MTSSL: Meta-Thresholding Semi-Supervised Learning
Provides theoretical insight into the role of thresholds in SSL, potentially simplifying future SSL algorithm design by relaxing the need for precise threshold tuning.
The paper establishes a unified theoretical framework explaining the role of the threshold τ in semi-supervised learning, showing that different τ values can yield similar performance. They propose MTSSL, which treats τ as an updatable parameter optimized via differentiation, achieving superior performance and demonstrating that τ selection can be relaxed.
A large body of Semi-supervised Learning~(SSL) algorithms encounter the threshold $τ$ to select pseudo-labels. The value of $τ$ across different SSL algorithms can vary depending on the learning perspective, yet they may achieve similar performance. It motivates us to establish a unified theoretical framework to explain the role of $τ$ in SSL. We statistically explained that the unsupervised loss is affected independently by correct and incorrect pseudo-labels, while $τ$ adjusts their numbers to balance the corresponding error term. This inherent trade-off indicates that SSL can reach the same loss with varying $τ$, precise optimal values of $τ$ during training may be unnecessary. With this, we treat $τ$ as an updatable parameter and optimize it via differentiation; the new policy is named \textbf{Meta-Thresholding Semi-Supervised Learning (MTSSL)}. Extensive experiments demonstrate the superior performance of MTSSL. We observe that the accuracy curves of SSL algorithms can overlap completely even when the values of $τ$ differ significantly, which supports our theoretical framework and indicates that the selection of $τ$ can be relaxed in the future design of SSL algorithms.