Mahdi Morafah
Machine Learning Researcher
Generative AI and Federated Learning Enthusiast
Former ML Research Intern at Tesla Autopilot and Qualcomm AI
![linkedin_photo.jpg](/homepage/assets/img/linkedin_photo.jpg)
Department of Electrical and Computer Engineering
University of California San Diego
9500 Gilman Dr, La Jolla, CA 92093
Email: mmorafah 'at' ucsd 'dot' edu
I am a PhD Candidate (third year) in Electrical and Computer Engineering at University of California San Diego (UCSD), luckly being advised by Prof. Bill Lin.
I am broadly interested in Machine Learning and Optimization. My current research focus is on Federated Learning, Generative AI and Business. Here are the research themes that I am currently working on:
- Large Scale Federated Learning Over Diverse Heterogeneous Devices
- Efficient Training and Fine-tuning of Large Language Models in Federated Learning
- Fast Sampling for Diffusion Models
- Multi-Modal Foundation Models
- Effect of Generative AI in Business Market
I previously received a M.S. in Electrical and Computer Engineering at UCSD in 2021, and a B.S. in Electrical Engineering at Tehran Polytechnique.
I also working as a machine learning research intern at Tesla Autopilot and Qualcomm AI.
I am looking for machine learning internship positions for summer 2024.
I am open to collaborate on interesting project, please feel free to reach out.
news
Nov 20, 2023 | I am excited to successfully pass my PhD qualifying exam and being advanced to candidacy! |
---|---|
Jul 20, 2023 | New accepted paper “A Practical Recipe for Federated Learning Under Statistical Heterogeneity Experimental Design” [Accepted to IEEE Transactions on AI (Jul, 2023)]. |
May 31, 2023 | Served as Reviewer for IEEE TCNS and IEEE CDC. |
Jan 5, 2023 | Awarded AAAI 2023 student scholarship for conference attendance. |
Sep 30, 2022 | New paper on arXiv “Rethinking Data Heterogeneity in Federated Learning: Introducing a New Notion and Standard Benchmarks” [Accepted to FL NeurIPS’22 Workshops (Oct, 2022)]. |
Sep 21, 2022 | New paper on arXiv “Efficient Distribution Similarity Identification in Clustered Federated Learning via Principal Angles Between Client Data Subspaces” [Accepted to AAAI’23 (acceptance rate=19.6%) Nov, 2022]. |
selected publications
- IEEE OJCSarXiv preprint arXiv:2208.09754 2022
- IEEE TCNSarXiv preprint arXiv:2110.10406 2021
- IEEE TAIIEEE Transactions on Artificial Intelligence 2023