About Me:
I am a research scientist at Meta AI where I am a part of the AI integrity team. I work on creating scalable and responsible AI systems that can solve challenging integrity problems such as hate speech and low-quality e-commerce detection. Prior to joining Meta AI, I was an AI Research Scientist at Electronic Arts where I was working on applying deep generative models in automating tasks in game design and computer graphics.
I used to be a postdoctoral scholar at USC Data Science and Operations Department, working with Jason Lee. Prior to joining USC, I was a postdoctoral scholar at UCLA Computer Science Department, working with Ameet Talwalkar. Before that, I was an R&D engineer at Starkey Hearing Technologies Research and Development department. I was fortunate to receive my Ph.D. and Masters in Electrical Engineering with a minor in Computer Science from the University of Minnesota under the supervision of Tom Luo.
Research Interests:
My current research interests broadly include the design of scalable & responsible AI for real-world challenging problems. This has meant working on areas such as self-supervised & multi-modal learning, generative modeling, federated learning, and efficient/scalable, fair, robust & explainable AI models.
Current/Past Interns (Co-)Mentored:
Yaodong Yu (UC Berkeley)
Nimit Sohoni (Stanford -> Citadel)
Aaron Chan (USC -> Meta AI)
Khalil Mrini (UC San Diego -> Meta AI)
Krishna Pillutla (University of Washington -> Google Research)
Woojeong Jin (USC)
Ajinkya Tejankar (UC Davis)
Neha Kalibhat (UMD)
Samuel Horvath (KAUST -> MBZUAI)
Jingwen Liang (UCSD -> EA Data & AI)
The items are color-coded based on the following topics*:
Scalable, and Efficient Learning π΅
Robustness, Fairness, and Explainability π’
Private & Federated Learning π΄
Self-Supervised, and Multi-Modal Learning π£
Generative Modeling π‘
*Some items might belong to multiple topics
Accepted/Published:
K. Pillutla, K. Malik, A. Mohamed, M. Rabbat, M. Sanjabi, and L. Xiao, "Federated Learning with Partial Model Personalization," ICML, 2022 π΄π΅
A. Chan, M. Sanjabi, L. Mathias, L. Tan, S. Nie, X. Peng, X. Ren, and H. Firooz, "UniREx: A Unified Learning Framework for Language Model Rationale Extraction," ICML, 2022 π’π΅
B. Joshi, A. Chan, Z. Liu, S. Nie, M. Sanjabi, H. Firooz, and X. Ren βER-TEST: Evaluating Explanation Regularization Methods for NLP Models,β EMNLP, 2022 π’
S. Horvath, M. Sanjabi, L. Xiao, P. Richtarik, and M. Rabbat, "FedShuffle: Recipes for Better Use of Local Work in Federated Learning," TMLR, 2022 π΄
K. Mrini, S. Nie, J. Gu, S. Wang, M. Sanjabi, and H. Firooz, "Detection, Disambiguation, Re-ranking: Autoregressive Entity Linking as a Multi-Task Problem," ACL, 2022 π΅
A. Tejankar, M. Sanjabi, B. Wu, S. Xie, M. Khabsa, H. Pirsiavash, and H. Firooz, "Can we train vision and language zero-shot classification models without syntax?", Self-Supervised Learning: Theory and Practice workshop, NeurIPS, 2022 π£
N. Sohoni, M. Sanjabi, N. Ballas, A. Grover, S. Nie, H. Firooz, and C. Re, "BARACK: Partially Supervised Group Robustness With Guarantees," Spurious correlations, Invariance, and Stability (SCIS) Workshop, ICML, 2022 π’
T. Li, A. Beirami, M. Sanjabi, and V. Smith. "Tilted Empirical Risk Minimization," ICLR, 2021 π’
T. Huang, P. Singhania, M. Sanjabi, P. Mitra, M. Razaviyayn, "Alternating Direction Method of Multipliers for Quantization," AISTAT, 2021 π΅
W. Jin, M. Sanjabi, S. Nie, L. Tan, X. Ren, H. Firooz, "MSD: Saliency-aware Knowledge Distillation for Multimodal Understanding," EMNLP, 2021 π΅π£
T. Li, M. Sanjabi, A. Beirami and V. Smith. "Fair Resource Allocation in Federated Learning," ICLR, 2020 π΄π’
T. Li, A. K. Sahu, M. Zaheer, M. Sanjabi, A. Talwalkar, and V. Smith, "On the convergence of federated optimization in heterogeneous networks," MLSYS, 2020 π΄
J. Liang, H. Liu, Y. Zhao, M. Sanjabi, "Building Placements In Urban Modeling Using Conditional Generative Latent Optimization," IEEE International Conference on Image Processing (ICIP), 2020 π‘
M. Razaviyayn, T. Huang, S. Lu, M. Nouiehed, M. Sanjabi, M. Hong, "Nonconvex min-max optimization: Applications, challenges, and recent theoretical advances," IEEE Signal Processing Magazine, 2020 π΅
M. Nouiehed, M. Sanjabi, T. Huang, J. D. Lee, and M. Razaviyayn, "Solving a Class of Non-Convex Min-Max Games Using Iterative First-Order Methods," NeurIPS, 2019 π΅π’
Y. Zhao, H. Liu, I. Borovikov, A. Beirami, M. Sanjabi, and K. Zaman, "Multi-Style Generative Adversarial Terrain Amplification," ACM SIGGRAPH Conference on Computer Graphics and Interactive Techniques in Asia (SIGGRAPH Asia) [Published in ACM Transactions on Graphics (TOG)], 2019 π‘
B. Barazandeh, M. Razaviyayn, and M. Sanjabi, "Training Generative Networks using Random Discriminators," IEEE Data Science Workshop (DSW), 2019 [Best Paper Award] π΅π‘
M. Sanjabi, J. Ba, M. Razaviyayn, and J. D. Lee, "On the Convergence and Robustness of Training GANs with Regularized Optimal Transport," NeurIPS, 2018 π΅π‘
V. Smith, C. K. Chiang, M. Sanjabi, and A. Talwalkar, "Federated Multi-Task Learning," NeurIPS, 2017 π΄
M. Razaviyayn, M. Sanjabi, and Z.-Q. Luo, "A Stochastic Successive Minimization Method for Nonsmooth Nonconvex Optimization with Applications to Transceiver Design in Wireless Communication Networks," Mathematical Programming, 2016 π΅
Preprints:
C. Guo, A. Sablayrolles, M. Sanjabi, "Analyzing Privacy Leakage in Machine Learning via Multiple Hypothesis Testing: A Lesson From Fano," arXiv preprint:2210.13662, 2022 [submitted to AISTATS'23] π΄
N. M. Kalibhat, K. Narang, L. Tan, H. Firooz, M. Sanjabi, and S. Feizi, "Understanding Failure Modes of Self-Supervised Learning," arXiv preprint arXiv:2203.01881, 2022 [submitted to CVPR'23] π’π£
T. Li, A. Beirami, M. Sanjabi, and V. Smith, "On tilted losses in machine learning: Theory and applications," arXiv preprint arXiv:2112.08802 [submitted to JMLR] π’
M. Sanjabi, S. Baharlouie, M. Razaviyayn, and J. D. Lee, "When Does Non-Orthogonal Tensor Decomposition Have No Spurious Local Minima?" arXiv preprint arXiv:1911.09815 π΅
Patents:
Y. Zhao, I. Borovikov, M. Sanjabi, M. Sardari, H. Chaput, N. Aghdaie, and K. Zaman, "Systems and methods for supervised and unsupervised animation style transfer," US Patent App. 16/896,541 π‘
Recent Talks
Fair Resource Allocation in Federated Learning, INFORMS 2020
Challenges of Modern Machine Learning at Scale: An Optimization Perspective, Facebook AI, Amazon Alexa AI, Bosch AI, UIC, Electronic Arts Data & AI, Samsung Research, JP Morgan AI (2019-2020)
Generative Adversarial Network Formulations with Convergence Guarantees, UCL, and INFORMS 2018
Contact:
Email: my first name dot my last name "at" gmail "dot" com