aiqbench analysis
This paper introduces MTSSL, a novel approach in semi-supervised learning (SSL) that treats the pseudo-label threshold as an updatable parameter rather than a fixed hyperparameter. By optimizing this value through differentiation and establishing a unified theoretical framework, the method demonstrates superior performance while relaxing the need for precise manual tuning of thresholds across different algorithms. The work advances fundamental machine learning theory by showing that accuracy curves can overlap significantly even with varying threshold values, simplifying future SSL algorithm designs.
MTSSL: Meta-Thresholding Semi-Supervised Learning
arXiv:2607.16363v1 Announce Type: cross Abstract: A large body of Semi-supervised Learning~(SSL) algorithms encounter the threshold $\tau$ to select pseudo-labels. The value of $\tau$ 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 $\tau$ in SSL. We statistically explained that the unsupervised loss is affected independently by correct and incorrect pseudo-labels, while $\tau$ adjusts their numbers to balance the corresponding error term. This inherent trade-off indicates that SSL can reach the same loss with varying $\tau$, precise optimal values of $\tau$ during training may be unnecessary. With this, we treat $\tau$ 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 $\tau$ differ significantly, which supports our theoretical framework and indicates that the selection of $\tau$ can be relaxed in the future design of SSL algorithms.
Abstract:A large body of Semi-supervised Learning~(SSL) algorithms encounter the threshold $\tau$ to select pseudo-labels. The value of $\tau$ 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 $\tau$ in SSL. We statistically explained that the unsupervised loss is affected independently by correct and incorrect pseudo-labels, while $\tau$ adjusts their numbers to balance the corresponding error term. This inherent trade-off indicates that SSL can reach the same loss with varying $\tau$, precise optimal values of $\tau$ during training may be unnecessary. With this, we treat $\tau$ 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 $\tau$ differ significantly, which supports our theoretical framework and indicates that the selection of $\tau$ can be relaxed in the future design of SSL algorithms.
Machine Learning (stat.ML); Machine Learning (cs.LG)
arXiv:2607.16363 [stat.ML]
(or arXiv:2607.16363v1 [stat.ML] for this version)
https://doi.org/10.48550/arXiv.2607.16363 Focus to learn more arXiv-issued DOI via DataCite (pending registration)
arXivLabs is a framework that allows collaborators to develop and share new arXiv features directly on our website.
Both individuals and organizations that work with arXivLabs have embraced and accepted our values of openness, community, excellence, and user data privacy. arXiv is committed to these values and only works with partners that adhere to them.
Have an idea for a project that will add value for arXiv's community? Learn more about arXivLabs.
Full text under the publisher's open licence · 272 words
Related coverage
SLUM-i: Semi-supervised Learning for Urban Mapping of Informal Settlements and Data Quality Benchmarking
arXiv · Jun 16, 2026
Structuring Relations Among Learning Paradigms via Protocol--Objective--Resource Reductions
arXiv · Sep 30, 2026
Imbalanced Semi-Supervised Learning via Label Refinement and Threshold Adjustment
arXiv · Jun 16, 2026
ReCalMatch:Reliability-Calibrated Semantic Guidance for Semi-Supervised Fine-Grained Recognition
arXiv · Sep 25, 2026