Binary Classification from Positive-Confidence Data

TL;DR

This paper introduces a novel binary classification method that learns from positive data with confidence scores, eliminating the need for negative or unlabeled data, and provides theoretical guarantees and practical neural network training results.

Contribution

It proposes a new positive-confidence classification framework with a simple empirical risk minimization approach, theoretical analysis, and demonstrated effectiveness in deep learning.

Findings

The method is consistent and has bounded estimation error.

It effectively trains deep neural networks using only positive-confidence data.

The approach outperforms traditional one-class classification in discrimination tasks.

Abstract

Can we learn a binary classifier from only positive data, without any negative data or unlabeled data? We show that if one can equip positive data with confidence (positive-confidence), one can successfully learn a binary classifier, which we name positive-confidence (Pconf) classification. Our work is related to one-class classification which is aimed at "describing" the positive class by clustering-related methods, but one-class classification does not have the ability to tune hyper-parameters and their aim is not on "discriminating" positive and negative classes. For the Pconf classification problem, we provide a simple empirical risk minimization framework that is model-independent and optimization-independent. We theoretically establish the consistency and an estimation error bound, and demonstrate the usefulness of the proposed method for training deep neural networks through experiments.