LEARNING FROM POSITIVE AND UNLABELED DATA USING OBSERVER-GAN
Omar Zamzam (University of Southern California); Haleh Akrami (Signal and Image Processing Institute at University of Southern California); Richard Leahy (Signal and Image Processing Institute at University of Southern California)
-
SPS
IEEE Members: $11.00
Non-members: $15.00
In the case of learning from positive and unlabeled data, the
input data consist of (1) observations from the positive class
and their corresponding labels and (2) unlabeled observations
from both positive and negative classes. Generative Adversar-
ial Networks (GANs) have been used to reduce the problem
to the supervised setting with the advantage that supervised
learning has state-of-the-art accuracy in classification tasks.
In order to generate pseudo-negative observations, GANs are
trained on positive and unlabeled observations with a modi-
fied loss. Using both positive and pseudo-negative observa-
tions leads to a supervised learning setting. The generation of
pseudo-negative observations that are realistic enough to re-
place missing negative class samples is a bottleneck for cur-
rent GAN-based algorithms. By including an additional clas-
sifier into the GAN architecture, we describe a novel GAN-
based approach. The GAN discriminator instructs the gen-
erator to only produce samples that fall into the unlabeled
data distribution, while a second classifier (observer) network
monitors the GAN training to: (i) prevent the generated sam-
ples from falling into the positive distribution; and (ii) learn
the features that are the key distinction between the posi-
tive and negative observations. Experiments on four image
datasets demonstrate that our trained observer network per-
forms better than existing techniques in discriminating be-
tween real unseen positive and negative samples.