I am currently a Master’s student in Computing Science at Simon Fraser University, supervised by Professor Mahdavi-Amiri. My research interests lie in Machine Learning and Computer Vision, particularly at its intersection with Natural Language Processing.
I earned my B.Sc. degree in Computer Engineering from Sharif University of Technology in 2024. During my time there, I worked as a research assistant in both the Machine Learning Lab and Data Analytics Lab. Moreover, I spent two years at Tapsi (the Iranian version of Uber), working as a Software Engineer for one year and as a Data Scientist for another.
Before my undergraduate studies, I was awarded a silver medal at the 13thInternational Olympiad on Astronomy and Astrophysics (IOAA) in Hungary in 2019, after earning a gold medal and ranking first in the national Olympiad.
We propose a novel framework for incorporating unlabeled data into semi-supervised classification problems, where scenarios involving the minimization of either i) adversarially robust or ii) non-robust loss functions have been considered. Notably, we allow the unlabeled samples to deviate slightly (in total variation sense) from the in-domain distribution. The core idea behind our framework is to combine Distributionally Robust Optimization (DRO) with self-supervised training. As a result, we also leverage efficient polynomial-time algorithms for the training stage. From a theoretical standpoint, we apply our framework on the classification problem of a mixture of two Gaussians in ℝd, where in addition to the m independent and labeled samples from the true distribution, a set of n (usually with n≫m) out of domain and unlabeled samples are given as well. Using only the labeled data, it is known that the generalization error can be bounded by ∝(d/m)1/2. However, using our method on both isotropic and non-isotropic Gaussian mixture models, one can derive a new set of analytically explicit and non-asymptotic bounds which show substantial improvement on the generalization error compared to ERM. Our results underscore two significant insights: 1) out-of-domain samples, even when unlabeled, can be harnessed to narrow the generalization gap, provided that the true data distribution adheres to a form of the “cluster assumption", and 2) the semi-supervised learning paradigm can be regarded as a special case of our framework when there are no distributional shifts. We validate our claims through experiments conducted on a variety of synthetic and real-world datasets.
@inproceedings{saberi2024outofdomain,title={Out-Of-Domain Unlabeled Data Improves Generalization},author={Saberi, Amir Hossein and Najafi, Amir and Heidari, Alireza and Movasaghinia, Mohammad Hosein and Motahari, Abolfazl and Khalaj, Babak},booktitle={The Twelfth International Conference on Learning Representations},year={2024},url={https://openreview.net/forum?id=Bo6GpQ3B9a},publisher={International Conference on Learning Representations}}
