SADA: Safe and Adaptive Aggregation of Multiple Black-Box Predictions in Semi-Supervised Learning
arXiv:2509.21707v3 Announce Type: replace-cross Abstract: Semi-supervised learning (SSL) arises in practice when labeled data are scarce or expensive to obtain, while large quantities of unlabeled data are readily available. With the growing adoption of machine learning techniques, it has become increasingly feasible to generate multiple predicted labels using a variety of models and algorithms, including deep learning, large language models, and generative AI. In this paper, we propose a novel approach that safely and adaptively aggregates multiple black-box predictions of uncertain quality f
The proliferation of various AI models, including large language models and generative AI, necessitates robust methods for aggregating predictions safely and adaptively in semi-supervised learning environments.
This paper addresses a critical challenge in AI development by proposing a method for reliably combining outputs from uncertain black-box models, which is essential for deploying AI in sensitive or high-stakes applications.
This research introduces a novel approach to enhance the trustworthiness and performance of AI systems that rely on diverse, potentially fallible black-box prediction sources through safe and adaptive aggregation.
- · AI developers
- · Machine learning researchers
- · Industries using semi-supervised learning
- · Deep learning practitioners
- · Systems relying on naive black-box aggregation
- · Applications with high risk from erroneous AI predictions
Improved reliability and accuracy of AI systems integrating multiple generative or deep learning models.
Accelerated adoption of AI in sectors requiring high levels of assurance and robustness due to enhanced prediction aggregation capabilities.
Potential for new AI services and products built around robust black-box model ensembles, leading to more complex AI architectures and distributed intelligence.
This signal links to a primary source. Continuum Brief monitors and indexes it as part of the live intelligence stream — we do not republish source content.
Read at arXiv cs.LG