Papers by Jinghan Zhang
SOUL: Unlocking the Power of Second-Order Optimization for LLM Unlearning (2024.emnlp-main)
Copied to clipboard
Jinghan Jia, Yihua Zhang, Yimeng Zhang, Jiancheng Liu, Bharat Runwal, James Diffenderfer, Bhavya Kailkhura, Sijia Liu
| Challenge: | Large Language Models (LLMs) have highlighted the need for effective unlearning mechanisms to comply with data regulations and ethical AI practices. |
| Approach: | They propose a second-order optimization-based LLM unlearning framework which extends the static, one-shot model update using influence unlearning to a dynamic, iterative unlearning process. |
| Outcome: | The proposed framework outperforms first-order methods across unlearning tasks, models, and metrics. |
VALU: A Benchmark for Video Anomaly Temporal Localization and Understanding at Multiple Semantic Levels (2026.acl-long)
Copied to clipboard
Yixiao He, Menghao Zhang, Haifeng Sun, Jing Wang, Kangheng Lin, Jinghan Wang, Chenye Xu, Pengfei Ren, Qi Qi, Jingyu Wang
| Challenge: | Recent advances in Video Large Language Models (Video-LLMs) enhance the ability of VAU models to describe and interpret anomalies. |
| Approach: | They propose a benchmark that explicitly defines anomalies across five semantic levels and provides detailed temporal boundaries and detailed textual descriptions for each. |
| Outcome: | The proposed benchmark defines anomalies across five semantic levels and provides detailed descriptions for each. |
DEEPER Insight into Your User: Directed Persona Refinement for Dynamic Persona Modeling (2025.acl-long)
Copied to clipboard
Aili Chen, Chengyu Du, Jiangjie Chen, Jinghan Xu, Yikai Zhang, Siyu Yuan, Zulong Chen, Liangyue Li, Yanghua Xiao
| Challenge: | Existing methods for generating personas from static historical data fail to capture dynamic behaviors and evolving preferences in real-world interactive scenarios. |
| Approach: | They propose a novel approach that iteratively updates personas using streaming user behavior data to continually enhance their quality. |
| Outcome: | The proposed approach delivers 32.2% reduction in user behavior prediction error over four update rounds, outperforming the best baseline by 22.92%. |
Blind Spot Navigation in Large Language Model Reasoning with Thought Space Explorer (2026.findings-eacl)
Copied to clipboard
Jinghan Zhang, Fengran Mo, Tharindu Cyril Weerasooriya, Xinyue Ye, Dongjie Wang, Yanjie Fu, Kunpeng Liu
| Challenge: | Existing studies show that large language models have strong reasoning capabilities through chain-structured methods. |
| Approach: | They propose a framework for navigating and expanding thought structures to overcome blind spots in LLM reasoning. |
| Outcome: | The proposed framework overcomes blind spots in large language models by expanding thought structures . the proposed framework improves accuracy of the final answer and intermediate reasoning steps . |
Diversity-oriented Data Augmentation with Large Language Models (2025.acl-long)
Copied to clipboard
| Challenge: | Existing data augmentation methods focus on increasing sample numbers while neglecting sample distribution diversity, which can lead to model overfitting. |
| Approach: | They propose a data augmentation framework that focuses on sample distribution diversity and trains a large language model as a diverse paraphraser. |
| Outcome: | The proposed framework achieves an average performance gain of 10.52% surpassing the runner-up baseline with more than three percentage points. |
SEUF: Is Unlearning One Expert Enough for Mixture-of-Experts LLMs? (2025.acl-long)
Copied to clipboard
| Challenge: | Recent advances in LLMs unlearning have shown remarkable success in removing unwanted data-model influences while preserving the model’s utility for legitimate knowledge. |
| Approach: | They propose a Selected-Expert Unlearning Framework (SEUF) that combines expert attribution and an anchor loss to ensure controlled unlearning. |
| Outcome: | Experiments show that the proposed framework improves forget quality and model utility by 35% on MoE LLMs across benchmarks and LLM architectures compared to standard unlearning algorithms . |
Reasoning Model Unlearning: Forgetting Traces, Not Just Answers, While Preserving Reasoning Skills (2025.emnlp-main)
Copied to clipboard
Changsheng Wang, Chongyu Fan, Yihua Zhang, Jinghan Jia, Dennis Wei, Parikshit Ram, Nathalie Baracaldo, Sijia Liu
| Challenge: | Existing methods for LRM unlearning overlook critical information leakage in reasoning traces, even when final answers are successfully removed. |
| Approach: | They propose a method that suppresses reasoning traces while preserving the model's general reasoning ability. |
| Outcome: | The proposed method significantly reduces reasoning trace leakage and achieves strong performance across reasoning and safety benchmarks, including WMDP, StrongReject, JBB-Behaviors and WildJailbreak. |
Prototypical Reward Network for Data-Efficient Model Alignment (2024.acl-long)
Copied to clipboard
| Challenge: | Reinforcement Learning from Human Feedback (RLHF) is a reward model that fine-tunes Large Language Models (LLMs) by utilizing Prototypical Networks. |
| Approach: | They propose a framework utilizing Prototypical Networks to enhance reward models under limited human feedback, enabling more stable and reliable structural learning from fewer samples. |
| Outcome: | The proposed framework improves reward models under limited human feedback, surpassing traditional methods, especially in data-limited scenarios. |
Entropy-based Exploration Conduction for Multi-step Reasoning (2025.findings-acl)
Copied to clipboard
| Challenge: | Existing methods to automatically decide the depth of exploration of the reasoning procedure lead to high cost and a lack of flexibility. |
| Approach: | They propose a method that dynamically adjusts the exploration depth during multi-step reasoning by monitoring LLM’s output entropy and variance entropic. |
| Outcome: | The proposed method captures the uncertainty of the current step and the fluctuation of uncertainty across consecutive reasoning steps and then selects whether to deepen, expand, or stop exploration according to the probability. |
How Can Synthetic Data Improve Multilingual Language Model Pretraining? A Data Quality Perspective (2026.acl-long)
Copied to clipboard
| Challenge: | Low-resource languages are a long-tail problem for multilingual LLMs due to limited high-quality training data. |
| Approach: | They propose a method that translates high-quality, knowledge-rich English data into low-resource languages . they propose SynRank, which leverages synthetic data as positive samples to train a classifier . |
| Outcome: | The proposed method matches handcrafted rule-based filtering by human experts and significantly improves knowledge-intensive tasks with less data. |
A Survey on Multi-modal Intent Recognition: Recent Advances and New Frontiers (2025.findings-emnlp)
Copied to clipboard
| Challenge: | Multi-modal intent recognition (MIR) requires integrating non-verbal cues from real-world contexts to enhance human intention understanding. |
| Approach: | They present a comprehensive review of multi-modal intent recognition . they provide a survey of the field covering textual, visual, and acoustic signals . |
| Outcome: | The present survey summarises the current state of multi-modal intent recognition . it includes a comprehensive taxonomy and advanced methods . |