Semi-Supervised Learning in the Few-Shot Zero-Shot Scenario
This addresses a practical limitation in SSL for real-world scenarios with limited labeled data, though it is incremental as it augments existing methods.
The paper tackles the problem of semi-supervised learning when some classes are missing from the labeled data, proposing a method that adds a KL-divergence penalty to existing SSL objectives, resulting in significant accuracy improvements on CIFAR-100 and STL-10 datasets, especially with few labeled examples per class.
Semi-Supervised Learning (SSL) is a framework that utilizes both labeled and unlabeled data to enhance model performance. Conventional SSL methods operate under the assumption that labeled and unlabeled data share the same label space. However, in practical real-world scenarios, especially when the labeled training dataset is limited in size, some classes may be totally absent from the labeled set. To address this broader context, we propose a general approach to augment existing SSL methods, enabling them to effectively handle situations where certain classes are missing. This is achieved by introducing an additional term into their objective function, which penalizes the KL-divergence between the probability vectors of the true class frequencies and the inferred class frequencies. Our experimental results reveal significant improvements in accuracy when compared to state-of-the-art SSL, open-set SSL, and open-world SSL methods. We conducted these experiments on two benchmark image classification datasets, CIFAR-100 and STL-10, with the most remarkable improvements observed when the labeled data is severely limited, with only a few labeled examples per class