Identifying noisy labels with a transductive semi-supervised leave-one-out filter
This addresses label noise in semi-supervised learning for real-world applications where labeling is error-prone, but it is incremental as it builds on existing methods like LGC.
The paper tackles the problem of label noise in semi-supervised learning by introducing LGC_LVOF, a leave-one-out filtering method that detects and removes wrong labels as a preprocessing step. Results show it is equally or more precise than an adapted gradient-based filter and yields accuracy comparable to robust classifiers on datasets like MNIST and ISOLET.
Obtaining data with meaningful labels is often costly and error-prone. In this situation, semi-supervised learning (SSL) approaches are interesting, as they leverage assumptions about the unlabeled data to make up for the limited amount of labels. However, in real-world situations, we cannot assume that the labeling process is infallible, and the accuracy of many SSL classifiers decreases significantly in the presence of label noise. In this work, we introduce the LGC_LVOF, a leave-one-out filtering approach based on the Local and Global Consistency (LGC) algorithm. Our method aims to detect and remove wrong labels, and thus can be used as a preprocessing step to any SSL classifier. Given the propagation matrix, detecting noisy labels takes O(cl) per step, with c the number of classes and l the number of labels. Moreover, one does not need to compute the whole propagation matrix, but only an $l$ by $l$ submatrix corresponding to interactions between labeled instances. As a result, our approach is best suited to datasets with a large amount of unlabeled data but not many labels. Results are provided for a number of datasets, including MNIST and ISOLET. LGCLVOF appears to be equally or more precise than the adapted gradient-based filter. We show that the best-case accuracy of the embedding of LGCLVOF into LGC yields performance comparable to the best-case of $\ell_1$-based classifiers designed to be robust to label noise. We provide a heuristic to choose the number of removed instances.