Timezone: »

Robustness to Adversarial Perturbations in Learning from Incomplete Data
Amir Najafi · Shin-ichi Maeda · Masanori Koyama · Takeru Miyato

Tue Dec 10 05:30 PM -- 07:30 PM (PST) @ East Exhibition Hall B + C #225

What is the role of unlabeled data in an inference problem, when the presumed underlying distribution is adversarially perturbed? To provide a concrete answer to this question, this paper unifies two major learning frameworks: Semi-Supervised Learning (SSL) and Distributionally Robust Learning (DRL). We develop a generalization theory for our framework based on a number of novel complexity measures, such as an adversarial extension of Rademacher complexity and its semi-supervised analogue. Moreover, our analysis is able to quantify the role of unlabeled data in the generalization under a more general condition compared to the existing theoretical works in SSL. Based on our framework, we also present a hybrid of DRL and EM algorithms that has a guaranteed convergence rate. When implemented with deep neural networks, our method shows a comparable performance to those of the state-of-the-art on a number of real-world benchmark datasets.

Author Information

Amir Najafi (Sharif University of Technology)

Amir Najafi received his B.Sc. and M.Sc. degrees in Electrical Engineering from Sharif University of Technology, Tehran, Iran, in 2012 and 2015, respectively. He is currently a Ph.D. student at Computer Engineering Dept. of Sharif University of Technology. From September 2015 to April 2016, he was with the Broad Institute of MIT and Harvard, MA, USA, as a visiting research scholar. In summer 2018, he was with Preferred Networks Inc., Tokyo, Japan, as an intern. His research interests include machine learning theory, information theory and bioinformatics.

Shin-ichi Maeda (Preferred Networks)
Masanori Koyama (Preferred Networks Inc. )
Takeru Miyato (Preferred Networks, Inc.)

More from the Same Authors