Learning from Indirect Observations

TL;DR

This paper introduces a probabilistic framework for weakly-supervised learning from various indirect observations, enabling models to learn effectively from noisy, coarse, or complementary labels with theoretical guarantees and practical implementation.

Contribution

It presents a general maximum likelihood-based method for learning from diverse weak supervision signals, extending beyond traditional single-source approaches.

Findings

Effective in learning from noisy labels and coarse-grained labels.

Demonstrates practical utility in two novel problem settings.

Theoretically sound with straightforward deep learning implementation.

Abstract

Weakly-supervised learning is a paradigm for alleviating the scarcity of labeled data by leveraging lower-quality but larger-scale supervision signals. While existing work mainly focuses on utilizing a certain type of weak supervision, we present a probabilistic framework, learning from indirect observations, for learning from a wide range of weak supervision in real-world problems, e.g., noisy labels, complementary labels and coarse-grained labels. We propose a general method based on the maximum likelihood principle, which has desirable theoretical properties and can be straightforwardly implemented for deep neural networks. Concretely, a discriminative model for the true target is used for modeling the indirect observation, which is a random variable entirely depending on the true target stochastically or deterministically. Then, maximizing the likelihood given indirect observations leads to an estimator of the true target implicitly. Comprehensive experiments for two novel problem settings --- learning from multiclass label proportions and learning from coarse-grained labels, illustrate practical usefulness of our method and demonstrate how to integrate various sources of weak supervision.