Abstract:Noise transition matrix estimation is a promising approach for learning with label noise. It can infer clean posterior probabilities, known as Label Distribution (LD), based on noisy ones and reduce the impact of noisy labels. However, this estimation is challenging, since the ground truth labels are not always available. Most existing methods estimate a global noise transition matrix using either correctly labeled samples (anchor points) or detected reliable samples (pseudo anchor points). These methods heavily rely on the existence of anchor points or the quality of pseudo ones, and the global noise transition matrix can hardly provide accurate label transition information for each sample, since the label noise in real applications is mostly instance-dependent. To address these challenges, we propose an Instance-dependent Label Distribution Estimation (ILDE) method to learn from noisy labels for image classification. The method's workflow has three major steps. First, we estimate each sample's noisy posterior probability, supervised by noisy labels. Second, since mislabeling probability closely correlates with inter-class correlation, we compute the inter-class correlation matrix to estimate the noise transition matrix, bypassing the need for (pseudo) anchor points. Moreover, for a precise approximation of the instance-dependent noise transition matrix, we calculate the inter-class correlation matrix using only mini-batch samples rather than the entire training dataset. Third, we transform the noisy posterior probability into instance-dependent LD by multiplying it with the estimated noise transition matrix, using the resulting LD for enhanced supervision to prevent DCNNs from memorizing noisy labels. The proposed ILDE method has been evaluated against several state-of-the-art methods on two synthetic and three real-world noisy datasets. Our results indicate that the proposed ILDE method outperforms all competing methods, no matter whether the noise is synthetic or real noise.

Learning with Bounded Instance- and Label-dependent Label Noise

Instance-dependent Label Distribution Estimation for Learning with Label Noise

Instance-Dependent Inaccurate Label Distribution Learning.

Feature-Induced Label Distribution for Learning with Noisy Labels

Learning with Instance-Dependent Noisy Labels by Anchor Hallucination and Hard Sample Label Correction

Label Noise: Ignorance Is Bliss

Instance-dependent Label-noise Learning under a Structural Causal Model

Tackling Instance-Dependent Label Noise Via a Universal Probabilistic Model.

Part-dependent Label Noise: Towards Instance-dependent Label Noise

Instance-specific Label Distribution Regularization for Learning with Label Noise

Learning With Noisy Labels Over Imbalanced Subpopulations

Inaccurate Label Distribution Learning with Dependency Noise

Learning with Instance-Dependent Label Noise: A Sample Sieve Approach

Multi-Instance Learning with One Side Label Noise

Instance-dependent Noisy-label Learning with Graphical Model Based Noise-rate Estimation

BadLabel: A Robust Perspective on Evaluating and Enhancing Label-noise Learning

IRNet: Iterative Refinement Network for Noisy Partial Label Learning

Confidence Scores Make Instance-dependent Label-noise Learning Possible

Tackling Instance-Dependent Label Noise with Dynamic Distribution Calibration