Papers by Xiaofeng Zhang

Taming "Zombie" Agents: A Markov State-Aware Framework for Resilient Multi-Agent Evolution (2026.acl-long)

Challenge:	Existing approaches of distantly supervised relation extraction (DSRE) focus on sentence-level or bag-level de-noising, neglecting the explicit interaction with cross levels.
Approach:	They propose a hierarchical contrastive learning framework for distantly supervised relation extraction to reduce noisy sentences.
Outcome:	The proposed framework outperforms baselines in various mainstream DSRE datasets.

Copied to clipboard

Taolin Zhang, Pukun Zhao, Qizhou Chen, Jiuheng Wan, Chen Chen, Xiaofeng He, Chengyu Wang, Richang Hong

KEHRL: Learning Knowledge-Enhanced Language Representations with Hierarchical Reinforcement Learning (2024.lrec-main)

Challenge:	Existing approaches to improve efficiency of multi-agent systems rely on aggressive graph topology evolution . however, such hard pruning overlooks the potential for "zombie" agents to recover and contribute in subsequent discussion rounds.
Approach:	They propose a Markov state-aware framework for resilient multi-agent evolution that manages agent collaboration through soft state transitions.
Outcome:	The proposed framework outperforms baselines and significantly reduces token consumption through state-aware agent scheduling.

Copied to clipboard

Dongyang Li, Taolin Zhang, Longtao Huang, Chengyu Wang, Xiaofeng He, Hui Xue

Revisiting and Advancing Chinese Natural Language Understanding with Accelerated Heterogeneous Knowledge Pre-training (2022.emnlp-industry)

Challenge:	General pre-trained language models (PLMs) leverage relation triples from knowledge graphs (KGs) and integrate external data sources into language models via self-supervised learning.
Approach:	They propose to learn Knowledge-Enhanced language representations with Hierarchical Reinforcement Learning (KEHRL) to detect positions for knowledge injection and integrate external knowledge into the model to avoid injecting inaccurate or irrelevant knowledge.
Outcome:	The proposed model can detect essential positions in texts for knowledge injection and integrate external knowledge into the model to avoid injecting inaccurate or irrelevant knowledge.

Copied to clipboard

Taolin Zhang, Junwei Dong, Jianing Wang, Chengyu Wang, Ang Wang, Yinghui Liu, Jun Huang, Yong Li, Xiaofeng He

Knowledge-Empowered Representation Learning for Chinese Medical Reading Comprehension: Task, Model and Resources (2021.findings-acl)

Challenge:	Existing knowledge-enhanced pre-trained language models (KEPLMs) can capture internal knowledge, but can't understand external background knowledge.
Approach:	They propose to use Chinese knowledge-enhanced pre-trained language models to improve context-aware representations via learning from structured relations in knowledge bases.
Outcome:	Experiments show that Chinese knowledge-enhanced pre-trained language models outperform strong baselines over various benchmark NLP tasks and in different model sizes.

Copied to clipboard

Taolin Zhang, Chengyu Wang, Minghui Qiu, Bite Yang, Zerui Cai, Xiaofeng He, Jun Huang

Enhancing Speech Large Language Models with Prompt-Aware Mixture of Audio Encoders (2025.emnlp-main)

Challenge:	MRC is a popular task in NLP, aiming to understand a passage and answer the relevant questions.
Approach:	They propose a multi-target machine learning task for the medical domain that predicts answers to medical questions and corresponding support sentences from medical information sources simultaneously.
Outcome:	The proposed model outperforms baselines by fusing context-aware and knowledge-awful token representations.

Copied to clipboard

Weiqiao Shan, Yuang Li, Yuhao Zhang, Yingfeng Luo, Chen Xu, Xiaofeng Zhao, Long Meng, Yunfei Lu, Min Zhang, Hao Yang, Tong Xiao, JingBo Zhu

TokenPenalty: Alleviating Attention Sinks and Positional Decay in LVLMs (2026.findings-acl)

Challenge:	Existing work on integrating audio encoders with large language models (LLMs) has focused on semantic understanding tasks, but different tasks may require distinct features that emphasize either semantic or acoustic aspects.
Approach:	They propose to use a prompt-aware mixture to enhance the Speech LLM that uses multiple audio encoders to extract different features based on the prompt.
Outcome:	The proposed approach outperforms all single-encoder Speech LLMs on ASR, speaker number verification, and AC tasks.

Copied to clipboard

Xiaofeng Zhang, Yuanchao Zhu, Qiyan Zhao, Xiaosong Yuan, Jiawei Cao, Xuhang Chen

MCS: An In-battle Commentary System for MOBA Games (2022.coling-1)

Challenge:	Multimodal large language models (MLLMs) often hallucinate due to two relevant phenomena: massive activation phenomenon and positional information decay.
Approach:	They propose a token-level intervention strategy that dynamically suppresses irrelevant visual tokens while preserving key contextual signals.
Outcome:	Experiments show that TokenTruth significantly improves factual consistency across MLLMs on standard image understanding benchmarks.

Copied to clipboard

Xiaofeng Qi, Chao Li, Zhongping Liang, Jigang Liu, Cheng Zhang, Yuanxin Wei, Lin Yuan, Guang Yang, Lanxiao Huang, Min Li

Challenge:	In-battle commentary is an important component of live streaming of e-sports competitions and is applicable to a wide range of scenarios like combat information analysis and live streaming.
Approach:	They propose a generative system for in-battle real-time commentary in mobile MOBA games and propose 'transform' method to convert match statistics and utterances into consistent encoding space.
Outcome:	The proposed system is based on real-time match statistics and events and can be used for live streaming, e-sports commentary and combat information analysis.

TRELM: Towards Robust and Efficient Pre-training for Knowledge-Enhanced Language Models (2024.lrec-main)

Copied to clipboard

Junbing Yan, Chengyu Wang, Taolin Zhang, Xiaofeng He, Jun Huang, Wei Zhang, Longtao Huang, Hui Xue

SalaMAnder: Shapley-based Mathematical Expression Attribution and Metric for Chain-of-Thought Reasoning (2025.findings-emnlp)

Challenge:	Existing methods for incorporating external knowledge into language models do not prioritize learning embeddings for entity-related tokens.
Approach:	They propose a framework for incorporating external knowledge into pre-training models that utilize entity-related tokens.
Outcome:	The proposed framework reduces pre-training time by 50% and outperforms other KEPLMs in knowledge probing tasks and multiple knowledge-aware language understanding tasks.

Copied to clipboard

Yue Xin, Chen Shen, Shaotian Yan, Xiaosong Yuan, Yaoming Wang, Xiaofeng Zhang, Chenxi Huang, Jieping Ye

From Redundancy to Relevance: Information Flow in LVLMs Across Reasoning Tasks (2025.naacl-long)

Challenge:	Chain-of-Thought prompting improves the math reasoning capability of large language models.
Approach:	They propose a method for attribution of component-level contributions in CoT reasoning using Shapley value and a stratified sampling algorithm that significantly reduces computational complexity.
Outcome:	The proposed method reduces computational complexity and provides robust correlations with model performance.

Copied to clipboard

Xiaofeng Zhang, Yihao Quan, Chen Shen, Xiaosong Yuan, Shaotian Yan, Liang Xie, Wenxiao Wang, Chaochen Gu, Hao Tang, Jieping Ye

QueueEDIT: Structural Self-Correction for Sequential Model Editing in LLMs (2026.findings-acl)

Challenge:	Large visionlanguage models (LVLMs) are a powerful visual-language reasoning tool.
Approach:	They propose to integrate attention analysis with LLaVA-CAM to determine interactions between visual representations.
Outcome:	The proposed approach can be used to determine interactions between visual representations.

Copied to clipboard

Taolin Zhang, Haidong Kang, Dongyang Li, Qizhou Chen, Xiaofeng He, Chengyu Wang, Richang Hong

Clustering and Ranking: Diversity-preserved Instruction Selection through Expert-aligned Quality Estimation (2024.emnlp-main)

Challenge:	Recent studies have shown that large language models (LLMs) can be effective for correcting factual inaccuracies but can still suffer from hallucinations.
Approach:	They propose a queue-based self-correction framework that addresses parameter bias during sequential model editing.
Outcome:	The proposed framework outperforms baseline models while maintaining competitive performance in single-turn editing.

Copied to clipboard

Yuan Ge, Yilun Liu, Chi Hu, Weibin Meng, Shimin Tao, Xiaofeng Zhao, Mahong Xia, Zhang Li, Boxing Chen, Hao Yang, Bei Li, Tong Xiao, JingBo Zhu

Logical Consistency is Vital: Neural-Symbolic Information Retrieval for Negative-Constraint Queries (2025.findings-acl)

Challenge:	Existing methods for instruction data selection have limitations such as relying on fragile external APIs, being affected by biases in GPT models, or reducing the diversity of the selected instruction dataset.
Approach:	They propose an industrial-friendly, expert-aligned and diversity-preserved instruction data selection method: Clustering and Ranking (CaR).
Outcome:	The proposed method outperforms Alpaca's existing methods by 32.1% in GPT-4 evaluations.

Copied to clipboard

Ganlin Xu, Zhoujia Zhang, Wangyi Mei, Jiaqing Liang, Weijia Lu, Xiaodong Zhang, Zhifei Yang, Xiaofeng Ma, Yanghua Xiao, Deqing Yang

Trident: Self-Supervised Preference Alignment via Triplet Regularization (2026.findings-acl)

Challenge:	Current dense retrieval methods compute similarities between dense vectors but overlook the real query intents.
Approach:	They propose a neuro-symbolic information retrieval method that leverages first-order logic to optimize the embeddings of naive natural language by considering the logical consistency between queries and documents.
Outcome:	The proposed method outperforms existing methods on negative-constraint queries under zero-shot and low-resource retrieval tasks.

Copied to clipboard

Yingnan Guo, Kejia Chen, Xiaofeng Zhang, Zifei Wu, Yu Zhang

Lifelong Knowledge Editing for LLMs with Retrieval-Augmented Continuous Prompt Learning (2024.emnlp-main)

Challenge:	Large vision-Language Models suffer from noisy supervision and semantic ambiguity in self-supervised settings.
Approach:	They propose a self-supervised framework that constructs reliable preference triplets . they propose 'trident' objective that enforces semantic separation between the triplet components .
Outcome:	The proposed framework outperforms state-of-the-art RLHF and RLAIF benchmarks on LLaVA-1.5-7B and achieves 95.70% precision on POPE using only 4k self-generated triplets and a single epoch.

Copied to clipboard

Qizhou Chen, Taolin Zhang, Xiaofeng He, Dongyang Li, Chengyu Wang, Longtao Huang, Hui Xue’

CulturalFrames: Assessing Cultural Expectation Alignment in Text-to-Image Models and Evaluation Metrics (2025.findings-emnlp)

Challenge:	Existing methods to correct outdated or erroneous knowledge in large language models (LLMs) are slow and cumbersome, resulting in catastrophic knowledge forgetting and degradation of model performance.
Approach:	They propose a RetriEval-augmented ContInuous Prompt lEarning method that converts knowledge statements into short and informative continuous prompts, prefixed to the LLM’s input query embedding.
Outcome:	The proposed method improves the performance of large language models (LLMs) while maintaining the overall performance of the model.

Copied to clipboard

Shravan Nayak, Mehar Bhatia, Xiaofeng Zhang, Verena Rieser, Lisa Anne Hendricks, Sjoerd Van Steenkiste, Yash Goyal, Karolina Stanczak, Aishwarya Agrawal

How to Determine the Most Powerful Pre-trained Language Model without Brute Force Fine-tuning? An Empirical Survey (2023.findings-emnlp)

Challenge:	CulturalFrames is a benchmark designed for rigorous human evaluation of cultural representation in visual generations.
Approach:	They propose to quantify the alignment of T2I models and evaluation metrics with respect to both explicit (stated) and implicit (unstated, implied by the prompt’s cultural context) cultural expectations.
Outcome:	The proposed model is based on 983 prompts, 3637 images and 10k human annotations from 10 countries and 5 socio-cultural domains.

Copied to clipboard

Jun Bai, Xiaofeng Zhang, Chen Li, Hanhua Hong, Xi Xu, Chenghua Lin, Wenge Rong

CausalGaze: Unveiling Hallucinations via Counterfactual Graph Intervention in Large Language Models (2026.findings-acl)

Challenge:	Transferability estimation has been a topic of great interest in computer vision fields . a lack of a comprehensive comparison between these estimation methods is a problem .
Approach:	They conduct a thorough survey of existing methods to find the most suitable model . they also outline difficulties of consideration of training details and applicability to text generation .
Outcome:	The proposed methods perform well with superiorities in effectiveness and efficiency.

Copied to clipboard

Linggang Kong, Lei Wu, Yunlong Zhang, Xiaofeng Zhong, Wang Zhen, Yongjie Wang, Yao Pan

DoCIA: An Online Document-Level Context Incorporation Agent for Speech Translation (2025.findings-acl)

Challenge:	Existing classification-based methods capture noise and spurious correlations while overlooking the underlying causal mechanisms.
Approach:	They propose a hallucination detection framework based on structural causal models that captures static and passive signals from internal states and employs counterfactual interventions to disentangle causal reasoning paths from incidental noise.
Outcome:	Experiments on four datasets and three widely used LLMs show that the proposed framework improves AUROC and interpretability.

Copied to clipboard

Xinglin Lyu, Wei Tang, Yuang Li, Xiaofeng Zhao, Ming Zhu, Junhui Li, Yunfei Lu, Min Zhang, Daimeng Wei, Hao Yang, Min Zhang

Text Generation

Challenge:	Document-level context is crucial for speech translation due to noise from ASR . incorporating document-level contextual information into ST remains a challenge .
Approach:	They develop an online framework that integrates document-level context into machine translation . they use document-based modules to integrate document- level context into ST .
Outcome:	The proposed framework outperforms baselines in sentence and discourse metrics . it can correct ASR transcription errors and improve translation performance .

Debt Collection Negotiations with Large Language Models: An Evaluation System and Optimizing Decision Making with Multi-Agent (2025.findings-acl)

Copied to clipboard

Xiaofeng Wang, Zhixin Zhang, Jin Guang Zheng, Yiming Ai, Rui Wang

From Complex to Simple: Unraveling the Cognitive Tree for Reasoning with Small Language Models (2023.findings-emnlp)

Challenge:	Debt collection negotiations (DCN) are vital for managing non-performing loans (NPLs) prior systems lacking dynamic negotiation and real-time decision-making capabilities.
Approach:	They propose a framework for debt negotiation that incorporates planning and judging modules to improve decision rationality.
Outcome:	The proposed framework improves decision rationality and integrates planning and judging modules to improve decision rationalness.

Copied to clipboard

Yan Junbing, Chengyu Wang, Taolin Zhang, Xiaofeng He, Jun Huang, Wei Zhang

On the Role of Long-tail Knowledge in Retrieval Augmented Large Language Models (2024.acl-short)

Challenge:	Existing methods to solve complex logical reasoning problems are cumbersome for language models.
Approach:	They propose to use iterative methodology to construct a cognitive tree using language models . they propose to generate multiple responses by utilizing in-context examples .
Outcome:	The proposed model achieves a performance level comparable to that of GPT-3.5 . the proposed model contains fewer parameters than 5% of the model with 175B parameters .

Copied to clipboard

Dongyang Li, Junbing Yan, Taolin Zhang, Chengyu Wang, Xiaofeng He, Longtao Huang, Hui Xue’, Jun Huang

Learning Knowledge-Enhanced Contextual Language Representations for Domain Natural Language Understanding (2023.emnlp-main)

Challenge:	Existing RAG methods focus on improving the task performance, without fine-grained process of knowledge.
Approach:	They propose a method that detects long-tail knowledge in large language models by analyzing retrieved documents and enhancing queries indiscriminately with retrieved information.
Outcome:	The proposed method achieves over 4x speedup in average inference time and consistent performance improvement in downstream tasks compared to existing pipelines.

Copied to clipboard

Taolin Zhang, Ruyao Xu, Chengyu Wang, Zhongjie Duan, Cen Chen, Minghui Qiu, Dawei Cheng, Xiaofeng He, Weining Qian

TRIDENT: Enhancing Large Language Model Safety with Tri-Dimensional Diversified Red-Teaming Data Synthesis (2025.acl-long)

Challenge:	Existing methods for pre-training KEPLMs with relational triples are difficult to adapt to close domains due to the lack of sufficient domain graph semantics.
Approach:	They propose a Knowledge-enhanced language representation learning framework for various closed domains that captures the implicit graph structure among the entities.
Outcome:	The proposed framework outperforms existing methods for pre-training KEPLMs in closed domains significantly.

Copied to clipboard

Xiaorui Wu, Xiaofeng Mao, Fei Li, Xin Zhang, Xuanhong Li, Chong Teng, Donghong Ji, Zhuang Li

ART: Attention Replacement Technique to Improve Factuality in LLMs (2026.acl-long)

Challenge:	Large Language Models (LLMs) excel in natural language processing tasks but are vulnerable to harmful content and being exploited for malicious purposes.
Approach:	They propose a framework to measure the risk coverage of alignment datasets across three dimensions: Lexical Diversity, Malicious Intent, and Jailbreak Tactics.
Outcome:	The proposed framework measures risk coverage across Lexical Diversity, Malicious Intent, and Jailbreak Tactics.

Copied to clipboard

Ziqin Luo, Yihao Quan, Xiaofeng Zhang, Xiaosong Yuan, Chen Shen

Prompt-Based Meta-Learning For Few-shot Text Classification (2022.emnlp-main)

Challenge:	Existing methods to mitigate hallucinations in large language models are expensive and require significant resources.
Approach:	They propose a training-free method that replaces uniform attention patterns in shallow layers with local attention patterns to reduce hallucinations.
Outcome:	The proposed method reduces hallucinations across multiple LLM architectures.

Copied to clipboard

Haoxing Zhang, Xiaofeng Zhang, Haibo Huang, Lei Yu

Diffusion-CAM: Faithful Visual Explanations for dMLLMs (2026.acl-long)

Challenge:	Existing methods to learn text labels require large amounts of data to build many few-shot tasks.
Approach:	They propose a Prompt-Based Meta-Learning model that adds the prompting mechanism to the meta-learning method.
Outcome:	The proposed method improves on four text classification datasets with high accuracy and robustness.

Copied to clipboard

Haomin Zuo, Yidi Li, Luoxiao Yang, Xiaofeng Zhang

Explainability & Interpretability in NLP

Challenge:	Existing Class Activation Mapping methods are ill-suited for interpreting non-autoregressive behaviors of diffusion-based architectures.
Approach:	They propose to use a method to generate parallel activation maps by probing intermediate representations in the transformer backbone to capture latent features and their class-specific gradients.
Outcome:	Experiments show that Diffusion-CAM significantly outperforms SoTA methods in localization accuracy and visual fidelity.

BELLE: A Bi-Level Multi-Agent Reasoning Framework for Multi-Hop Question Answering (2025.acl-long)

Copied to clipboard

Taolin Zhang, Dongyang Li, Qizhou Chen, Chengyu Wang, Xiaofeng He

Question Answering

Challenge:	Existing studies on multi-hop question answering employ specific methods regardless of question types . complexity of multihop question answerrs often exceeds knowledge boundaries of LLMs .
Approach:	They propose a framework that uses chain-of-thought prompting to prompt LLMs to answer multi-hop questions.
Outcome:	The proposed framework outperforms baseline models in multi-hop QA scenarios.

SMedBERT: A Knowledge-Enhanced Pre-trained Language Model with Structured Semantics for Medical Text Mining (2021.acl-long)

Copied to clipboard

Taolin Zhang, Zerui Cai, Chengyu Wang, Minghui Qiu, Bite Yang, Xiaofeng He

SConU: Selective Conformal Uncertainty in Large Language Models (2025.acl-long)

Challenge:	Existing knowledge-based PLMs are based on linked-entity information, but they only use linked-enemy information as auxiliary information.
Approach:	They propose to integrate semantic knowledge from neighbours of linked-entity into a medical PLM that integrates heterogeneous-entities into the homogeneously neighbouring entity structure.
Outcome:	Experiments show that SMedBERT outperforms baselines in knowledge-intensive Chinese medical tasks.

Copied to clipboard

Zhiyuan Wang, Qingni Wang, Yue Zhang, Tianlong Chen, Xiaofeng Zhu, Xiaoshuang Shi, Kaidi Xu

A Novel Global Feature-Oriented Relational Triple Extraction Model based on Table Filling (2021.emnlp-main)

Challenge:	Existing frameworks fail to identify outliers that violate the exchangeability assumption, leading to unbounded miscoverage rates and unactionable prediction sets.
Approach:	They propose a method that implements significance tests to determine whether a given sample deviates from the uncertainty distribution of the calibration set.
Outcome:	The proposed approach facilitates rigorous management of miscoverage rates across single-domain and interdisciplinary contexts, and enhances the efficiency of predictions.

Copied to clipboard

Feiliang Ren, Longhui Zhang, Shujuan Yin, Xiaofeng Zhao, Shilei Liu, Bochao Li, Yaduo Liu

Information Extraction & Text Mining

Challenge:	Table filling based relational triple extraction methods focus on using local features but ignore the global associations of relations and token pairs, which increases the possibility of overlooking some important information during triple extraction.
Approach:	They propose a global feature-oriented triple extraction model that makes full use of the two kinds of global associations of relations and token pairs.
Outcome:	The proposed model achieves state-of-the-art on three benchmark datasets.

Fixing Semantic Blind Spots in Anchor Tokens of dMLLMs (2026.findings-acl)

Copied to clipboard

Ruixuan Xu, Jiexi Xu, Qiyan Zhao, Xiaofeng Zhang

A Three-Stage Learning Framework for Low-Resource Knowledge-Grounded Dialogue Generation (2021.emnlp-main)

Challenge:	Autoregressive models (ARMs) are prone to hallucinations due to their sequential text generation and high latency.
Approach:	They propose a training-free decoding strategy that augments the attention key space with a static, distance-aware matrix to reduce the attention sink effect on semantic anchors.
Outcome:	The proposed method reduces the attention sink effect on semantic anchors while enhancing their ability to aggregate global visual information.

Copied to clipboard

Shilei Liu, Xiaofeng Zhao, Bochao Li, Feiliang Ren, Longhui Zhang, Shujuan Yin

Dialogue Systems & Conversational Agents

Challenge:	Existing knowledge-grounded dialogues perform poorly when transfer into new domains with limited training samples.
Approach:	They propose a weakly supervised three-stage learning framework based on weakly-supervised learning based upon large scale ungrounded dialogues and unstructured knowledge base.
Outcome:	The proposed framework outperforms state-of-the-art methods even in zero-resource setting.

Building the Directed Semantic Graph for Coherent Long Text Generation (2021.emnlp-main)

Copied to clipboard

Ziao Wang, Xiaofeng Zhang, Hongwei Du

ConU: Conformal Uncertainty in Large Language Models with Correctness Coverage Guarantees (2024.findings-emnlp)

Challenge:	Existing methods for conditional long text generation ignore the coherence issue of the generated texts.
Approach:	They propose a two-stage approach to generate coherent long text based on short input text . they first build a document-level path for each output text with each sentence embedding as its node .
Outcome:	The proposed approach is superior to state-of-the-art approaches on three real-world datasets.

Copied to clipboard

Zhiyuan Wang, Jinhao Duan, Lu Cheng, Yue Zhang, Qingni Wang, Xiaoshuang Shi, Kaidi Xu, Heng Tao Shen, Xiaofeng Zhu

ShieldHead: Decoding-time Safeguard for Large Language Models (2025.findings-acl)

Challenge:	Uncertainty quantification (UQ) in natural language generation tasks remains an open challenge . however, black-box uncertainty measures require investigating with the proliferation of LLMs served via APIs.
Approach:	They propose a conformal uncertainty measure and a method to transform heuristic uncertainty notions into rigorous prediction sets.
Outcome:	Empirical results show that the proposed method outperforms state-of-the-art methods and can provide reliable guarantees for open-ended NLG tasks.

Copied to clipboard

Zitao Xuan, Xiaofeng Mao, Da Chen, Xin Zhang, Yuhan Dong, Jun Zhou

Mixture of Attention Heads: Selecting Attention Heads Per Token (2022.emnlp-main)

Challenge:	Recent advances in LLM-based moderation methods have demonstrated remarkable promise in identifying safety risks associated with both inputs and outputs in human-AI interactions.
Approach:	They propose to learn a classification head on the last-layer hidden states of a dialogue model and use it to detect harmful content.
Outcome:	The proposed framework is 300 faster (1ms) than previous LLM-based moderation models with 99% less parameters than LlamaGuard.

Copied to clipboard

Xiaofeng Zhang, Yikang Shen, Zeyu Huang, Jie Zhou, Wenge Rong, Zhang Xiong

Shallow Focus, Deep Fixes: Enhancing Shallow Layers Vision Attention Sinks to Alleviate Hallucination in LVLMs (2025.emnlp-main)

Challenge:	Mixture-of-Experts (MoE) networks have been proposed as an efficient way to scale up model capacity and implement conditional computing.
Approach:	They propose a new architecture that combines multi-head attention with the MoE mechanism and a sparsely gated architecture that allows for faster computations.
Outcome:	The proposed architecture can scale up the number of attention heads and the number parameters while preserving computational efficiency.

Copied to clipboard

Xiaofeng Zhang, Yihao Quan, Chen Shen, Chaochen Gu, Xiaosong Yuan, Shaotian Yan, Jiawei Cao, Hao Cheng, Kaijie Wu, Jieping Ye

AMATA: Adaptive Multi-Agent Trajectory Alignment for Knowledge-Intensive Question Answering (2026.acl-long)

Challenge:	Multimodal large language models (MLLMs) demonstrate excellent abilities for understanding visual information, but the hallucination remains a challenging problem.
Approach:	They propose a training-free approach to enhance vision attention sinks to facilitate convergence of the image token attention sink within shallow layers.
Outcome:	The proposed approach improves the convergence of the image token attention sink within shallow layers and strengthens the layer’s focus on the image itself.

Copied to clipboard

Taolin Zhang, Dongyang Li, Chen Chen, Qizhou Chen, Jiuheng Wan, Xiaofeng He, Chengyu Wang, Richang Hong

Question Answering

Challenge:	Empirical results show that AMATA outperforms baseline approaches, knowledge-augmented frameworks, and LLMs on knowledge-intensive QA benchmarks.
Approach:	They propose an Adaptive Multi-Agent Trajectory Alignment framework that integrates external knowledge to improve response interpretability and factual grounding.
Outcome:	The proposed framework outperforms baseline approaches, knowledge-augmented frameworks, and LLM-based trajectory systems on five established knowledge-intensive QA benchmarks.

DAPE-BR: Distance-Aware Positional Encoding for Mitigating Object Hallucination in LVLMs (2025.findings-emnlp)

Copied to clipboard

Mingrui Xie, Tianxiang Xu, Qianhai Tang, Shanming Yao, Xiaofeng Zhang, Junliang Du

Reasoning Fails Where Step Flow Breaks (2026.acl-long)

Challenge:	Large Vision–Language Models (LVLMs) suffer from object hallucination, generating descriptions for objects that are absent from the image, which undermines reliability and hinders real-world deployment.
Approach:	They propose a positional-alignment scheme that preserves pretrained weight order while globally—- visual–text distances, embeds an isotropic fused patch-distance metric, and applies a patch-delay causal mask to enforce spatial causality.
Outcome:	Extensive experiments on POPE, MMStar and SQA show that DAPE-BR reduces hallucinations and boosts performance.

Copied to clipboard

Xiaoyu Xu, Yulan Pan, Xiaosong Yuan, Zhihong Shen, Minghao Su, Yuanhao Su, Xiaofeng Zhang

WildFeedback: Aligning LLMs With In-situ User Interactions And Feedback (2026.acl-long)

Challenge:	Existing analysis tools struggle with long chain of thought traces.
Approach:	They propose a saliency-inspired test-time intervention that adjusts shallow saliencies to improve accuracy on math, science, and coding tasks.
Outcome:	The proposed model improves accuracy on math, science, and coding tasks without retraining.

Copied to clipboard

Taiwei Shi, Zhuoer Wang, Longqi Yang, Ying-Chun Lin, Zexue He, Mengting Wan, Pei Zhou, Sujay Kumar Jauhar, Sihao Chen, Shan Xia, Hongfei Zhang, Jieyu Zhao, Xiaofeng Xu, Xia Song, Jennifer Neville

Challenge:	Traditional alignment methods rely on human annotations and are subjective and misalignment with real-world user preferences.
Approach:	They propose a framework that leverages in-situ user feedback during conversations with LLMs to create preference datasets automatically.
Outcome:	The proposed framework identifies and classifies user feedback to LLM responses between conversation turns and creates examples of preferred and dispreferred responses according to user preferences.

UniPSDA: Unsupervised Pseudo Semantic Data Augmentation for Zero-Shot Cross-Lingual Natural Language Understanding (2024.lrec-main)

Copied to clipboard

Dongyang Li, Taolin Zhang, Jiali Deng, Longtao Huang, Chengyu Wang, Xiaofeng He, Hui Xue

Low-Resource NLP

Challenge:	Existing studies rely on shallow unsupervised data generated by token surface matching regardless of global context-aware semantics of the surrounding text tokens.
Approach:	They propose an Unsupervised Pseudo Semantic Data Augmentation mechanism to enrich training data without human intervention.
Outcome:	The proposed model improves on general zero-shot cross-lingual understanding tasks on different languages without human intervention.

Unlocking the Black Box of Latent Reasoning: An Interpretability-Guided Approach to Intervention (2026.acl-long)

Copied to clipboard

Shuochen Chang, Tong Bai, Xiaofeng Zhang, Qianli Ma, Qingyang Liu, Zhaohe Liao, Yibo Miao, Li Niu