On Leveraging Unlabeled Data for Concurrent Positive-Unlabeled Classification and Robust Generation

TL;DR

This paper introduces a joint framework for positive-unlabeled classification and conditional data generation, effectively utilizing unlabeled data, including out-of-distribution samples, to improve both tasks.

Contribution

It proposes a novel training approach combining PU classification and robust conditional generation, leveraging extra unlabeled data and a noise-invariant GAN.

Findings

Enhanced PU classifier performance with CNI-CGAN

Effective utilization of out-of-distribution unlabeled data

Theoretical proof of optimal conditions for CNI-CGAN

Abstract

The scarcity of class-labeled data is a ubiquitous bottleneck in many machine learning problems. While abundant unlabeled data typically exist and provide a potential solution, it is highly challenging to exploit them. In this paper, we address this problem by leveraging Positive-Unlabeled~(PU) classification and the conditional generation with extra unlabeled data \emph{simultaneously}. We present a novel training framework to jointly target both PU classification and conditional generation when exposed to extra data, especially out-of-distribution unlabeled data, by exploring the interplay between them: 1) enhancing the performance of PU classifiers with the assistance of a novel Classifier-Noise-Invariant Conditional GAN~(CNI-CGAN) that is robust to noisy labels, 2) leveraging extra data with predicted labels from a PU classifier to help the generation. Theoretically, we prove the optimal condition of CNI-CGAN and experimentally, we conducted extensive evaluations on diverse datasets.