My name is Farnood Salehi. I am a Ph.D. student, supervised by Prof. Patrick Thiran and Dr. Elisa Celis. I earned my B.S. in Electrical Engineering from Sharif University of Technology 2014.
- In my research, I have worked on accelerating the stochastic optimization algorithms for the cost functions used commonly in machine learning, where we developed some theory and tested it on real datasets.
- I have also worked on dictionary learning problem. We developed an algorithm to learn the dictionary, which is a basis for representing a signal and tested it on real images for denoising.
- I have worked on a problem related to auctions, where we forced constraints (e.g., fairness, budget, etc) on the repeated auctions (e.g., online advertisement) and derived an algorithm for allocating items to users under such settings. We tested our algorithm on an online ad dataset provided by Yahoo!.
You can find my CV here.
Learning under fairness constraints.
Accelerating stochastic optimization algorithms using multi-armed bandit algorithms for sampling.
|P. Pad, F. Salehi, E. Celis, P. Thiran and M. Unser
International Conference on Machine Learning (ICML), 2017.
|Dictionary Learning Based on Sparse Distribution Tomography|
|E. Celis, F. Salehi and S. Salamatian
Manufacturing & Service Operations Management Society Conference (MSOM), 2017.
|Auctions for Online Advertising with Constraints|
|F. Salehi, P. Thiran, E. Celis
Thirty-second Conference on Neural Information Processing Systems (NeurIPS), 2018.
|Coordinate Descent with Bandit Sampling|
|E. Celis, S. Kapoor, F. Salehi, N. Vishnoi
ACM Conference on Fairness, Accountability, and Transparency (FAT*), 2019.
|Controlling Polarization in Personalization: An Algorithmic Framework|