Papers by Ronghao Zhang
ALRPHFS: Adversarially Learned Risk Patterns with Hierarchical Fast & Slow Reasoning for Robust Agent Defense (2025.findings-emnlp)
Copied to clipboard
| Challenge: | Existing safety checks fail to capture complex semantic risks posed by harmful user inputs or unsafe agent behaviors. |
| Approach: | They propose a framework to bridge the semantic gap between safety checks and real-world risks. |
| Outcome: | The proposed framework achieves superior overall performance compared to existing baselines. |
LiveCANNBench: Benchmark SWE AI Coding for Ascend CANN (2026.findings-acl)
Copied to clipboard
Sijie Wang, Kai Zhao, Wee Peng Tay, Shuo Zhang, Chengwen Liu, Quanjiang Guo, Ren Junhao, Xin Li, Heng Lian, Jingdi Lei, Rui She, Huacan Wang, Ronghao Chen
| Challenge: | Recent advances in agents have enabled multi-file, multi-language, and dependency-aware AI coding. |
| Approach: | They propose an SWE-level benchmark for AI coding in the Huawei Ascend CANN software stack. |
| Outcome: | The proposed benchmark is constructed from real-world CANN repositories and consists of over 400 task instances spanning multiple file, multi-language, and execution-aware coding challenges. |
Chain of Ideas: Revolutionizing Research Via Novel Idea Development with LLM Agents (2025.findings-emnlp)
Copied to clipboard
Long Li, Weiwen Xu, Jiayan Guo, Ruochen Zhao, Xingxuan Li, Yuqian Yuan, Boqiang Zhang, Yuming Jiang, Yifei Xin, Ronghao Dang, Yu Rong, Deli Zhao, Tian Feng, Lidong Bing
| Challenge: | Existing methods for idea generation either trivially prompt LLMs or expose LLM to extensive literature without indicating useful information. |
| Approach: | They propose a chain-of-ideas agent that organizes literature in a chains structure . they propose evaluating idea-generation methods from different perspectives . |
| Outcome: | The proposed agent outperforms existing methods and matches human quality in idea generation. |
Lost in Literalism: How Supervised Training Shapes Translationese in LLMs (2025.acl-long)
Copied to clipboard
| Challenge: | Large language models exhibit translationese errors and generate unexpected unnatural translations . Neural machine translation (NMT) has become the dominant method in machine translation research . |
| Approach: | They evaluate the prevalence of translationese in LLM-generated translations and investigate its roots during supervised fine-tuning. |
| Outcome: | The proposed methods reduce translationese while improving translation naturalness . the proposed methods are validated by human evaluations and automatic metrics . |
Beyond Surface-Level Patterns: An Essence-Driven Defense Framework Against Jailbreak Attacks in LLMs (2025.findings-acl)
Copied to clipboard
| Challenge: | Existing methods focus on surface-level patterns, overlooking the deeper attack essences. |
| Approach: | They propose an Essence-Driven Defense Framework Against Jailbreak Attacks in Aligned Large Language Models that extracts the "attack essence" from a diverse set of known attack instances and stores it in an offline vector database. |
| Outcome: | The proposed framework outperforms existing methods by reducing the Attack Success Rate by at least 20%, underscoring its superior robustness against jailbreak attacks. |
CloneMem: Benchmarking Long-Term Memory for AI Clones (2026.acl-long)
Copied to clipboard
| Challenge: | Existing memory benchmarks rely on user–agent conversational histories, which are temporally fragmented and insufficient for capturing continuous life trajectories. |
| Approach: | They propose a benchmark for evaluating long-term memory in AI Clone scenarios grounded in non-conversational digital traces, including diaries, social media posts, and emails, spanning one to three years. |
| Outcome: | Experiments show that existing memory benchmarks struggle in this setting, highlighting open challenges for life-grounded personalized AI. |
Tiny Scales, Great Challenges: The Limits of Multimodal LLMs in Scale Recognition (2026.acl-long)
Copied to clipboard
| Challenge: | Existing benchmarks focus on a single type of quantity or a specific format, lacking a comprehensive evaluation of scale recognition capabilities. |
| Approach: | They propose a visual scale recognition benchmark built using images from COCO, Open Images, and Flickr to evaluate scale recognition capabilities of multimodal large language models. |
| Outcome: | The proposed model achieves 42.60% accuracy, lower than the 97.40% of humans. |
RealMem: Benchmarking LLMs in Real-World Memory-Driven Interaction (2026.findings-acl)
Copied to clipboard
Haonan Bian, Zhiyuan Yao, Sen Hu, Zishan Xu, Shaolei Zhang, Yifu Guo, Ziliang Yang, Xueran Han, Huacan Wang, Ronghao Chen
| Challenge: | Existing benchmarks focus on casual conversation or task-oriented dialogue, failing to capture “long-term project-oriented” interactions where agents must track evolving goals. |
| Approach: | They propose a benchmark that simulates the dynamic evolution of memory in real-world projects. |
| Outcome: | The proposed benchmarks simulate the dynamic evolution of memory in real-world projects. |
KnowMe-Bench: Benchmarking Person Understanding for Lifelong Digital Companions (2026.acl-long)
Copied to clipboard
Tingyu Wu, Zhisheng Chen, Ziyan Weng, Shuhe Wang, Shuo Zhang, Sen Hu, Silin Wu, Qizhen Lan, Huacan Wang, Ronghao Chen
| Challenge: | Existing long-horizon memory benchmarks use multi-turn dialogues or synthetic user histories . despite rapid progress on long-term memory evaluation, there are gaps in existing benchmarks . |
| Approach: | They propose a long-form autobiographical narrative benchmark that reconstructs each narrative into a flashback-aware, time-anchored stream and evaluates models with evidence-linked questions. |
| Outcome: | The proposed benchmarks build from long-form autobiographical narratives . they show that retrieval-augmented systems improve factual accuracy while errors persist on temporally grounded explanations and higher-level inferences. |