Approximately Correct Label Distribution Learning

Label distribution learning (LDL) is a powerful learning paradigm that emulates label polysemy by assigning label distributions over the label space. However, existing LDL evaluation metrics struggle to capture meaningful performance differences due to their insensitivity to subtle distributional changes, and existing LDL learning objectives often exhibit biases by disproportionately emphasizing a small subset of samples with extreme predictions. As a result, the LDL metrics lose their discriminability, and the LDL objectives are also at risk of overfitting. In this paper, we propose DeltaLDL, a percentage of predictions that are approximately correct within the context of LDL, as a solution to the above problems. DeltaLDL can serve as a novel evaluation metric, which is parameter-free and reflects more on real performance improvements. DeltaLDL can also serve as a novel learning objective, which is differentiable and encourages most samples to be predicted as approximately correct, thereby mitigating overfitting. Our theoretical analysis and empirical results demonstrate the effectiveness of the proposed solution.

Teaching ML models to handle ambiguous labels, like recognizing someone is "70% happiness and 30% surprise", is tricky. Current evaluation metrics often miss small but important differences in these percentages, while current learning objectives focus too much on extreme cases (like 99% happiness). This makes the system less reliable and prone to overfitting.We introduce DeltaLDL, a new way to measure and train these systems. Instead of demanding perfect predictions, it rewards being "close enough", like accepting that 65-35 is reasonably close to 70-30. This makes evaluations more meaningful and reduces overfitting by encouraging balanced learning across all examples, not just the easiest or hardest ones.Tests show our method reflects real-world performance and improves generalization better. Think of it like grading a student on overall understanding rather than penalizing every small mistake. It’s a fairer and more practical approach for messy, real-life labeling tasks.