Jinghan Jia

Room 3210

428 S Shaw LN

East Lansing, MI, USA

I am Jinghan Jia, a third-year Ph.D. student affiliated with the OPTML Group at Michigan State University, guided by Prof. Sijia Liu. My research is committed to advancing the trustworthiness and operational efficiency of AI systems through cutting-edge optimization strategies and machine learning techniques. My areas of focus include:

Machine Unlearning and AI Alignment: My research targets the development of methodologies to align generative AI models, such as Large Language Models (LLMs) and Diffusion Models, with human ethical standards. This involves not only aligning behaviors with human values but also enhancing the capacity for selective data removal from trained models. I am particularly interested in extending machine unlearning processes to conventional computer vision tasks, reinforcing privacy and control over data.
Adversarial Robustness: I am dedicated to bolstering the resilience of generative AI systems against adversarial attacks, commonly referred to as “Jailbreaking”. My work ensures that these systems maintain reliability and performance integrity under diverse and challenging scenarios.
AI for Code: I leverage AI to fortify robustness in programming, software development, and code understanding.
Gradient-Free Optimization: I explore alternative optimization techniques that bypass the reliance on traditional gradient methods, aiming to decrease memory usage and computational costs, thereby facilitating more efficient and scalable AI solutions.

My research strives to produce AI innovations that are not only effective but are also reliable and adhere to ethical guidelines.

news

Sep 25, 2024	Two papers were accepted to the Neurips 2024 and one paper was accepted to the Neurips 2024 DB track!
Sep 19, 2024	A first-author paper was accepted at EMNLP’24 main track.
Jul 2, 2024	A first-author paper was accepted at ECCV’24.
Oct 24, 2023	Grateful to be awarded the NeurIPS 2023 Scholar Award !
Oct 22, 2023	One preprint released To Generate or Not? Safety-Driven Unlearned Diffusion Models Are Still Easy To Generate Unsafe Images… For Now.

Selected publications

See a full publication list at here.

NeurIPS’24
WAGLE: Strategic Weight Attribution for Effective and Modular Unlearning in Large Language Models

Jinghan Jia, Jiancheng Liu, Yihua Zhang, and 3 more authors

In The Thirty-eighth Annual Conference on Neural Information Processing Systems 2024

Abs Paper Code

"The need for effective unlearning mechanisms in large language models (LLMs) is increasingly urgent, driven by the necessity to adhere to data regulations and foster ethical generative AI practices. LLM unlearning is designed to reduce the impact of undesirable data influences and associated model capabilities without diminishing the utility of the model if unrelated to the information being forgotten. Despite growing interest, much of the existing research has focused on varied unlearning method designs to boost effectiveness and efficiency. However, the inherent relationship between model weights and LLM unlearning has not been extensively examined. In this paper, we systematically explore how model weights interact with unlearning processes in LLMs and we design the weight attribution-guided LLM unlearning method, WAGLE, which unveils the interconnections between ’influence’ of weights and ’influence’ of data to forget and retain in LLM generation. By strategically guiding the LLM unlearning across different types of unlearning methods and tasks, WAGLE can erase the undesired content, while maintaining the performance of the original tasks. We refer to the weight attribution-guided LLM unlearning method as WAGLE, which unveils the interconnections between ’influence’ of weights and ’influence’ of data to forget and retain in LLM generation. Our extensive experiments show that WAGLE boosts unlearning performance across a range of LLM unlearning methods such as gradient difference and (negative) preference optimization, applications such as fictitious unlearning (TOFU benchmark), malicious use prevention (WMDP benchmark), and copyrighted information removal, and models including Zephyr-7b-beta and Llama2-7b. To the best of our knowledge, our work offers the first principled method for attributing and pinpointing the influential weights in enhancing LLM unlearning. It stands in contrast to previous methods that lack weight attribution and simpler weight attribution techniques."
@inproceedings{jia2024wagle, title = {WAGLE: Strategic Weight Attribution for Effective and Modular Unlearning in Large Language Models}, author = {Jia, Jinghan and Liu, Jiancheng and Zhang, Yihua and Ram, Parikshit and Baracaldo, Nathalie and Liu, Sijia}, year = {2024}, booktitle = {The Thirty-eighth Annual Conference on Neural Information Processing Systems}, }

EMNLP’24

Soul: Unlocking the power of second-order optimization for llm unlearning

Jinghan Jia, Yihua Zhang, Yimeng Zhang, and 5 more authors

In The 2024 Conference on Empirical Methods in Natural Language Processing 2024

Paper Code

@inproceedings{jia2024soul,
  title = {Soul: Unlocking the power of second-order optimization for llm unlearning},
  author = {Jia, Jinghan and Zhang, Yihua and Zhang, Yimeng and Liu, Jiancheng and Runwal, Bharat and Diffenderfer, James and Kailkhura, Bhavya and Liu, Sijia},
  journal = {arXiv preprint arXiv:2404.18239},
  year = {2024},
  publisher = {EMNLP 2024},
  booktitle = {The 2024 Conference on Empirical Methods in Natural Language Processing},
}

ECCV’24

To Generate or Not? Safety-Driven Unlearned Diffusion Models Are Still Easy To Generate Unsafe Images... For Now

Yimeng Zhang*, Jinghan Jia*, Xin Chen, and 5 more authors

In European Conference on Computer Vision 2024

Paper Code

@inproceedings{zhang2023generate,
  title = {To Generate or Not? Safety-Driven Unlearned Diffusion Models Are Still Easy To Generate Unsafe Images... For Now},
  author = {Zhang*, Yimeng and Jia*, Jinghan and Chen, Xin and Chen, Aochuan and Zhang, Yihua and Liu, Jiancheng and Ding, Ke and Liu, Sijia},
  publisher = {ECCV 2024},
  booktitle = {European Conference on Computer Vision},
  year = {2024}
}

NAACL’24

Leveraging LLMs for dialogue quality measurement

Jinghan Jia, Abi Komma, Timothy Leffel, and 5 more authors

In 2024 Annual Conference of the North American Chapter of the Association for Computational Linguistics 2024

Paper

NeurIPS’23

Model Sparsity Can Simplify Machine Unlearning

Jinghan Jia*, Jiancheng Liu*, Parikshit Ram, and 5 more authors

In Thirty-seventh Conference on Neural Information Processing Systems 2023

Paper Code

@inproceedings{jia2023model,
  title = {Model Sparsity Can Simplify Machine Unlearning},
  author = {Jia*, Jinghan and Liu*, Jiancheng and Ram, Parikshit and Yao, Yuguang and Liu, Gaowen and Liu, Yang and Sharma, Pranay and Liu, Sijia},
  year = {2023},
  publisher = {NeurIPS 2023},
  booktitle = {Thirty-seventh Conference on Neural Information Processing Systems},
}

SANER’23

CLAWSAT: Towards Both Robust and Accurate Code Models

Jinghan Jia*, Shashank Srikant*, Tamara Mitrovska, and 4 more authors

In 2023 IEEE International Conference on Software Analysis, Evolution and Reengineering (SANER) 2023

Paper Code Poster

  doi = {10.48550/ARXIV.2211.11711},
  author = {Jia*, Jinghan and Srikant*, Shashank and Mitrovska, Tamara and Gan, Chuang and Chang, Shiyu and Liu, Sijia and O'Reilly, Una-May},
  keywords = {Machine Learning (cs.LG), Programming Languages (cs.PL), Software Engineering (cs.SE), FOS: Computer and information sciences, FOS: Computer and information sciences},
  booktitle = {2023 IEEE International Conference on Software Analysis, Evolution and Reengineering (SANER)},
  title = {CLAWSAT: Towards Both Robust and Accurate Code Models},
  year = {2023},
  copyright = {Creative Commons Attribution 4.0 International}
}