Get our free extension to see links to code for papers anywhere online!

Chrome logo Add to Chrome

Firefox logo Add to Firefox

Picture for Mahdi Soltanolkotabi

Understanding Overparameterization in Generative Adversarial Networks


Apr 12, 2021
Yogesh Balaji, Mohammadmahdi Sajedi, Neha Mukund Kalibhat, Mucong Ding, Dominik Stöger, Mahdi Soltanolkotabi, Soheil Feizi

* Accepted in ICLR 2021 

  Access Paper or Ask Questions

PipeTransformer: Automated Elastic Pipelining for Distributed Training of Transformers


Feb 12, 2021
Chaoyang He, Shen Li, Mahdi Soltanolkotabi, Salman Avestimehr


  Access Paper or Ask Questions

Theoretical Insights Into Multiclass Classification: A High-dimensional Asymptotic View


Nov 16, 2020
Christos Thrampoulidis, Samet Oymak, Mahdi Soltanolkotabi

* To Appear at NeurIPS 2020. 62 pages, 7 figures 

  Access Paper or Ask Questions

Precise Statistical Analysis of Classification Accuracies for Adversarial Training


Oct 21, 2020
Adel Javanmard, Mahdi Soltanolkotabi


  Access Paper or Ask Questions

Minimax Lower Bounds for Transfer Learning with Linear and One-hidden Layer Neural Networks


Jun 16, 2020
Seyed Mohammadreza Mousavi Kalan, Zalan Fabian, A. Salman Avestimehr, Mahdi Soltanolkotabi


  Access Paper or Ask Questions

Approximation Schemes for ReLU Regression


May 26, 2020
Ilias Diakonikolas, Surbhi Goel, Sushrut Karmalkar, Adam R. Klivans, Mahdi Soltanolkotabi


  Access Paper or Ask Questions

Compressive sensing with un-trained neural networks: Gradient descent finds the smoothest approximation


May 07, 2020
Reinhard Heckel, Mahdi Soltanolkotabi

* arXiv admin note: text overlap with arXiv:1910.14634 

  Access Paper or Ask Questions

High-Dimensional Robust Mean Estimation via Gradient Descent


May 04, 2020
Yu Cheng, Ilias Diakonikolas, Rong Ge, Mahdi Soltanolkotabi

* Under submission to ICML'20 

  Access Paper or Ask Questions

Precise Tradeoffs in Adversarial Training for Linear Regression


Feb 24, 2020
Adel Javanmard, Mahdi Soltanolkotabi, Hamed Hassani


  Access Paper or Ask Questions

Convergence and sample complexity of gradient methods for the model-free linear quadratic regulator problem


Dec 26, 2019
Hesameddin Mohammadi, Armin Zare, Mahdi Soltanolkotabi, Mihailo R. Jovanović

* 38 pages, 4 figures 

  Access Paper or Ask Questions

Denoising and Regularization via Exploiting the Structural Bias of Convolutional Generators


Oct 31, 2019
Reinhard Heckel, Mahdi Soltanolkotabi


  Access Paper or Ask Questions

Generalization Guarantees for Neural Networks via Harnessing the Low-rank Structure of the Jacobian


Jul 04, 2019
Samet Oymak, Zalan Fabian, Mingchen Li, Mahdi Soltanolkotabi


  Access Paper or Ask Questions

Gradient Descent with Early Stopping is Provably Robust to Label Noise for Overparameterized Neural Networks


Apr 07, 2019
Mingchen Li, Mahdi Soltanolkotabi, Samet Oymak


  Access Paper or Ask Questions

Towards moderate overparameterization: global convergence guarantees for training shallow neural networks


Feb 12, 2019
Samet Oymak, Mahdi Soltanolkotabi


  Access Paper or Ask Questions

Fitting ReLUs via SGD and Quantized SGD


Jan 19, 2019
Seyed Mohammadreza Mousavi Kalan, Mahdi Soltanolkotabi, A. Salman Avestimehr


  Access Paper or Ask Questions

Overparameterized Nonlinear Learning: Gradient Descent Takes the Shortest Path?


Dec 25, 2018
Samet Oymak, Mahdi Soltanolkotabi


  Access Paper or Ask Questions

Polynomially Coded Regression: Optimal Straggler Mitigation via Data Encoding


May 24, 2018
Songze Li, Seyed Mohammadreza Mousavi Kalan, Qian Yu, Mahdi Soltanolkotabi, A. Salman Avestimehr


  Access Paper or Ask Questions

End-to-end Learning of a Convolutional Neural Network via Deep Tensor Decomposition


May 16, 2018
Samet Oymak, Mahdi Soltanolkotabi

* 29 pages, 12 figures 

  Access Paper or Ask Questions

Fundamental Resource Trade-offs for Encoded Distributed Optimization


Mar 31, 2018
A. Salman Avestimehr, Seyed Mohammadreza Mousavi Kalan, Mahdi Soltanolkotabi


  Access Paper or Ask Questions

Gradient Methods for Submodular Maximization


Aug 15, 2017
Hamed Hassani, Mahdi Soltanolkotabi, Amin Karbasi


  Access Paper or Ask Questions

Learning ReLUs via Gradient Descent


Aug 01, 2017
Mahdi Soltanolkotabi

* arXiv admin note: text overlap with arXiv:1702.06175 

  Access Paper or Ask Questions

Theoretical insights into the optimization landscape of over-parameterized shallow neural networks


Jul 16, 2017
Mahdi Soltanolkotabi, Adel Javanmard, Jason D. Lee


  Access Paper or Ask Questions

Structured signal recovery from quadratic measurements: Breaking sample complexity barriers via nonconvex optimization


Feb 20, 2017
Mahdi Soltanolkotabi


  Access Paper or Ask Questions

Fast and Reliable Parameter Estimation from Nonlinear Observations


Oct 23, 2016
Samet Oymak, Mahdi Soltanolkotabi

* 23 pages, 4 figures 

  Access Paper or Ask Questions

Approximate Subspace-Sparse Recovery with Corrupted Data via Constrained $\ell_1$-Minimization


Feb 13, 2016
Ehsan Elhamifar, Mahdi Soltanolkotabi, Shankar Sastry


  Access Paper or Ask Questions

Sharp Time--Data Tradeoffs for Linear Inverse Problems


Jan 05, 2016
Samet Oymak, Benjamin Recht, Mahdi Soltanolkotabi


  Access Paper or Ask Questions

Experimental robustness of Fourier Ptychography phase retrieval algorithms


Dec 18, 2015
Li-Hao Yeh, Jonathan Dong, Jingshan Zhong, Lei Tian, Michael Chen, Gongguo Tang, Mahdi Soltanolkotabi, Laura Waller

* Opt. Express 23, 33214-33240 (2015) 

  Access Paper or Ask Questions

Isometric sketching of any set via the Restricted Isometry Property


Oct 06, 2015
Samet Oymak, Benjamin Recht, Mahdi Soltanolkotabi

* 17 pages 

  Access Paper or Ask Questions