

Challenge:	SDiaReward is an end-to-end spoken dialogue system that integrates paralinguistic nuances and spontaneous nature of human conversation.
Approach:	They propose an end-to-end multi-turn reward model trained on SDiaReward-Dataset . it is a collection of episode-level preference pairs targeting modality and colloquiality gaps .
Outcome:	The proposed model outperforms general-purpose audio LLMs in episode-level evaluation.

F2RL: Factuality and Faithfulness Reinforcement Learning Framework for Claim-Guided Evidence-Supported Counterspeech Generation (2024.emnlp-main)

Copied to clipboard

Haiyang Wang, Yuchen Pan, Xin Song, Xuechen Zhao, Minghao Hu, Bin Zhou

TAVT: Towards Transferable Audio-Visual Text Generation (2023.acl-long)

Challenge:	Existing methods for generating evidence-supported counterspeech lack clear guidance with a core claim for organizing evidence.
Approach:	They propose a Factuality and Faithfulness Reinforcement Learning framework for generating claim-guided and evidence-supported counterspeech (F2RL) they generate counter-claims based on hate speech and design a self-evaluation mechanism to select the most appropriate one.
Outcome:	The proposed framework achieves excellent performance on three benchmark datasets with strong factuality and faithfulness.

Copied to clipboard

Wang Lin, Tao Jin, Wenwen Pan, Linjun Li, Xize Cheng, Ye Wang, Zhou Zhao

Breaking Language Barriers: Cross-Lingual Continual Pre-Training at Scale (2024.emnlp-main)

Challenge:	Existing transfer learning techniques focus on uni-modal analysis and lack consideration of multi-modal content and cross-modal relation.
Approach:	They propose a transferable audio-visual text generation framework that incorporates two components: Audio-Visual Meta-Mapper and Dual Counterfactual Contrastive Learning.
Outcome:	The proposed framework outperforms the state-of-the-art methods across multiple domains and modal settings.

Copied to clipboard

Wenzhen Zheng, Wenbo Pan, Xu Xu, Libo Qin, Li Yue, Ming Zhou

MathCoder-VL: Bridging Vision and Code for Enhanced Multimodal Mathematical Reasoning (2025.findings-acl)

Challenge:	Large Language Models (LLMs) have made significant strides towards Artificial General Intelligence, but training them from scratch is prohibitively expensive.
Approach:	They propose to continuously pre-train LLMs from existing pre-trained LLM models by using a set of parameters instead of randomly initializing them.
Outcome:	The proposed approach saves significant resources and accelerates convergence and performance.

Copied to clipboard

Ke Wang, Junting Pan, Linda Wei, Aojun Zhou, Weikang Shi, Zimu Lu, Han Xiao, Yunqiao Yang, Houxing Ren, Mingjie Zhan, Hongsheng Li

Learning to Plan for Retrieval-Augmented Large Language Models from Knowledge Graphs (2024.findings-emnlp)

Challenge:	Large Language Models (LMMs) struggle with simple tasks such as geometry, e.g., arithmetic, and reasoning.
Approach:	They propose to leverage code as supervision for cross-modal alignment . they propose to use FigCodifier and ImgCode-8.6M to synthesize novel mathematical figures .
Outcome:	The proposed model surpasses GPT-4o and Claude 3.5 Sonnet in the geometry problem-solving subset of MathVista, achieving improvements of 8.9% and 9.2%.

Copied to clipboard

Junjie Wang, Mingyang Chen, Binbin Hu, Dan Yang, Ziqi Liu, Yue Shen, Peng Wei, Zhiqiang Zhang, Jinjie Gu, Jun Zhou, Jeff Pan, Wen Zhang, Huajun Chen

Less is More: Making Smaller Language Models Competent Subgraph Retrievers for Multi-hop KGQA (2024.findings-emnlp)

Challenge:	Recent studies have attempted to enhance the performance of large language models (LLMs) in complex question-answering (QA) tasks by combining step-wise planning with external retrieval.
Approach:	They propose a framework for enhancing LLMs’ planning capabilities by using planning data derived from knowledge graphs (KGs).
Outcome:	The proposed framework improves LLMs’ planning capabilities by using knowledge graphs (KGs) the proposed framework is compared with existing frameworks on multiple datasets and shows that it is effective for large language models.

Copied to clipboard

Wenyu Huang, Guancheng Zhou, Hongru Wang, Pavlos Vougiouklis, Mirella Lapata, Jeff Pan

Virtual Context Enhancing Jailbreak Attacks with Special Token Injection (2024.findings-emnlp)

Challenge:	Recent studies suggest that Knowledge Graphs (KGs) contain valuable external knowledge for LLMs.
Approach:	They propose to model a conditional subgraph retrieval task handled by small language models and use a subgraph identifier as a special token to retrieve subgraphs.
Outcome:	The proposed model achieves competitive retrieval performance compared to state-of-the-art models relying on 7B parameters.

Copied to clipboard

Yuqi Zhou, Lin Lu, Ryan Sun, Pan Zhou, Lichao Sun

SQL-ASTRA: Alleviating Sparse Feedback in Agentic SQL via Column-Set Matching and Trajectory Aggregation (2026.findings-acl)

Challenge:	Existing jailbreak attacks target the two phases of user interaction: prompt input and model computation.
Approach:	They propose a new tool that leverages special tokens to improve jailbreak attacks . they found that the tool can increase success rates of existing jailbreak methods by 40% .
Outcome:	The proposed solution can improve success rates of four widely used jailbreak methods by approximately 40% across various LLMs.

Copied to clipboard

Long Li, Zhijian Zhou, Jiangxuan Long, Peiyang Liu, Weidi Xu, Zhe Wang, Shirui Pan, Chao Qu

Maximum Score Routing For Mixture-of-Experts (2025.findings-acl)

Challenge:	Agentic SQL is a framework for multiturn agent learning, but it is limited to single-turn paradigms.
Approach:	They propose a framework that provides a universal two-tiered reward mechanism for credit assignment . they propose 'Aggregated Trajectory Reward' to resolve multi-turn credit assignment.
Outcome:	The proposed framework outperforms SOTA Arctic-Text2SQL-R1-7B on BIRD and Spider 2.0 using identical models.

Copied to clipboard

Bowen Dong, Yilong Fan, Yutao Sun, Zhenyu Li, Tengyu Pan, Zhou Xun, Jianyong Wang

Green & Sustainable NLP

Challenge:	Traditional mixture-of-experts (MoE) networks impose an expert capacity constraint to ensure GPU-friendly computation.
Approach:	They propose a routing paradigm that dynamically allocates input tokens to top-k experts through differentiable sparse transformations, enabling scalable model capacity while preserving computational efficiency.
Outcome:	The proposed model achieves lower training losses and higher evaluation scores at equivalent FLOPs compared to constrained and unconstrained baselines.

Intent-Aware and Hate-Mitigating Counterspeech Generation via Dual-Discriminator Guided LLMs (2024.lrec-main)

Copied to clipboard

Haiyang Wang, Zhiliang Tian, Xin Song, Yue Zhang, Yuchen Pan, Hongkui Tu, Minlie Huang, Bin Zhou

Merger-as-a-Stealer: Stealing Targeted PII from Aligned LLMs with Model Merging (2025.emnlp-main)

Challenge:	Hate speech is an aggressive expression that incites hatred towards specific groups based on their group identity.
Approach:	They propose an LLMs-based framework for counterspeech generation that uses intent-aware discriminators to decode intents of LLM models.
Outcome:	The proposed framework matches intents with hate mitigation intents and performs well.

Copied to clipboard

Lin Lu, Zhigang Zuo, Ziji Sheng, Pan Zhou

ReflectionCoder: Learning from Reflection Sequence for Enhanced One-off Code Generation (2025.acl-long)

Challenge:	Model merging is a promising approach for updating large language models . but unmonitored mergers can introduce significant security vulnerabilities .
Approach:	They propose a model merging attack surface where a malicious merger can extract PII from an aligned model with model merg.
Outcome:	The proposed framework can extract PII from an aligned model with model merging.

Copied to clipboard

Houxing Ren, Mingjie Zhan, Zhongyuan Wu, Aojun Zhou, Junting Pan, Hongsheng Li

TCSinger: Zero-Shot Singing Voice Synthesis with Style Transfer and Multi-Level Style Control (2024.emnlp-main)

Challenge:	Existing methods to enhance code generation performance include integrating compiler feedback.
Approach:	They propose a method that integrates compiler feedback to improve one-off code generation performance.
Outcome:	The proposed method improves one-off code generation performance on three benchmarks and can be applied to other domains that focus on final results and require long reasoning paths.

Copied to clipboard

Yu Zhang, Ziyue Jiang, Ruiqi Li, Changhao Pan, Jinzheng He, Rongjie Huang, Chuxin Wang, Zhou Zhao

Low-Resource NLP

Challenge:	Existing models fail to generate singing voices rich in stylistic nuances for unseen singers due to multifaceted nature of singing styles.
Approach:	They propose a zero-shot SVS model for style transfer across cross-lingual speech and singing styles and multi-level style control.
Outcome:	Experimental results show that TCSinger outperforms baseline models in synthesis quality, singer similarity, and style controllability.

CrowdSelect: SyntheticInstruction Data Selection with Multi-LLM Wisdom (2026.findings-eacl)

Copied to clipboard

Yisen Li, Lingfeng Yang, Wenxuan Shen, Pan Zhou, Yao Wan, Weiwei Lin, Dongping Chen

Machine Translation

Challenge:	Existing methods for capturing instruction-following complexity rely on single-dimensional signals, but they fail to capture complexity across diverse fields.
Approach:	They propose three foundational metrics that leverage Multi-LLMs wisdom to capture instruction-response pair characteristics and propose CrowdSelect, an integrated metric incorporating a clustering-based approach to maintain response diversity.
Outcome:	The proposed metrics outperform existing models on MT-bench and Arena-hard and show improvements of 4.81% on full and LoRA fine-tuning.

Probability-Consistent Preference Optimization for Enhanced LLM Reasoning (2025.findings-acl)

Copied to clipboard

Yunqiao Yang, Houxing Ren, Zimu Lu, Ke Wang, Weikang Shi, Aojun Zhou, Junting Pan, Mingjie Zhan, Hongsheng Li

RESIN: A Dockerized Schema-Guided Cross-document Cross-lingual Cross-media Information Extraction and Event Tracking System (2021.naacl-demos)

Challenge:	Recent advances in preference optimization have demonstrated significant potential for improving mathematical reasoning capabilities in large language models.
Approach:	They propose a framework that establishes two quantitative metrics for preference selection: surface-level answer correctness and intrinsic token-level probability consistency.
Outcome:	The proposed framework outperforms existing outcome-only criterion approaches across a diverse range of LLMs and benchmarks.

Copied to clipboard

Haoyang Wen, Ying Lin, Tuan Lai, Xiaoman Pan, Sha Li, Xudong Lin, Ben Zhou, Manling Li, Haoyu Wang, Hongming Zhang, Xiaodong Yu, Alexander Dong, Zhenhailong Wang, Yi Fung, Piyush Mishra, Qing Lyu, Dídac Surís, Brian Chen, Susan Windisch Brown, Martha Palmer, Chris Callison-Burch, Carl Vondrick, Jiawei Han, Dan Roth, Shih-Fu Chang, Heng Ji

Visual Data in NLP

Challenge:	We present a new information extraction system that can construct temporal event graphs from news documents.
Approach:	They propose a temporal event graph extraction system that can extract news documents . they extend the system from sentence-level event extraction to cross-document cross-media event extraction .
Outcome:	The proposed system can extract temporal event graphs from news documents in multiple languages and multiple data modalities.

PsychEval: A Multi-Session and Multi-Therapy Benchmark for High-Realism AI Psychological Counselor (2026.findings-acl)

Copied to clipboard

Qianjun Pan, Junyi Wang, Jie Zhou, Yutao Yang, Junsong Li, Kaiyin Xu, Yougen Zhou, Yihan Li, JingYuan Zhao, Qin Chen, Ningning Zhou, Kai Chen, Liang He

Challenge:	Existing models focus on a single therapy, but complex cases require flexible strategies among various therapies.
Approach:	They propose a multi-session, multi-therapy, and highly realistic benchmark . it is designed to address three key challenges: 1) can we train a highly realistic AI counselor? 2) How to systematically evaluate an AI counselor?"
Outcome:	The proposed benchmark is annotated with extensive professional skills and includes over 677 meta-skills and 4577 atomic skills.

AntiLeakBench: Preventing Data Contamination by Automatically Constructing Benchmarks with Updated Real-World Knowledge (2025.acl-long)

Copied to clipboard

Xiaobao Wu, Liangming Pan, Yuxi Xie, Ruiwen Zhou, Shuai Zhao, Yubo Ma, Mingzhe Du, Rui Mao, Anh Tuan Luu, William Yang Wang

From Curated Data to Scalable Models: Continual Pre-training of Dense and MoE Large Language Models for Tibetan (2026.acl-long)

Challenge:	Existing studies solve this challenge by updating benchmarks with newly collected data, but they fail to guarantee contamination-free evaluation as the newly collected knowledge may contain pre-existing knowledge.
Approach:	They propose an automated anti-leakage benchmarking framework that builds and updates benchmarks without human labor instead of using newly collected data.
Outcome:	The proposed framework significantly reduces the cost of benchmark maintenance to accommodate emerging LLMs.

Copied to clipboard

Lei Yang, Leiyu Pan, Bojian Xiong, Renren Jin, Shaowei Zhang, Yue Chen, Ling Shi, Jiang Zhou, Junru Wu, Zhen Wang, Jianxiang Peng, Juesi Xiao, Tianyu Dong, Zhuowen Han, Zhuo Chen, Yuqi Ren, Deyi Xiong

A Survey on Training-free Alignment of Large Language Models (2025.findings-emnlp)

Challenge:	Large language models have achieved remarkable success across a wide range of tasks, yet their performance remains heavily biased toward high-resource languages.
Approach:	They propose a pipeline for advancing Tibetan language modeling through multilingual continual pre-training with Tibetan, Chinese, and English.
Outcome:	The proposed model outperforms open-source and Tibetan-focused models on diverse tasks.

Copied to clipboard

Birong Pan, Yongqi Li, Weiyu Zhang, Wenpeng Lu, Mayi Xu, Shen Zhou, Yuanyuan Zhu, Ming Zhong, Tieyun Qian

Adversarial Category Alignment Network for Cross-domain Sentiment Classification (N19-1)

Challenge:	a survey of large language models (LLMs) aims to ensure outputs adhere to human values, ethical standards, and legal norms.
Approach:	They present the first systematic review of TF alignment methods . they categorize them by stages of pre-decoding, in-decoder and post-decoration .
Outcome:	The proposed methods are based on training-free (TF) alignment techniques . they are able to be used in open-source and closed-source environments without retraining .

Copied to clipboard

Xiaoye Qu, Zhikang Zou, Yu Cheng, Yang Yang, Pan Zhou

Sentiment Analysis

Challenge:	Existing methods for cross-domain sentiment classification focus on aligning marginal distribution without taking category-specific decision boundaries into consideration.
Approach:	They propose an adversarial category alignment network to enhance category consistency . experimental results show the proposed method can achieve state-of-the-art performance .
Outcome:	The proposed method achieves state-of-the-art performance and produces more discriminative features on benchmark datasets.

Comprehensive Benchmarking of Long-Form Speech Generation in Diverse Scenarios (2026.findings-acl)

Copied to clipboard

Changhao Pan, Rui Yang, Han Wang, Zhuan Zhou, Xuming He, Wenxiang Guo, Ziyue Jiang, Ruiqi Li, Yu Zhang, Chenyuhao Wen, Ke Lei, Xiang Yin, Jingyu Lu, Zhiyuan Zhu, Zhou Zhao

GroundingGPT: Language Enhanced Multi-modal Grounding Model (2024.acl-long)

Challenge:	Existing evaluation benchmarks for long-form speech are limited to limited domains, creating a significant gap with the diverse downstream applications.
Approach:	They propose a benchmark that decomposes "long-form speech quality" into specific, disentangled dimensions.
Outcome:	The proposed benchmark decomposes “long-form speech quality” into specific, disentangled dimensions.

Copied to clipboard

Zhaowei Li, Qi Xu, Dong Zhang, Hang Song, YiQing Cai, Qi Qi, Ran Zhou, Junting Pan, Zefeng Li, Vu Tu, Zhida Huang, Tao Wang

MedWriter: Knowledge-Aware Medical Text Generation (2020.coling-main)

Challenge:	Existing multi-modal large language models focus on capturing global information while neglecting the fine-grained local information in multimodal inputs.
Approach:	They propose an end-to-end language enhanced multi-modal grounding model that performs fine-grained grounding tasks for image, video and audio.
Outcome:	The proposed model achieves impressive fine-grained understanding of multi-modal inputs while maintaining or improving its global comprehension capabilities.

Copied to clipboard

Youcheng Pan, Qingcai Chen, Weihua Peng, Xiaolong Wang, Baotian Hu, Xin Liu, Junying Chen, Wenxiu Zhou

Structured Data in NLP

Challenge:	Recent studies focus on the information of unstructured text rather than structured information of the knowledge graph.
Approach:	They propose a knowledge-aware text generation model for medical domains that incorporates knowledge graphs into the model to improve the quality of generated text.
Outcome:	The proposed model improves the quality of generated text and has robust superiority over other methods.

Understanding and Mitigating Overrefusal in LLMs from an Unveiling Perspective of Safety Decision Boundary (2025.emnlp-main)

Copied to clipboard

Licheng Pan, Yongqi Tong, Xin Zhang, Xiaolu Zhang, Jun Zhou, Zhixuan Chu

Glance-or-Gaze: Incentivizing LMMs to Adaptively Focus Search via Reinforcement Learning (2026.findings-acl)

Challenge:	Large language models (LLMs) often refuse to answer legitimate queries, causing models to treat many reasonable prompts as potentially risky.
Approach:	They propose a framework that automatically generates and selects overrefusal prompts near the safety boundary.
Outcome:	The proposed framework identifies and curates boundary-aligned prompts, enabling more effective and targeted mitigation of overrefusal.

Copied to clipboard

Hongbo Bai, Yujin Zhou, Yile Wu, Chi-Min Chan, Pengcheng Wen, Kunhao Pan, Sirui Han, Yike Guo

FR-Spec: Accelerating Large-Vocabulary Language Models via Frequency-Ranked Speculative Sampling (2025.acl-long)

Challenge:	Existing search-augmented approaches rely on indiscriminate whole-image retrieval and lack deep iterative reflection, limiting their effectiveness on complex visual queries.
Approach:	They propose a fully autonomous framework that shifts from passive perception to active visual planning and introduces a Selective Gaze mechanism that dynamically chooses whether to glance at global context or gaze into high-value regions.
Outcome:	Experiments across six benchmarks demonstrate state-of-the-art performance.

Copied to clipboard

Weilin Zhao, Tengyu Pan, Xu Han, Yudi Zhang, Sun Ao, Yuxiang Huang, Kaihuo Zhang, Weilun Zhao, Yuxuan Li, Jie Zhou, Hao Zhou, Jianyong Wang, Maosong Sun, Zhiyuan Liu

Pre-training Multilingual Neural Machine Translation by Leveraging Alignment Information (2020.emnlp-main)

Challenge:	Speculative sampling is an efficient way to accelerate the auto-regressive generation process of large language models.
Approach:	They propose a frequency-ranked speculative sampling framework that optimizes draft candidate selection through vocabulary space compression.
Outcome:	Experiments show that FR-Spec reduces LM Head computation overhead by 75% while ensuring the equivalence of the final output distribution.

Copied to clipboard

Zehui Lin, Xiao Pan, Mingxuan Wang, Xipeng Qiu, Jiangtao Feng, Hao Zhou, Lei Li

Multilinguality

Challenge:	Existing pre-training methods are not effective for machine translation tasks.
Approach:	They propose a method to pre-train a universal multilingual neural machine translation model . they use random aligned substitution technique to bring words and phrases with similar meanings closer in the representation space.
Outcome:	The proposed approach improves translation quality on low, medium, rich resource languages.

MiCEval: Unveiling Multimodal Chain of Thought’s Quality via Image Description and Reasoning Steps (2025.naacl-long)

Copied to clipboard

Xiongtao Zhou, Jie He, Lanyu Chen, Jingyu Li, Haojing Chen, Victor Gutierrez Basulto, Jeff Z. Pan, Hanjie Chen

Discourse Marker Augmented Network with Reinforcement Learning for Natural Language Inference (P18-1)

Challenge:	Existing methods for evaluating the quality of reasoning steps in multimodal chain-of-thought are lacking.
Approach:	They propose a framework to evaluate the correctness of reasoning chains by evaluating the quality of both the description and each reasoning step.
Outcome:	The proposed framework improves interpretability and human judgments on four state-of-the-art MLLMs.

Copied to clipboard

Boyuan Pan, Yazheng Yang, Zhou Zhao, Yueting Zhuang, Deng Cai, Xiaofei He

LearnerCoMPASS: Intelligent Tutoring System with Dynamic Cognitive Diagnosis and Multi-Model Path Planning (2026.acl-long)

Challenge:	Existing approaches to natural language inference focus on interaction architectures of sentences . but, we propose to transfer knowledge from discourse markers to augment the model .
Approach:	They propose to transfer knowledge from discourse markers to augment the quality of the NLI model.
Outcome:	The proposed method achieves state-of-the-art performance on large-scale datasets.

Copied to clipboard

Ziji Sheng, Guiyao Tie, Weidong Wang, Pan Zhou, Daizong Liu

Living in the Moment: Can Large Language Models Grasp Co-Temporal Reasoning? (2024.acl-long)

Challenge:	Existing adaptive learning systems struggle to achieve deep personalization, dynamic adaptability, and content trustworthiness.
Approach:	They propose a framework that integrates large language models into adaptive learning systems . they propose 'cognitive multi-model planning adapted system' to enable deep personalization .
Outcome:	The proposed framework outperforms state-of-the-art learning paths and improves trustworthiness.

Copied to clipboard

Zhaochen Su, Juntao Li, Jun Zhang, Tong Zhu, Xiaoye Qu, Pan Zhou, Yan Bowen, Yu Cheng, Min Zhang

End-to-end Task-oriented Dialogue: A Survey of Tasks, Methods, and Future Directions (2023.emnlp-main)

Challenge:	Current temporal reasoning datasets are limited to questions about single or isolated events, falling short in mirroring the realistic temporal characteristics involving concurrent nature and intricate temporal interconnections.
Approach:	They propose a co-temporal Question Answering benchmark that contains four co-time scenarios with 4,748 samples for evaluating the co-timing abilities of large language models.
Outcome:	The proposed benchmarks show that current LLMs struggle on CoTempQA tasks even when enhanced with Chain of Thought methodologies.

Copied to clipboard

Libo Qin, Wenbo Pan, Qiguang Chen, Lizi Liao, Zhou Yu, Yue Zhang, Wanxiang Che, Min Li

Dialogue Systems & Conversational Agents

Challenge:	End-to-end task-oriented dialogue (EToD) can generate responses in an end-to end fashion without modular training, which attracts escalating popularity.
Approach:	They present a systematic review of EToD and propose a unified perspective to summarize existing approaches and recent trends.
Outcome:	The proposed approaches can generate responses in an end-to-end fashion without modular training, which attracts escalating popularity.

DMSD: Dual-Modal Semantic Disentanglement for Compositional Zero-Shot Learning (2026.findings-acl)

Copied to clipboard

Pan Yang, Jing Yang, Ruan Xiao li, Yuling Chen, Yuankai Wu, Quan Zhou, Xu Wang

Progressively Guide to Attend: An Iterative Alignment Framework for Temporal Sentence Grounding (2021.emnlp-main)

Challenge:	Compositional Zero-Shot Learning (CZSL) is a new research paradigm that learns sub-concepts from seen compositions and recognizes unseen novel combinations.
Approach:	They propose a Dual-Modal Semantic Disentanglement framework that integrates visual and textual information to achieve effective sub-concept disentangling.
Outcome:	The proposed framework achieves state-of-the-art performance on three benchmark datasets . it integrates a class-centroid bridge module to guide class centroids toward the textual space .

Copied to clipboard

Daizong Liu, Xiaoye Qu, Pan Zhou

Beyond Timestamps: Bridging Forward and Backward Reasoning in Temporal Numerical and Relational Understanding (2026.acl-long)

Challenge:	Existing methods to learn effective alignment between vision and language features are insufficient in practice due to complicated multi-step reasoning.
Approach:	They propose an iterative alignment network which iterates inter- and intra-modal features within multiple steps for more accurate grounding.
Outcome:	The proposed model performs better than the state-of-the-arts on three challenging benchmarks.

Copied to clipboard

Xinying Qian, Ying Zhang, Xuhui Sui, Yu Zhao, Baohang Zhou, Jeff Z. Pan

Alignment with Fill-In-the-Middle for Enhancing Code Generation (2025.emnlp-main)

Challenge:	Existing benchmarks for Temporal Numerical and Relational reasoning rely on single-task evaluation paradigms.
Approach:	They propose a benchmark to evaluate Temporal Numerical and Relational reasoning . they propose QA and verification, and a Consistency Rate to quantify robustness .
Outcome:	The proposed framework evaluates both Temporal Numerical and Relational reasoning . it measures the alignment between QA and FV and the Consistency Rate measures robustness across these directions.

Copied to clipboard

Houxing Ren, Zimu Lu, Weikang Shi, Haotian Hou, Yunqiao Yang, Ke Wang, Aojun Zhou, Junting Pan, Mingjie Zhan, Hongsheng Li

CodeIP: A Grammar-Guided Multi-Bit Watermark for Large Language Models of Code (2024.findings-emnlp)

Challenge:	Existing methods for generating test cases with limited training data are not reliable and may be counterproductive.
Approach:	They propose a method that splits code snippets into smaller, granular blocks, creating more diverse DPO pairs from the same test cases.
Outcome:	The proposed approach shows significant improvements in code generation tasks on benchmark datasets such as HumanEval (+), MBPP (+), and APPS.

Copied to clipboard

Batu Guan, Yao Wan, Zhangqian Bi, Zheng Wang, Hongyu Zhang, Pan Zhou, Lichao Sun

CE-GPPO: Coordinating Entropy via Gradient-Preserving Clipping Policy Optimization in Reinforcement Learning (2026.acl-long)

Challenge:	Large Language Models (LLMs) have shown significant potential in code generation, but they also present challenges regarding the protection of Intellectual Property (IP) related to model architectures, weights, and training data.
Approach:	They propose a multi-bit watermarking technique that embeds additional information to preserve provenance details, such as the vendor ID of an LLM.
Outcome:	The proposed technique preserves provenance details while maintaining syntactical correctness of generated code.

Copied to clipboard

Zhenpeng Su, Leiyu Pan, Minxuan Lv, Yuntao Li, Wenping Hu, Fuzheng Zhang, Kun Gai, Guorui Zhou

Merge Hijacking: Backdoor Attacks to Model Merging of Large Language Models (2025.acl-long)

Challenge:	Existing methods for proximal policy optimization discard valuable gradient signals from low-probability tokens due to the clipping mechanism.
Approach:	They propose an algorithm that reintroduces gradients from clipped tokens in native PPO in a gentle and bounded manner.
Outcome:	The proposed algorithm outperforms strong baselines on reasoning benchmarks on different model scales.

Copied to clipboard

Zenghui Yuan, Yangming Xu, Jiawen Shi, Pan Zhou, Lichao Sun

CoMave: Contrastive Pre-training with Multi-scale Masking for Attribute Value Extraction (2023.findings-acl)

Challenge:	Existing research on model merging focuses on optimizing model performance and minimizing backdoors.
Approach:	They propose a backdoor attack targeting model merging in Large Language Models that creates a unified model for multi-domain tasks.
Outcome:	The proposed attack is effective across models, merging algorithms, and tasks while maintaining utility across tasks.

Copied to clipboard

Xinnan Guo, Wentao Deng, Yongrui Chen, Yang Li, Mengdi Zhou, Guilin Qi, Tianxing Wu, Dong Yang, Liubin Wang, Yong Pan

Versatile Framework for Song Generation with Prompt-based Control (2025.findings-emnlp)

Challenge:	Existing methods to extract product features from unstructured text still suffer from problems . e-commerce platforms are focusing on multi-scale values, which can be confusing .
Approach:	They propose a pre-training technique to automatically obtain attribute value pairs from product descriptions to aid e-commerce.
Outcome:	The proposed method improves on the existing token-level masking strategy and achieves state-of-the-art on four benchmarks.

Copied to clipboard

Yu Zhang, Wenxiang Guo, Changhao Pan, Zhiyuan Zhu, Ruiqi Li, Jingyu Lu, Rongjie Huang, Ruiyuan Zhang, Zhiqing Hong, Ziyue Jiang, Zhou Zhao

TCSinger 2: Customizable Multilingual Zero-shot Singing Voice Synthesis (2025.findings-acl)

Challenge:	Existing methods for song generation fail to generate vocals with prompt-based control and proper alignment.
Approach:	VersBand is a multi-task song generation framework for synthesizing high-quality songs with prompt-based control.
Outcome:	Experimental results show that VersBand performs better than baseline models across multiple song generation tasks.

Copied to clipboard

Yu Zhang, Wenxiang Guo, Changhao Pan, Dongyu Yao, Zhiyuan Zhu, Ziyue Jiang, Yuhan Wang, Tao Jin, Zhou Zhao

MarkLLM: An Open-Source Toolkit for LLM Watermarking (2024.emnlp-demo)

Challenge:	Existing zero-shot singing voice synthesis models depend on phoneme and note boundary annotations, limiting their robustness and producing poor transitions between phonemes and notes.
Approach:	They propose a multi-task multilingual zero-shot SVS model with style transfer and style control based on various prompts.
Outcome:	Experimental results show that TCSinger 2 outperforms baseline models in subjective and objective metrics across multiple related tasks.

Copied to clipboard

Leyi Pan, Aiwei Liu, Zhiwei He, Zitian Gao, Xuandong Zhao, Yijian Lu, Binglin Zhou, Shuliang Liu, Xuming Hu, Lijie Wen, Irwin King, Philip Yu

MPBench: A Comprehensive Multimodal Reasoning Benchmark for Process Errors Identification (2025.findings-acl)

Challenge:	Large Language Models (LLMs) embed imperceptible yet algorithmically detectable signals in outputs to identify LLM-generated text.
Approach:	They propose to develop an open-source toolkit for LLM watermarking that embeds imperceptible yet algorithmically detectable signals in model outputs to identify LLM-generated text.
Outcome:	MarkLLM provides a unified framework for implementing LLM watermarking algorithms, while providing user-friendly interfaces to ensure ease of access.

Copied to clipboard

xu Zhao Pan, Pengfei Zhou, Jiaxin Ai, Wangbo Zhao, Kai Wang, Xiaojiang Peng, Wenqi Shao, Hongxun Yao, Kaipeng Zhang

COIG-P: A High-Quality and Large-Scale Chinese Preference Dataset for Alignment with Human Values (2026.findings-eacl)

Challenge:	Existing benchmarks of large language models focus on error detection, neglecting other scenarios like reasoning search.
Approach:	et al. propose a multi-task, multimodal benchmark to assess effectiveness of PRMs . step correctness, answers aggregation and reasoning process search are evaluated . ethical principles of MPBench are based on a set of evaluation paradigms based in a text-based benchmark .
Outcome:	a new benchmark assesses the effectiveness of large language models (LLMs) in multiple scenarios . it uses three evaluation paradigms to assess the effectiveness and compares them with existing models . a the proposed model improves reasoning accuracy by providing stepwise feedback for multi-step reasoning results .

Copied to clipboard

Siwei Wu, JinCheng Ren, Xeron Du, Shuyue Guo, Xingwei Qu, Yiming Liang, Jie Liu, Yunwen Li, Tyler Loakman, Tianyu Zheng, Boyu Feng, Huaqing Yuan, Zili Wang, Jiaheng Liu, Wenhao Huang, Chenglin Cai, Haoran Que, Jian Yang, Yuelin Bai, Zekun Moore Wang, Zhouliang Yu, Qunshu Lin, Ding Pan, Yuchen Eleanor Jiang, Tiannan Wang, Wangchunshu Zhou, Shenzhi Wang, Xingyuan Bu, Minghao Liu, Guoyin Wang, Ge Zhang, Chenghua Lin

Improving Open-Domain Dialogue Systems via Multi-Turn Incomplete Utterance Restoration (D19-1)

Challenge:	Existing Chinese preference datasets suffer from limited scale, restricted domain coverage, and insufficiently rigorous data validation.
Approach:	They propose an LLM-based data annotation pipeline with no human intervention to annotate Chinese preference datasets.
Outcome:	The proposed pipeline outperforms existing Chinese preference datasets on AlignBench and Chinese Reward Benchmark.

Copied to clipboard

Zhufeng Pan, Kun Bai, Yan Wang, Lianqiang Zhou, Xiaojiang Liu

Natural Language Interfaces

Challenge:	Experimental results show that restoring incomplete utterances from context improves the performance of open-domain dialogue systems.
Approach:	They propose to use a dataset to restore incomplete utterances from context . they propose to pick and combine the data to restore the incomplete .
Outcome:	The proposed model significantly boosts response quality of open-domain dialogue systems.

Battling against Tough Resister: Strategy Planning with Adversarial Game for Non-collaborative Dialogues (2025.acl-long)

Copied to clipboard

Haiyang Wang, Zhiliang Tian, Yuchen Pan, Xin Song, Xin Niu, Minlie Huang, Bin Zhou

DSMoE: Matrix-Partitioned Experts with Dynamic Routing for Computation-Efficient Dense LLMs (2025.emnlp-main)

Challenge:	Non-collaborative dialogue involves two participants with conflicting interests engaging in multiround dialogue to achieve their own goals.
Approach:	They propose a Game-based Adversarial self-play InterActive training paradigm which constructs an adversarial two-player (a persuader and a resister) zero-sum game and guides the game to approximate Nash Equilibrium (NE) via reinforcement learning.
Outcome:	The proposed model achieves state-of-the-art performance on three datasets.

Copied to clipboard

Minxuan Lv, Zhenpeng Su, Leiyu Pan, Yizhe Xiong, Zijia Lin, Hui Chen, Wei Zhou, Jungong Han, Guiguang Ding, Wenwu Ou, Di Zhang, Kun Gai, Songlin Hu

Deputy: Accelerating Large Language Model Inference with Dynamic Low-Rank Substitution (2026.findings-acl)

Challenge:	Existing sparsification methods like pruning can lose model knowledge through parameter removal.
Approach:	They propose a novel approach that achieves sparsification by partitioning pre-trained FFN layers into computational blocks.
Outcome:	The proposed approach achieves superior performance across language modeling and downstream tasks under equivalent computational constraints.

Copied to clipboard

Yuhua Zhou, Shichao Weng, Changhai Zhou, Yuhan Wu, Qian Qiao, Jun Gao, Fei Yang, Aimin Pan

Rethinking the Video Sampling and Reasoning Strategies for Temporal Sentence Grounding (2022.findings-emnlp)

Challenge:	Existing dynamic schemes such as early-exit and layer-drop reduce FLOPs but break batch processing or introduce KV-cache inconsistency.
Approach:	They propose a dynamic low-rank substitution framework that employs a lightweight decision module at each layer to dynamically determine the execution branch for different tokens.
Outcome:	The proposed model reduces computation by approximately 40% compared to the original dense model while outperforming existing baseline methods.

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

Doolittle: Benchmarks and Corpora for Academic Writing Formalization (2023.emnlp-main)

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.

Copied to clipboard

Shizhe Diao, Yongyu Lei, Liangming Pan, Tianqing Fang, Wangchunshu Zhou, Sedrick Keh, Min-Yen Kan, Tong Zhang

Towards Robust Temporal Activity Localization Learning with Noisy Labels (2024.lrec-main)

Challenge:	Existing methods of language refinement focus on narrow, specific linguistic features within isolated sentences, such as grammatical errors and improper word use.
Approach:	They propose a task to improve the overall quality of academic writing at paragraph level by integrating automatic feedback into the training process.
Outcome:	The proposed task improves the overall quality of formal academic writing at the paragraph level.

Copied to clipboard

Daizong Liu, Xiaoye Qu, Xiang Fang, Jianfeng Dong, Pan Zhou, Guoshun Nan, Keke Tang, Wanlong Fang, Yu Cheng

The Impact of Large Language Models in Academia: from Writing to Speaking (2025.findings-acl)

Challenge:	Existing methods for temporal activity localization are expensive and difficult to satisfy due to subjective labeling.
Approach:	They propose a new TAL setting where a TAL model should be robust to mixed training data with noisy moment boundaries.
Outcome:	The proposed method is significantly more robust to noisy training data than existing methods.

Copied to clipboard

Mingmeng Geng, Caixi Chen, Yanru Wu, Yao Wan, Pan Zhou, Dongping Chen

CIA: Inferring the Communication Topology from LLM-based Multi-Agent Systems (2026.acl-long)

Challenge:	Large language models (LLMs) are impacting human society, especially in textual information.
Approach:	They propose to build an automated monitoring platform to track the impact of large language models on human expression.
Outcome:	The results show that LLM-style words such as significant are used more frequently in abstracts and oral presentations.

Copied to clipboard

Yongxuan Wu, Xixun Lin, He Zhang, Nan Sun, Kun Wang, Chuan Zhou, Shirui Pan, Yanan Cao

RuleArena: A Benchmark for Rule-Guided Reasoning with LLMs in Real-World Scenarios (2025.acl-long)

Challenge:	LLM-based multi-agent systems (MAS) have demonstrated remarkable capabilities in solving complex tasks.
Approach:	They propose a communication inference attack that constructs new adversarial queries to induce intermediate agents’ reasoning outputs and models their semantic correlations through the global bias disentanglement and LLM-guided weak supervision.
Outcome:	The proposed attack achieves an average AUC of 0.87 and a peak AUC up to 0.99, revealing the privacy risk in MAS.

Copied to clipboard

Ruiwen Zhou, Wenyue Hua, Liangming Pan, Sitao Cheng, Xiaobao Wu, En Yu, William Yang Wang

Noise Robust Named Entity Understanding for Voice Assistants (2021.naacl-industry)

Challenge:	RuleArena assesses the ability of large language models (LLMs) to follow complex, real-world rules in reasoning.
Approach:	They propose a benchmark to evaluate the ability of large language models (LLMs) to follow complex, real-world rules in reasoning.
Outcome:	The proposed benchmark covers airline baggage fees, NBA transactions, and tax regulations.

Copied to clipboard

Deepak Muralidharan, Joel Ruben Antony Moniz, Sida Gao, Xiao Yang, Justine Kao, Stephen Pulman, Atish Kothari, Ray Shen, Yinying Pan, Vivek Kaul, Mubarak Seyed Ibrahim, Gang Xiang, Nan Dun, Yidan Zhou, Andy O, Yuan Zhang, Pooja Chitkara, Xuan Wang, Alkesh Patel, Kushal Tayal, Roger Zheng, Peter Grasch, Jason D Williams, Lin Li

SpiderFlow: Efficient Topology-Aware Scheduling for LLM Training Across Decentralized GPU Clusters (2026.acl-long)

Challenge:	Named Entity Recognition and Entity Linking are challenging for voice assistants . utterances are relatively short, so there is not much context to help disambiguate .
Approach:	They propose a Named Entity Understanding system that combines NER and EL in a joint reranking module.
Outcome:	The proposed framework improves NER accuracy by up to 3.13% and EL accuracy by 3.6% in F1 score . it also leads to better accuracies in other natural language understanding tasks .

Copied to clipboard

Zihan Chang, Shuibing He, Bo Zhou, Sheng Xiao, Siling Yang, Rui Wang, Zhe Pan

Information Extraction & Text Mining

Challenge:	Existing approaches to training large language models lack topologyaware task scheduling mechanisms and model parallelization strategies.
Approach:	They propose a topology-aware scheduling system specifically designed for decentralized GPU clusters . they propose heuristic methods at the inter-cluster level with ILP-based optimization within clusters.
Outcome:	The proposed system reduces job completion time by 1.2-1.3 and improves throughput by 1.12-1.25 . it also reduces scheduling overhead by 20-90 on average compared to state-of-the-art scheduling systems.

Takin-VC: Expressive Zero-Shot Voice Conversion via Adaptive Hybrid Content Encoding and Enhanced Timbre Modeling (2025.acl-long)

Copied to clipboard

Yang Yuguang, Yu Pan, Jixun Yao, Xiang Zhang, Jianhao Ye, Hongbin Zhou, Lei Xie, Lei Ma, Jianjun Zhao

Unlocking Exploration in RLVR: Uncertainty-aware Advantage Shaping for Deeper Reasoning (2026.findings-acl)

Challenge:	Expressive zero-shot voice conversion (VC) aims to modify source timbre to match unseen speaker . existing zero- shot VC systems struggle to reproduce paralinguistic information in highly expressive speech .
Approach:	They propose a framework for expressive zero-shot voice conversion that uses hybrid content encoding and memory-augmented context-aware timbre modeling.
Outcome:	The proposed framework surpasses state-of-the-art VC systems in speech naturalness, speaker similarity, and speaker similarness.

Copied to clipboard

Can Xie, Ruotong Pan, Xiangyu Wu, Zhang Yunfei, Jiayi Fu, Tingting Gao, Guorui Zhou

Towards Intrinsic Interpretability of Large Language Models: A Survey of Design Principles and Architectures (2026.acl-long)

Challenge:	Reinforcement Learning with Verifiable Rewards (RLVR) has shown significant promise for enhancing the reasoning capabilities of large language models (LLMs).
Approach:	They propose a model-free method that refines credit assignment by leveraging the model's internal uncertainty signals.
Outcome:	Extensive experiments on five mathematical reasoning benchmarks show that the proposed method outperforms strong RLVR baselines on multiple model scales, including 1.5B and 7B.

Copied to clipboard

Yutong Gao, Qinglin Meng, Yuan Zhou, Liangming Pan

Explainability & Interpretability in NLP

Challenge:	Existing studies on explainable AI focus on post-hoc explanation methods that interpret trained models through external approximations.
Approach:	They propose to categorize existing approaches into five design paradigms: functional transparency, concept alignment, representational decomposability, explicit modularization, and latent sparsity induction.
Outcome:	The proposed approaches are categorized into five design paradigms: functional transparency, concept alignment, representational decomposability, explicit modularization, and latent sparsity induction.

A Usage-centric Take on Intent Understanding in E-Commerce (2024.emnlp-main)

Copied to clipboard

Wendi Zhou, Tianyi Li, Pavlos Vougiouklis, Mark Steedman, Jeff Pan

Challenge:	Identifying and understanding user intents is a crucial task for E-Commerce.
Approach:	They propose to use intent understanding as a natural language reasoning task independent of product ontologies to identify and understand user intents.
Outcome:	The proposed framework can't be used to strongly align user intents with products with desirable properties and recommend useful products across diverse categories.

PEK: A Parameter-Efficient Framework for Knowledge-Grounded Dialogue Generation (2024.findings-acl)

Copied to clipboard

Pan Yang, Dandan Song, Zhijing Wu, Yanru Zhou

Entropy Ratio Clipping as a Soft Global Constraint for Stable Reinforcement Learning (2026.findings-acl)

Challenge:	Pre-trained language models have shown great dialogue generation capability in different scenarios, but the huge VRAM consumption when fine-tuning them is one of their drawbacks.
Approach:	They propose a parameter-efficient framework for knowledge-enhanced dialogue generation that leverages external knowledge documents and knowledge graphs to enhance its generation capabilities.
Outcome:	The proposed framework outperforms baseline methods on multiple evaluation metrics on Wizard of Wikipedia and CMU_DoG datasets.

Copied to clipboard

Zhenpeng Su, Leiyu Pan, Minxuan Lv, Tiehua Mei, Zijia Lin, Yuntao Li, Wenping Hu, Ruiming Tang, Kun Gai, Guorui Zhou

Wav-BERT: Cooperative Acoustic and Linguistic Representation Learning for Low-Resource Speech Recognition (2021.findings-emnlp)

Challenge:	Large language model post-training often adopts an off-policy training paradigm . however, the off-poliicy training model introduces distribution shifts that push the policy beyond the trust region.
Approach:	They propose to use the entropy ratio as a global metric to measure the relative change in policy exploration throughout updates.
Outcome:	Experiments show that the proposed metric improves performance across multiple benchmarks.

Copied to clipboard

Guolin Zheng, Yubei Xiao, Ke Gong, Pan Zhou, Xiaodan Liang, Liang Lin

Adaptive Proposal Generation Network for Temporal Sentence Localization in Videos (2021.emnlp-main)

Challenge:	Existing methods to learn the transfer from speech to text are unexplored . how to solve the representation discrepancy of speech and text is unexplorable .
Approach:	They propose a cooperative acoustic and linguistic representation learning method to fuse and utilize contextual information of speech and text.
Outcome:	The proposed method outperforms existing methods on low-resource speech recognition.

Copied to clipboard

Daizong Liu, Xiaoye Qu, Jianfeng Dong, Pan Zhou

Leveraging Unimodal Self-Supervised Learning for Multimodal Audio-Visual Speech Recognition (2022.acl-long)

Challenge:	Temporal sentence localization in videos is an important yet challenging task in natural language processing.
Approach:	They propose an Adaptive Proposal Generation Network to maintain the segment-level interaction while speeding up the efficiency.
Outcome:	The proposed model outperforms state-of-the-art methods on three challenging benchmarks.

Copied to clipboard

Xichen Pan, Peiyu Chen, Yichen Gong, Helong Zhou, Xinbing Wang, Zhouhan Lin

An Empirical Study on Parameter-Efficient Fine-Tuning for MultiModal Large Language Models (2024.findings-acl)

Challenge:	Existing methods for audio-visual speech recognition use extra data to increase performance . a recent study shows that the use of unimodal self-supervised learning improves performance on multimodal tasks.
Approach:	They propose to use unimodal self-supervised learning to train AVSR models on unlabelled unilateral data.
Outcome:	The proposed model improves on lip reading sentences 2 by 30% even without an external language model.

Copied to clipboard

Xiongtao Zhou, Jie He, Yuhua Ke, Guangyao Zhu, Victor Gutierrez Basulto, Jeff Pan

Reasoning Step-by-Step: Temporal Sentence Localization in Videos via Deep Rectification-Modulation Network (2020.coling-main)

Challenge:	Multimodal Large Language Models fine-tuned with multimodal instruction-following data have demonstrated formidable capabilities in multimodal tasks.
Approach:	They propose to employ four PEFT methods to fine-tune the LLM component of open-source MLLMs.
Outcome:	The proposed method is the best performing on seven datasets, while fine-tuning the connector layers leads to improved performance in most MLLMs.

Copied to clipboard

Daizong Liu, Xiaoye Qu, Jianfeng Dong, Pan Zhou

AiraXiv: An AI-Driven Open-Access Platform for Human and AI Scientists (2026.acl-demo)

Challenge:	Existing methods for temporal sentence localization in videos focus on visual content, but they are insufficient to model complex video contents.
Approach:	They propose a deep rectification-modulation network to correct attention misalignment . they use sentence information to capture frame-to-frame relation .
Outcome:	The proposed method achieves state-of-the-art performance on three public datasets.

Copied to clipboard

Junshu Pan, Panzhong Lu, Yixuan Weng, QiYao Sun, Fang Guo, Zijie Yang, Qiji Zhou, Yue Zhang

Challenge:	Recent advances in artificial intelligence (AI) have accelerated the growth of both human-authored and AI-generated research outputs.
Approach:	They propose an AI-driven open-access platform built on open preprints, AI-augmented analysis and review, and reader feedback.
Outcome:	The proposed platform supports human scientists through an interactive UI and AI scientists through Model Context Protocol (MCP)-based interactions.

Stealing Training Data from Large Language Models in Decentralized Training through Activation Inversion Attack (2025.acl-long)

Copied to clipboard

Chenxi Dai, Lin Lu, Pan Zhou

LexGenius: An Expert-Level Benchmark for Large Language Models in Legal General Intelligence (2026.findings-acl)

Challenge:	Decentralized training is a resource-efficient framework to democratize training of large language models.
Approach:	They propose an activation inversion attack to exploit privacy leakage from training data . they construct a shadow dataset comprising text labels and corresponding activations .
Outcome:	The proposed attack surface is based on a shadow dataset and public datasets . the proposed attack model reconstructs training data from activations in victim decentralized training.

Copied to clipboard

Wenjin Liu, Haoran Luo, Xin Feng, Xiang Ji, Lijuan Zhou, Rui Mao, Jiapu Wang, Shirui Pan, Erik Cambria

SOLAR-RL: Semi-Online Long-horizon Assignment Reinforcement Learning (2026.findings-acl)

Challenge:	Existing benchmarks for legal general intelligence (GI) are result-oriented and do not evaluate the legal intelligence of large language models (LLMs).
Approach:	They propose a Chinese legal benchmark for evaluating legal GI in large language models . they use recent legal cases and exam questions to create multiple-choice questions .
Outcome:	The proposed benchmarks lack a systematic evaluation of the legal intelligence of large language models (LLMs) the results show that even the best LLMs lagging behind human legal professionals.

Copied to clipboard

Jichao Wang, Liuyang Bian, Yufeng Zhou, Han Xiao, Yue Pan, Guozhi Wang, Hao Wang, Zhaoxiong Wang, Yafei Wen, Xiaoxin Chen, Shuai Ren, Lingfang Zeng

MoExtend: Tuning New Experts for Modality and Task Extension (2024.acl-srw)

Challenge:	Existing approaches to training GUI agents on dynamic tasks are based on SFT or Behavior Cloning.
Approach:	They propose a framework that integrates global trajectory insights directly into offline learning . they reconstruct diverse rollout candidates from static data and detect first failure point .
Outcome:	The proposed framework improves long-horizon task completion rates and robustness compared to baselines.

Copied to clipboard

Shanshan Zhong, Shanghua Gao, Zhongzhan Huang, Wushao Wen, Marinka Zitnik, Pan Zhou

GenTool: Enhancing Tool Generalization in Language Models through Zero-to-One and Weak-to-Strong Simulation (2025.findings-acl)

Challenge:	Existing instruction tuning methods for large language models (LLMs) are costly and difficult to implement.
Approach:	They propose a framework to streamline the modality adaptation and extension of Mixture-of-Experts (MoE) models.
Outcome:	The proposed framework enables rapid adaptation and extension to new modal data or tasks without tuning pretrained models.

Copied to clipboard

Jie He, Jennifer Neville, Mengting Wan, Longqi Yang, Hui Liu, Xiaofeng Xu, Xia Song, Jeff Z. Pan, Pei Zhou

Reward Generalization in RLHF: A Topological Perspective (2025.findings-acl)

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.

Copied to clipboard

Tianyi Alex Qiu, Fanzhi Zeng, Jiaming Ji, Dong Yan, Kaile Wang, Jiayi Zhou, Yang Han, Josef Dai, Xuehai Pan, Yaodong Yang

Challenge:	Existing alignment methods share a common topology of information flow, but their alternatives have not been thoroughly explored.
Approach:	They propose a theory of reward generalization in reinforcement learning from human feedback . they propose induced Bayesian networks to model the impact of dataset topologies on reward generalisation .
Outcome:	The proposed method achieves an average win rate of 65% on three NLP tasks.

CompKBQA: Component-wise Task Decomposition for Knowledge Base Question Answering (2025.emnlp-main)

Copied to clipboard

Yuhang Tian, Dandan Song, Zhijing Wu, Pan Yang, Changzhi Zhou, Jun Yang, Hao Wang, Huipeng Ma, Chenhao Li, Luan Zhang

Grounding Visual Illusions in Language: Do Vision-Language Models Perceive Illusions Like Humans? (2023.emnlp-main)

Challenge:	Existing knowledge base question answering methods struggle with complex queries.
Approach:	They propose a framework that optimizes the process of fine-tuning a LLM for generating logical forms by enabling it to learn relevant sub-tasks like skeleton generation, topic entity generation, and relevant relations generation.
Outcome:	The proposed framework achieves state-of-the-art on two benchmark KBQA datasets, WebQSP and CWQ.

Copied to clipboard

Yichi Zhang, Jiayi Pan, Yuchen Zhou, Rui Pan, Joyce Chai

Visual Data in NLP

Challenge:	Visual illusions are a phenomenon that is often seen in human perception but are not always faithful to the physical world.
Approach:	They build a dataset containing five types of visual illusions and formulate four tasks to examine visual illusion in state-of-the-art VLMs.
Outcome:	The proposed dataset reveals that larger models are closer to human perception and more susceptible to visual illusions.

Synapse: Empowering LLM Agents with Episodic-Semantic Memory via Spreading Activation (2026.findings-acl)

Copied to clipboard

Hanqi Jiang, Junhao Chen, Yi Pan, Ling Chen, Weihang You, Yifan Zhou, Ruidong Zhang, Yohannes Abate, Tianming Liu

Learning from Few Samples: A Novel Approach for High-Quality Malcode Generation (2025.emnlp-main)

Challenge:	Large Language Models excel at generalized reasoning, but lack the ability to accumulate experiences and maintain narrative coherence over long horizons.
Approach:	They propose a unified memory architecture that transcends static vector similarity.
Outcome:	The proposed model outperforms state-of-the-art methods in temporal and multihop reasoning tasks.

Copied to clipboard

Haijian Ma, Daizong Liu, Xiaowen Cai, Pan Zhou, Yulai Xie