AsyCo: An Asymmetric Dual-task Co-training Model for Partial-label Learning

TL;DR

AsyCo introduces an asymmetric dual-task co-training model for partial-label learning, effectively reducing error accumulation by training two networks with different tasks and views, leading to improved performance on various datasets.

Contribution

The paper proposes a novel asymmetric co-training framework with distinct tasks for each network, addressing error accumulation in partial-label learning.

Findings

Outperforms existing PLL methods on multiple datasets.

Effectively mitigates error accumulation through information distillation.

Demonstrates robustness on both uniform and instance-dependent datasets.

Abstract

Partial-Label Learning (PLL) is a typical problem of weakly supervised learning, where each training instance is annotated with a set of candidate labels. Self-training PLL models achieve state-of-the-art performance but suffer from error accumulation problem caused by mistakenly disambiguated instances. Although co-training can alleviate this issue by training two networks simultaneously and allowing them to interact with each other, most existing co-training methods train two structurally identical networks with the same task, i.e., are symmetric, rendering it insufficient for them to correct each other due to their similar limitations. Therefore, in this paper, we propose an asymmetric dual-task co-training PLL model called AsyCo, which forces its two networks, i.e., a disambiguation network and an auxiliary network, to learn from different views explicitly by optimizing distinct tasks. Specifically, the disambiguation network is trained with self-training PLL task to learn label confidence, while the auxiliary network is trained in a supervised learning paradigm to learn from the noisy pairwise similarity labels that are constructed according to the learned label confidence. Finally, the error accumulation problem is mitigated via information distillation and confidence refinement. Extensive experiments on both uniform and instance-dependent partially labeled datasets demonstrate the effectiveness of AsyCo. The code is available at https://github.com/libeibeics/AsyCo.