Papers by Biao Zhang

Self-training Reduces Flicker in Retranslation-based Simultaneous Translation (2023.eacl-main)

Challenge:	Existing methods to improve NLP convergence and computational overhead are limited by stacking more layers.
Approach:	They propose a depth-scaled initialization method which reduces parameter variance at initialization and reduces output variance of residual connections to ease gradient back-propagation.
Outcome:	The proposed method outperforms the base model on translation tasks with five translation directions while matching the decoding speed of the baseline model.

Copied to clipboard

Sukanta Sen, Rico Sennrich, Biao Zhang, Barry Haddow

Machine Translation

Challenge:	Existing approaches to reduce flicker in simultaneous translation have increased the latency through masking and specialised inference, thus losing the simplicity of the approach.
Approach:	They propose to train a machine translation system to reduce flicker by controlling monotonicity and biased beam search to achieve the same flicker-latency tradeoff.
Outcome:	The proposed approach reduces flicker by controlling monotonicity while maintaining similar translation quality to the original.

LLMs Can Achieve High-quality Simultaneous Machine Translation as Efficiently as Offline (2025.findings-acl)

Copied to clipboard

Biao Fu, Minpeng Liao, Kai Fan, Chengxi Li, Liang Zhang, Yidong Chen, Xiaodong Shi

CTRAP: Embedding Collapse Trap to Safeguard Large Language Models from Harmful Fine-Tuning (2026.acl-long)

Challenge:	Large language models perform well in offline machine translation when the complete source sentence is provided . however, in many real scenarios, the source tokens arrive in a streaming manner and simultaneous machine translation is required .
Approach:	They propose a new paradigm that includes constructing supervised fine-tuning data for simultaneous machine translation (SiMT) to achieve SiMT, source and target tokens are rearranged into interleaved sequences, separated by special tokens according to varying latency requirements.
Outcome:	The proposed approach achieves state-of-the-art performance across various SiMT benchmarks and evaluation metrics while maintaining efficient auto-regressive decoding.

Copied to clipboard

Biao Yi, Tiansheng Huang, Baolei Zhang, Tong Li, Lihai Nie, Zheli Liu, Li Shen

SEAD: A Surrogate-free Label-only Membership Inference Attack against Pre-trained LLMs with Semantic-Aware Density (2026.findings-acl)

Challenge:	Fine-tuning-as-a-service exposes models to harmful fine-tuneing attacks . however, inherent general adaptability of LLMs allows them to bypass selective unlearning by rapidly relearning or repurposing their general capabilities for harmful tasks.
Approach:	They propose a paradigm shift that inducing model collapse instead of selective removal by relearning or repurposing general capabilities for harmful tasks.
Outcome:	The proposed model collapse mechanism neutralizes the very general capabilities that attackers exploit, tackling the core issue unaddressed by selective unlearning.

Copied to clipboard

Biao Yi, Jiahao Li, Yiming Li, Yu He, Baolei Zhang, Zheli Liu, Dacheng Tao

Robustness in NLP

Challenge:	Existing membership inference attacks require access to complete logits, but such access is often unavailable in real-world deployments where only the generated text is exposed.
Approach:	They propose a surrogate-free label-only MIA approach that directly estimates token probabilities through Monte Carlo sampling of the target model.
Outcome:	The proposed approach outperforms existing label-only attacks and serves as a foundational density estimator in the label-exclusive setting.

AFMRL: Attribute-Enhanced Fine-Grained Multi-Modal Representation Learning in E-commerce (2026.findings-acl)

Copied to clipboard

Biao Zhang, Lixin Chen, Bin Zhang, Wang Zongwei, Tong Liu, Bo Zheng

What’s Missing in Screen-to-Action? Towards a UI-in-the-Loop Paradigm for Multimodal GUI Reasoning (2026.findings-acl)

Challenge:	Multimodal representation is crucial for E-commerce tasks such as identical product retrieval.
Approach:	They propose an approach which leverages the generative power of Multimodal Large Language Models to extract key attributes from product images and text and enhances representation learning through a two-stage training framework.
Outcome:	The proposed model achieves state-of-the-art on multiple downstream retrieval tasks, validating the effectiveness of harnessing generative models to advance fine-grained representation learning.

Copied to clipboard

Songze Li, Xiaoke Guo, Tianqi Liu, Biao Yi, Zhaoyan Gong, Zhiqiang Liu, Huajun Chen, Wen Zhang

Prompt-Guided Internal States for Hallucination Detection of Large Language Models (2025.acl-long)

Challenge:	Existing GUI reasoning methods rely on direct screen-based decision-making, which lacks interpretability and overlooks a comprehensive understanding of UI elements, ultimately leading to task failure.
Approach:	They propose a GUI reasoning paradigm that treats the GUI reasoning task as a cyclic *Screen-UI elements-Action process.
Outcome:	The proposed paradigm achieves state-of-the-art UI understanding performance while yielding superior results in GUI reasoning tasks.

Copied to clipboard

Fujie Zhang, Peiqi Yu, Biao Yi, Baolei Zhang, Tong Li, Zheli Liu

Multi-Level Cross-Modal Alignment for Speech Relation Extraction (2024.emnlp-main)

Challenge:	Large Language Models (LLMs) generate incorrect or logically incorrect responses, which is known as LLM hallucinations.
Approach:	They propose a framework for supervised hallucination detection using in-domain data by prompting changes to the structure related to text truthfulness in LLMs’ internal states.
Outcome:	The proposed framework enhances the cross-domain generalization of existing hallucination detection methods.

Copied to clipboard

Liang Zhang, Zhen Yang, Biao Fu, Ziyao Lu, Liangying Shao, Shiyu Liu, Fandong Meng, Jie Zhou, Xiaoli Wang, Jinsong Su

Multilingual Document-Level Translation Enables Zero-Shot Transfer From Sentences to Documents (2022.acl-long)

Challenge:	Existing studies use synthetic speech to train and evaluate SpeechRE models, hindering their development . modality gap issue limits performance of existing models, limiting future researches .
Approach:	They propose to use speech data to train and evaluate SpeechRE models by using real speech . they propose to train a cross-modal alignment model to bridge the modality gap .
Outcome:	The proposed model can train to bridge the modality gap between speech encoder and text decoder . the proposed model is based on two real SpeechRE datasets .

Copied to clipboard

Biao Zhang, Ankur Bapna, Melvin Johnson, Ali Dabirmoghaddam, Naveen Arivazhagan, Orhan Firat

PresentAgent: Multimodal Agent for Presentation Video Generation (2025.emnlp-demos)

Challenge:	Document-level neural machine translation (DocNMT) is a powerful tool for integrating cross-sentence context into translations.
Approach:	They explore whether and how contextual modeling in DocNMT is transferable via multilingual modeling.
Outcome:	The proposed model can be used to transfer from teacher languages to student languages with no documents but sentence level data.

Copied to clipboard

Jingwei Shi, Zeyu Zhang, Biao Wu, Yanjie Liang, Meng Fang, Ling Chen, Yang Zhao

Accelerating Neural Transformer via an Average Attention Network (P18-1)

Challenge:	Existing methods for generating static slides or text summaries are limited to producing narrated presentations.
Approach:	They propose a multimodal agent that transforms long-form documents into narrated presentations.
Outcome:	The present agent produces fully synchronized visual and spoken content that closely mimics human-style presentations.

Copied to clipboard

Biao Zhang, Deyi Xiong, Jinsong Su

A Study in Improving BLEU Reference Coverage with Diverse Automatic Paraphrasing (2020.findings-emnlp)

Challenge:	Using parallelizable attention networks, the neural Transformer is slow to train due to auto-regressive architecture and self-attention in the decoder.
Approach:	They propose an average attention network to replace the original self-attention model in the decoder of the neural Transformer.
Outcome:	The proposed network can decode sentences over four times faster than the original version with almost no loss in training time and translation performance.

Copied to clipboard

Rachel Bawden, Biao Zhang, Lisa Yankovskaya, Andre Tättar, Matt Post

Efficient CTC Regularization via Coarse Labels for End-to-End Speech Translation (2023.eacl-main)

Challenge:	Using neural paraphrasing techniques, we investigate whether automatically generating additional diverse references can provide better coverage of the space of valid translations.
Approach:	They propose to use neural paraphrasing techniques to generate additional references that provide better coverage of the space of valid translations.
Outcome:	The proposed approach beats human paraphrases in the BLEU evaluation.

Copied to clipboard

Biao Zhang, Barry Haddow, Rico Sennrich

Speech & Audio in NLP

Challenge:	Developing techniques to support end-to-end speech translation is non-trivial because of the speech-text modality gap.
Approach:	They propose a coarse labeling approach that merges vocabulary labels via simple heuristic rules . they propose to use 256-bit truncation, division or modulo operations to regularize the encoder .
Outcome:	The proposed method can increase training efficiency while delivering better performance.

SMEC:Rethinking Matryoshka Representation Learning for Retrieval Embedding Compression (2025.emnlp-main)

Copied to clipboard

Biao Zhang, Lixin Chen, Tong Liu, Bo Zheng

ProphetNet-X: Large-Scale Pre-training Models for English, Chinese, Multi-lingual, Dialog, and Code Generation (2021.acl-demo)

Challenge:	Large language models generate high-dimensional embeddings that capture rich semantic and syntactic information.
Approach:	They propose a training framework to reduce dimensionality and complexity of large language models.
Outcome:	Experiments on image, text, and multimodal datasets show that the proposed training framework reduces dimensionality while maintaining performance.

Copied to clipboard

Weizhen Qi, Yeyun Gong, Yu Yan, Can Xu, Bolun Yao, Bartuer Zhou, Biao Cheng, Daxin Jiang, Jiusheng Chen, Ruofei Zhang, Houqiang Li, Nan Duan

A Lightweight Recurrent Network for Sequence Modeling (P19-1)

Challenge:	Existing models for pre-training are not convenient for users to find and set them up.
Approach:	They propose to extend ProphetNet into other domains and languages by pre-training models . they pre-train a cross-lingual generation model ProphetNet-Multi and a Chinese generation model .
Outcome:	The proposed models achieve new state-of-the-art on 10 benchmarks.

Copied to clipboard

Biao Zhang, Rico Sennrich

Green & Sustainable NLP

Challenge:	Recent studies show that recurrent networks suffer from severe computational inefficiency due to weak parallelization.
Approach:	They propose a lightweight recurrent network (LRN) that uses input and forget gates to handle long-range dependencies and gradient vanishing and explosion.
Outcome:	The proposed recurrent network yields the best running efficiency on six NLP tasks.

VaseVQA: Multimodal Agent and Benchmark for Ancient Greek Pottery (2026.findings-eacl)

Copied to clipboard

Jinchao Ge, Tengfei Cheng, Biao Wu, Zeyu Zhang, Shiya Huang, Judith Bishop, Gillian Shepherd, Meng Fang, Ling Chen, Yang Zhao

MMCLIP: Cross-Modal Attention Masked Modelling for Medical Language-Image Pre-Training (2026.acl-long)

Challenge:	MLLMs that use domain-specific data are limited in understanding cultural heritage artifacts such as ancient Greek pottery . supervised fine-tuning improves adaptation to domain knowledge, but it struggles with deeper reasoning tasks.
Approach:	They propose a visual question-answer tool that augments SFT with reinforcement learning using verifiable rewards.
Outcome:	The proposed model outperforms baseline models on reasoning-intensive questions on ancient Greek pottery.

Copied to clipboard

Biao Wu, Yutong Xie, Zeyu Zhang, Vu Minh Hieu Phan, Qi Chen, Ling Chen, Qi Wu

Adaptive Simultaneous Sign Language Translation with Confident Translation Length Estimation (2024.lrec-main)

Challenge:	Existing vision-and-language pretraining methods face challenges in reconstructing pathological features due to limited data.
Approach:	They propose a method that uses masked modeling to enhance visual and linguistic learning.
Outcome:	MMCLIP integrates unpaired data through disease-kind prompts to achieve state-of-the-art performance in zero-shot and fine-tuning across five benchmarks.

Copied to clipboard

Tong Sun, Biao Fu, Cong Hu, Liang Zhang, Ruiquan Zhang, Xiaodong Shi, Jinsong Su, Yidong Chen

Simplifying Neural Machine Translation with Addition-Subtraction Twin-Gated Recurrent Networks (D18-1)

Challenge:	Existing non-simultaneous sign language translation methods suffer from inherent inference delays in real-time scenarios.
Approach:	They propose an adaptive policy for simultaneous sign language translation that progressively converts incrementally received sign video into its corresponding natural sentence.
Outcome:	The proposed policy excels in situations requiring extremely low latency.

Copied to clipboard

Biao Zhang, Deyi Xiong, Jinsong Su, Qian Lin, Huiji Zhang

Adaptive Feature Selection for End-to-End Speech Translation (2020.findings-emnlp)

Challenge:	Existing gated recurrent networks have a vanishing gradient, allowing for more matrix transformations and less transparent functions.
Approach:	They propose an additionsubtraction twin-gated recurrent network (ATR) to simplify neural machine translation.
Outcome:	The proposed system is more transparent than LSTM/GRU due to the simplification.

Copied to clipboard

Biao Zhang, Ivan Titov, Barry Haddow, Rico Sennrich

Metric-guided Distillation: Distilling Knowledge from the Metric to Ranker and Retriever for Generative Commonsense Reasoning (2022.emnlp-main)

Challenge:	End-to-end speech translation (E2E) models that directly maps audio to a foreign text are not efficient.
Approach:	They propose adaptive feature selection (AFS) for encoder-decoder based E2E ST.
Outcome:	The proposed model outperforms the existing model on LibriSpeech En-Fr and MuST-C with a BLEU score of 18.56.

Copied to clipboard

Xingwei He, Yeyun Gong, A-Long Jin, Weizhen Qi, Hang Zhang, Jian Jiao, Bartuer Zhou, Biao Cheng, Sm Yiu, Nan Duan

Adaptive Threshold Selective Self-Attention for Chinese NER (2022.coling-1)

Challenge:	Existing work on commonsense generation requires models to have relational reasoning and compositional generalization capabilities.
Approach:	They propose a metric distillation rule to distill knowledge from a standard metric to a ranker and transfer it to re-ranking a retriever.
Outcome:	The proposed method surpasses the previous SOTA.

Copied to clipboard

Biao Hu, Zhen Huang, Minghao Hu, Ziwen Zhang, Yong Dou

On Sparsifying Encoder Outputs in Sequence-to-Sequence Models (2021.findings-acl)

Challenge:	Named entity recognition (NER) is a computationally difficult task in Chinese since there is no natural delimiter between words in sentences.
Approach:	They propose a data-driven Adaptive Threshold Selective Self-Attention mechanism to select the most relevant characters to enhance Transformer architecture for Chinese named entity recognition.
Outcome:	Experiments on four benchmark Chinese NER datasets show the proposed mechanism improves performance.

Copied to clipboard

Biao Zhang, Ivan Titov, Rico Sennrich

Exploring Dynamic Selection of Branch Expansion Orders for Code Generation (2021.acl-long)

Challenge:	Using sequence-to-sequence models, encoder outputs are usually transferred to the decoder for generation, but in this study, encoded outputs can be compressed to shorten the sequence for decoding.
Approach:	They propose to use a stochastic gate-based algorithm to mask encoder outputs to shorten the sequence delivered for decoding.
Outcome:	The proposed model can be used to shorten encoder outputs to short a sequence . the proposed model yields a speedup of up to 1.65 on document summarization and 1.20 on character-based machine translation tasks.

Copied to clipboard

Hui Jiang, Chulun Zhou, Fandong Meng, Biao Zhang, Jie Zhou, Degen Huang, Qingqiang Wu, Jinsong Su

Programming Languages in NLP

Challenge:	Existing code generation models model abstract syntax tree (AST) but not suitable for all multi-branch nodes.
Approach:	They propose to equip a Seq2Tree model with a branch selector to determine optimal expansion orders for multi-branch nodes.
Outcome:	The proposed model can determine optimal expansion orders of branches for multi-branch nodes.

Improving Massively Multilingual Neural Machine Translation and Zero-Shot Translation (2020.acl-main)

Copied to clipboard

Biao Zhang, Philip Williams, Ivan Titov, Rico Sennrich

WMT24++: Expanding the Language Coverage of WMT24 to 55 Languages & Dialects (2025.findings-acl)

Challenge:	Existing approaches to improve multilingual neural machine translation (NMT) are weak, and lack robustness to support language pairs with varying typological characteristics.
Approach:	They propose to deepen NMT models to support language pairs with varying typological characteristics by random online backtranslation.
Outcome:	The proposed approach narrows the performance gap with bilingual models and improves zero-shot performance by 10 BLEU, approaching conventional pivot-based methods.

Copied to clipboard

Daniel Deutsch, Eleftheria Briakou, Isaac Rayburn Caswell, Mara Finkelstein, Rebecca Galor, Juraj Juraska, Geza Kovacs, Alison Lui, Ricardo Rei, Jason Riesa, Shruti Rijhwani, Parker Riley, Elizabeth Salesky, Firas Trabelsi, Stephanie Winkler, Biao Zhang, Markus Freitag

Sparse Attention with Linear Units (2021.emnlp-main)

Challenge:	In order to evaluate large language models (LLMs), it is important to collect benchmark datasets in order to assess their multilingual performance.
Approach:	They extend the WMT24 dataset to cover 55 languages by collecting new human-written references and post-edits for 46 new languages/dialects.
Outcome:	The proposed dataset covers 55 languages and provides best-performing MT systems in all 55 languages.

Copied to clipboard

Biao Zhang, Ivan Titov, Rico Sennrich

Machine Translation

Challenge:	Recent studies have suggested that sparse attention mechanisms can be made more interpretable by replacing the softmax activation with its sparser variants.
Approach:	They propose a method to replace softmax activation with a ReLU to achieve sparsity in attention by layer normalization with either a specialized initialization or an additional gating function.
Outcome:	The proposed model is easy to implement and more efficient than previously proposed sparse attention mechanisms.

LLMRefine: Pinpointing and Refining Large Language Models via Fine-Grained Actionable Feedback (2024.findings-naacl)

Copied to clipboard

Wenda Xu, Daniel Deutsch, Mara Finkelstein, Juraj Juraska, Biao Zhang, Zhongtao Liu, William Yang Wang, Lei Li, Markus Freitag

OS Agents: A Survey on MLLM-based Agents for Computer, Phone and Browser Use (2025.acl-long)

Challenge:	Recent large language models (LLMs) are leveraging human feedback to improve their output quality. however, human feedback is costly to collect, especially at inference time when the model provides new, unseen input.
Approach:	They propose an inference-time optimization method to refine large language models' output based on fine-grained feedback to pinpoint defects and guide iterative refinement .
Outcome:	The proposed method consistently outperforms baseline approaches on three text generation tasks, including machine translation, long-form question answering, and topical summarization.

Copied to clipboard

Xueyu Hu, Tao Xiong, Biao Yi, Zishu Wei, Ruixuan Xiao, Yurun Chen, Jiasheng Ye, Meiling Tao, Xiangxin Zhou, Ziyu Zhao, Yuhuai Li, Shengze Xu, Shenzhi Wang, Xinchen Xu, Shuofei Qiao, Zhaokai Wang, Kun Kuang, Tieyong Zeng, Liang Wang, Jiwei Li, Yuchen Eleanor Jiang, Wangchunshu Zhou, Guoyin Wang, Keting Yin, Zhou Zhao, Hongxia Yang, Fan Wu, Shengyu Zhang, Fei Wu

Revisiting Low-Resource Neural Machine Translation: A Case Study (P19-1)

Challenge:	a new generation of (M)LLMs is enabling the creation of superintelligent AI assistants . OS Agents can complete tasks autonomously and have the potential to significantly enhance the lives of billions of users worldwide.
Approach:	They propose to build OS Agents that operate within operating systems' GUIs and GUIs . they examine evaluation metrics and benchmarks to identify promising directions .
Outcome:	The proposed agents are based on operating systems (OS) and operating systems frameworks.

Copied to clipboard

Rico Sennrich, Biao Zhang

BadActs: A Universal Backdoor Defense in the Activation Space (2024.findings-acl)

Challenge:	Recent research has shown that neural machine translation models are highly data-inefficient and underperform phrase-based statistical machine translation (PBSMT) in low-resource settings.
Approach:	They propose to use auxiliary data to train low-resource neural machine translation systems without auxiliary monolingual or multilingual data.
Outcome:	The proposed methods outperform PBSMT and other statistical machine translation models in Korean–English with minimal data.

Copied to clipboard

Biao Yi, Sishuo Chen, Yiming Li, Tong Li, Baolei Zhang, Zheli Liu

Signer Diversity-driven Data Augmentation for Signer-Independent Sign Language Translation (2024.findings-naacl)

Challenge:	Backdoor attacks pose an increasingly severe security threat to Deep Neural Networks . existing methods focused on the word space are ineffective against feature-space triggers - a recent study has shown .
Approach:	They propose a backdoor defense that purifies backdoor samples in the activation space . they aim to eliminate backdoor triggers while preserving the integrity of clean data .
Outcome:	The proposed method achieves state-of-the-art against backdoor attacks on clean data.

Copied to clipboard

Honghao Fu, Liang Zhang, Biao Fu, Rui Zhao, Jinsong Su, Xiaodong Shi, Yidong Chen