Papers by Kun Liang

16 papers
Backdoor Collapse: Eliminating Unknown Threats Via Known Backdoor Aggregation In Language Models (2026.acl-long)

Copied to clipboard

Challenge: Existing defenses rely on impractical assumptions about trigger settings to mitigate backdoor attacks . a recent study found that small amounts of training data can systematically induce harmful behaviors in large language models.
Approach: They propose a backdoor defense framework that requires no prior knowledge of trigger settings . they use a two-stage process to aggregate backdoor representations and fine-tune recovery .
Outcome: The proposed defense reduces the average Attack Success Rate to 4.41% across multiple benchmarks . the proposed framework generalizes across different types of backdoors, confirming its robustness in practical deployment scenarios.
C-World: A Computer Use Agent Environment Creator (2026.acl-long)

Copied to clipboard

Challenge: C-World enables users to build agent environments on demand.
Approach: They propose a system that enables users to build agent environments on demand.
Outcome: The proposed system outperforms baselines on 119k samples and achieves Spearman = 0.883 ranking correlation with real execution.
Emphasising Structured Information: Integrating Abstract Meaning Representation into LLMs for Enhanced Open-Domain Dialogue Evaluation (2025.findings-emnlp)

Copied to clipboard

Challenge: Existing evaluation metrics struggle to evaluate adversarial negative examples . existing metrics struggle in handling adversarials, resulting in low correlations with human judgments.
Approach: They propose a framework that integrates AMR and domain-specific language models for automatic open-domain dialogue evaluation.
Outcome: The proposed evaluation framework achieves strong correlations with human judgments across multiple datasets.
Mind Reader: Latent User Demand-Guided Content Optimization for Generative Search Engine (2026.acl-long)

Copied to clipboard

Challenge: Generative Search Engines (GSEs) have reshaped information retrieval and Generating Engine Optimization (GEO) emerges to improve the content visibility in GSEs’ responses.
Approach: They propose a method to optimize content to cover latent semantic information of GSEs by decomposing query into diverse perspectives and capturing underlying semantic information.
Outcome: The proposed method outperforms baselines and effectively improves content visibility (with up to 2.44x objective metrics and 1.23x subjective metrics on average).
RSA-Bench: Benchmarking Audio Large Models in Real-World Acoustic Scenarios (2026.findings-acl)

Copied to clipboard

Challenge: Existing evaluations rely on synthetic Gaussian noise or simplistic single-source interference, failing to capture the intricate, multi-layered acoustic dynamics that characterize authentic physical environments.
Approach: They propose a robustness benchmark to stress-test Audio Large Models (ALLMs) using high-fidelity auditory scene simulations.
Outcome: The proposed model performs well on a wide range of tasks, including automatic speech recognition, speech translation, and audio-based reasoning.
SLIDE: A Framework Integrating Small and Large Language Models for Open-Domain Dialogues Evaluation (2024.findings-acl)

Copied to clipboard

Challenge: Existing approaches to evaluate open domain dialogues have a one-to-many problem . existing approaches lack commonsense reasoning biases and perform poorly in domain-specific scenarios.
Approach: They propose a framework that leverages both a small, specialised model and LLMs for the evaluation of open-domain dialogues.
Outcome: The proposed framework achieves state-of-the-art performance in both classification and evaluation tasks and exhibits better correlation with human judgements.
Selection Bias Explorations and Debias Methods for Natural Language Sentence Matching Datasets (P19-1)

Copied to clipboard

Challenge: Natural Language Sentence Matching (NLSM) is a popular NLP task.
Approach: They propose to use QuoraQP to train and evaluate NLSM models using a selection bias framework.
Outcome: The proposed framework can improve generalization ability of trained models and give more trustworthy evaluation results for real-world adoptions.
iAgent: LLM Agent as a Shield between User and Recommender Systems (2025.findings-acl)

Copied to clipboard

Challenge: Traditional recommender systems focus on the user-platform paradigm, where users are directly exposed under the control of the platform's recommendation algorithms.
Approach: They propose a user-agent-platform paradigm where agent serves as the protective shield between user and recommender system that enables indirect exposure.
Outcome: The proposed model improves 16.6% over baselines on four datasets and mitigates echo chamber effects and reduces model bias in disadvantaged users.
Extracting and Combining Abilities For Building Multi-lingual Ability-enhanced Large Language Models (2025.emnlp-main)

Copied to clipboard

Challenge: Existing work relies on training with multi-lingual ability-related data, which may not be available for low-resource languages.
Approach: They propose a multi-lingual ability-enhanced LLM that extracts language-agnostic ability-related weights from LLMs and combine them across different languages by simple addition and subtraction operations without training.
Outcome: The proposed approach extracts language-agnostic ability-related weights from LLMs and combine them across different languages without training.
Question Tells You Where the Answer Is: Intention-aware Long-Context KV Cache Compression (2026.acl-long)

Copied to clipboard

Challenge: Recent methods to reduce the KV cache size fail to identify crucial KVs for generation while excluding others accurately, resulting in severe information loss.
Approach: They propose an intention-aware KV cache eviction method that identifies and retains crucial KVs according to the attention distribution of intention, which semantically reflects the user’s goal and determines which part of the context is relevant.
Outcome: The proposed method can maintain the model performance while reducing the KV cache size from 128K to 2K, leading to a 6.3x increase in decoding speed and 7.8x enhancement in memory efficiency compared to the default setting.
HearSay Benchmark: Do Audio LLMs Leak What They Hear? (2026.findings-acl)

Copied to clipboard

Challenge: Recent advances in audio large language models have led to their potential privacy implications unexplored.
Approach: They propose a benchmark to examine whether ALLMs leak user privacy through acoustic voiceprints.
Outcome: The proposed benchmark is constructed from over 22,000 real-world audio clips.
Composable Cross-prompt Essay Scoring by Merging Models (2025.emnlp-main)

Copied to clipboard

Challenge: Existing approaches to cross-prompt automated essay scoring use all available sources . however, using multiple sources for continual adaptation raises privacy concerns .
Approach: They propose a source-free adaptation approach that selectively merges the parameters of individual models without further access to the source datasets.
Outcome: The proposed method outperforms joint-training methods on all sources while maintaining computational efficiency.
X-ray Made Simple: Lay Radiology Report Generation and Robust Evaluation (2026.findings-acl)

Copied to clipboard

Challenge: Technical language and templated nature of professional reports hinder patient comprehension and allow models to artificially boost lexical metrics such as BLEU by reproducing common report patterns.
Approach: They propose a layman's RRG framework that leverages layperson-friendly language to enhance patient accessibility and promote robust evaluation and report generation by encouraging models to focus on semantic accuracy over rigid templates.
Outcome: The proposed framework improves model performance with more layman-style data, compared to templated professional language and inflated lexical scores.
Following the Autoregressive Nature of LLM Embeddings via Compression and Alignment (2025.emnlp-main)

Copied to clipboard

Challenge: Experimental results demonstrate that our method significantly outperforms traditional contrastive learning approaches when using the same amount of data.
Approach: They propose a new contrastive learning method built on embedding conditional probability distributions that integrates two tasks: information compression and conditional distribution alignment.
Outcome: The proposed method outperforms traditional contrastive learning approaches and achieves comparable performance to state-of-the-art models when using the same amount of data.
OS Agents: A Survey on MLLM-based Agents for Computer, Phone and Browser Use (2025.acl-long)

Copied to clipboard

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.
Scaling Law for Multimodal Large Language Model Supervised Fine-Tuning (2026.acl-long)

Copied to clipboard

Challenge: supervised fine-tuning (SFT) is crucial for multimodal large language models, yet a comprehensive scaling law is lacking . et al.: scaling laws focus on model size, pre-training tokens, and MLLM SFT data volumes .
Approach: They propose two scaling laws to guide optimal model-data configuration . they propose one applicable when training data volumes are well defined by researchers .
Outcome: The proposed scaling laws provide valuable recommendations for optimal resource allocation . they show that the proposed laws are more accurate than existing models .

What is GenGO?

GenGO is an NLP powered publication search system. It currenctly indexes 30k+ papers from ACL Anthology, and implements multi-aspect summarization, semantic search, and more!

Information

About
Limitations