Saeed Vahidian

I am a postdoctoral researcher at Duke University, working with Prof. Yiran Chen. Previously, I got my Ph.D. from the Department of Electrical and Computer Engineering at the University of California San Diego (UCSD), where I was supervised by Prof. Bill Lin.

We've hit "peak data." The web is tapped out. Our solution? Synthetic Data—programmatically creating targeted, privacy-preserving data to build the next generation of AI.
My research explores Multimodal Synthetic Data Generation (Vision ↔ Language ↔ Audio) with a focus on advancing how foundation models learn across diverse modalities. I design methods to create high-fidelity, controllable synthetic datasets that enable robust training pipelines for next-generation multimodal AI systems. A core aspect of my work also addresses robust learning with synthetic data. My work is built upon three core pillars:

Controllability: Fine-grained control over generated Synthetic Data for Vision–Language, Multimodal, and Video

Explainability: how and why a particular piece of synthetic data was created

Robust Learning on Synthetic Data: Designing robust optimization methods to enhance the generalization and robustness of models trained on synthetic data

Edge Intelligence: Enables privacy-aware synthetic data generation and federated training under computational and privacy budgets.

@misc{wang2024coreinferacceleratinglargelanguage, title={CoreInfer: Accelerating Large Language Model Inference with Semantics-Inspired Adaptive Sparse Activation}, author={Qinsi Wang and Saeed Vahidian and Hancheng Ye and Jianyang Gu and Jianyi Zhang and Yiran Chen}, year={2024}, eprint={2410.18311}, archivePrefix={arXiv}, primaryClass={cs.LG}, url={https://arxiv.org/abs/2410.18311}}

@Article{SVahidian-RobustDD, author = {Saeed Vahidian and Mingyu Wang and Jianyang Gu and Vyacheslav Kungurtsev and Wei Jiang and Yiran Chen}, title = {Group Distributionally Robust Dataset Distillation with Risk Minimization}, journal = {CoRR}, volume = {abs/2402.04676}, year = {2024}, url = {https://doi.org/10.48550/arXiv.2402.04676}, doi = {10.48550/ARXIV.2402.04676}, eprinttype = {arXiv}}

@Article{SVahidian-ICEPL-2024, title={Dataset Distillation from First Principles: Integrating Core Information Extraction and Purposeful Learning}, author={Vyacheslav Kungurtsev and Yuanfang Peng and Jianyang Gu and Saeed Vahidian and Anthony Quinn and Fadwa Idlahcen and Yiran Chen}, year={2024}, eprint={2409.01410}, archivePrefix={arXiv}, url={https://arxiv.org/abs/2409.01410}}

@Article{SVahidian-minmax-Diffusion2024, title={Efficient dataset distillation via minimax diffusion}, author={Gu, Jianyang and Vahidian, Saeed and Kungurtsev, Vyacheslav and Wang, Haonan and Jiang, Wei and You, Yang and Chen, Yiran}, booktitle={Proceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition}, pages={15793--15803}, year={2024} }

@Article{SVahidian-BiasedDD2024, author = {Yao Lu and Jianyang Gu and Xuguang Chen and Saeed Vahidian and Qi Xuan}, title = {Exploring the Impact of Dataset Bias on Dataset Distillation}, journal = {CoRR}, volume = {abs/2403.16028}, year = {2024}, url = {https://doi.org/10.48550/arXiv.2403.16028}, doi = {10.48550/ARXIV.2403.16028}, eprinttype = {arXiv}, eprint = {2403.16028}, timestamp = {Tue, 09 Apr 2024 15:12:39 +0200}, biburl = {https://dblp.org/rec/journals/corr/abs-2403-16028.bib}, bibsource = {dblp computer science bibliography, https://dblp.org} }

@Article{SVahidian-FedGPT2024, title={Towards Building the Federated GPT: Federated Instruction Tuning}, author={Jianyi Zhang and Saeed Vahidian and Martin Kuo and Chunyuan Li and Ruiyi Zhang and Guoyin Wang and Yiran Chen}, year={2023}, eprint={2305.05644}, archivePrefix={arXiv}, primaryClass={cs.CL}}

@Article{SVahidian-DDF-ECCV2024, title={Unlocking the Potential of Federated Learning: The Symphony of Dataset Distillation via Deep Generative Latents}, author={Yuqi Jia and Saeed Vahidian and Jingwei Sun and Jianyi Zhang and Vyacheslav Kungurtsev and Neil Zhenqiang Gong and Yiran Chen}, year={2023}, eprint={2312.01537}, archivePrefix={arXiv}, primaryClass={cs.LG}, url={https://arxiv.org/abs/2312.01537}}

@Article{SVahidian-DDF-ICCV2023, author = {Saeed Vahidian and Sreevatsank Kadaveru and Woonjoon Baek and Weijia Wang and Vyacheslav Kungurtsev and Chen Chen and Mubarak Shah and Bill Lin}, title = {When Do Curricula Work in Federated Learning?}, booktitle = {{IEEE/CVF} International Conference on Computer Vision, {ICCV} 2023, Paris, France, October 1-6, 2023}, pages = {5061--5071}, publisher = {{IEEE}}, year = {2023}}

@Article{SVahidian-DDF-AAAI2023, author = {Saeed Vahidian and Mahdi Morafah and Weijia Wang and Vyacheslav Kungurtsev and Chen Chen and Mubarak Shah and Bill Lin}, editor = {Brian Williams and Yiling Chen and Jennifer Neville}, title = {Efficient Distribution Similarity Identification in Clustered Federated Learning via Principal Angles between Client Data Subspaces}, booktitle = {Thirty-Seventh {AAAI} Conference on Artificial Intelligence, {AAAI} 2023, Thirty-Fifth Conference on Innovative Applications of Artificial Intelligence, {IAAI} 2023, Thirteenth Symposium on Educational Advances in Artificial Intelligence, {EAAI} 2023, Washington, DC, USA, February 7-14, 2023}, pages = {10043--10052}, publisher = {{AAAI} Press}, year = {2023}}

@Article{SVahidian-DDF-OJCS2023, author={Morafah, Mahdi and Vahidian, Saeed and Wang, Weijia and Lin, Bill}, journal={IEEE Open Journal of the Computer Society}, title={FLIS: Clustered Federated Learning Via Inference Similarity for Non-IID Data Distribution}, year={2023}, volume={4}, number={}, pages={109-120}}

@Article{SVahidian-DDF-IROS2023, author = {Umar Khalid and Hasan Iqbal and Saeed Vahidian and Jing Hua and Chen Chen}, title = {{CEFHRI:} {A} Communication Efficient Federated Learning Framework for Recognizing Industrial Human-Robot Interaction}, booktitle = {{IROS}}, pages = {10141--10148}, year = {2023}, url = {https://doi.org/10.1109/IROS55552.2023.10341467}}

@Article{SVahidian-ICE-UAI2023, author = {Saeed Vahidian and Baharan Mirzasoleiman and Alexander Cloninger}, title = {Coresets for Estimating Means and Mean Square Error with Limited Greedy Samples}, booktitle = {Proceedings of the Thirty-Sixth Conference on Uncertainty in Artificial Intelligence, {UAI} 2020, virtual online, August 3-6, 2020}, series = {Proceedings of Machine Learning Research}, volume = {124}, pages = {350--359}, publisher = {{AUAI} Press}, year = {2020}}

@Article{SVahidian-ICE-CVPR2020, author = {Mohsen Joneidi and Saeed Vahidian and Ashkan Esmaeili and Weijia Wang and Nazanin Rahnavard and Bill Lin and Mubarak Shah}, title = {Select to Better Learn: Fast and Accurate Deep Learning Using Data Selection From Nonlinear Manifolds}, booktitle = {2020 {IEEE/CVF} Conference on Computer Vision and Pattern Recognition, {CVPR} 2020, Seattle, WA, USA, June 13-19, 2020}, pages = {7816--7826}, publisher = {Computer Vision Foundation / {IEEE}}, year = {2020}}

@Article{SVahidian-Meta-ICLR2020, @article{DBLP:journals/corr/abs-2006-10236, author = {Siavash Khodadadeh and Sharare Zehtabian and Saeed Vahidian and Weijia Wang and Bill Lin and Ladislau B{\"{o}}l{\"{o}}ni}, title = {Unsupervised Meta-Learning through Latent-Space Interpolation in Generative Models}, journal = {CoRR}, volume = {abs/2006.10236}, year = {2020}, url = {https://arxiv.org/abs/2006.10236}}