# CAVA: Complementary-Label-Aware Data Valuation

DevFeed: [CAVA: Complementary-Label-Aware Data Valuation](<https://devfeed.tech/articles/cava-complementary-label-aware-data-valuation-93666.md>)

Original publisher: [Read original article](<https://arxiv.org/abs/2610.26734>)

Author: Tan-Ha Mai, Hsuan-Tien Lin, Vu Nguyen

Published: 2026-10-21T04:00:00Z

Content type: article

Language: en

Sources: [arXiv: Machine Learning](<https://devfeed.tech/sources/arxiv-machine-learning.md>)

Topics: [Machine learning](<https://devfeed.tech/topics/machine-learning.md>)

Tags: [annotation](<https://devfeed.tech/tags/annotation.md>), [benchmarks](<https://devfeed.tech/tags/benchmarks.md>), [cost](<https://devfeed.tech/tags/cost.md>), [cs-cv](<https://devfeed.tech/tags/cs-cv.md>), [cs-lg](<https://devfeed.tech/tags/cs-lg.md>), [learning](<https://devfeed.tech/tags/learning.md>), [margin](<https://devfeed.tech/tags/margin.md>), [noise](<https://devfeed.tech/tags/noise.md>), [optimal-transport](<https://devfeed.tech/tags/optimal-transport.md>), [precision](<https://devfeed.tech/tags/precision.md>), [random](<https://devfeed.tech/tags/random.md>), [supervision](<https://devfeed.tech/tags/supervision.md>), [synthetic](<https://devfeed.tech/tags/synthetic.md>), [training](<https://devfeed.tech/tags/training.md>), [validation](<https://devfeed.tech/tags/validation.md>), [valuation](<https://devfeed.tech/tags/valuation.md>), [vision-language-models-vlms](<https://devfeed.tech/tags/vision-language-models-vlms.md>)

## AI overview

CAVA addresses annotation errors in complementary-label learning (CLL), where mistakenly naming the true class as the class an instance does not belong to can train a model to suppress the correct answer. Directly transferring optimal-transport data valuation from ordinary labels fails because training and clean validation labels have opposite meanings: its score reverses below the 1/M noise crossover, a range the abstract says is typical in practice. CAVA adds a complementary-label cost and label margin to isolate each label's contribution to the transport potential. The margin retains a consistent direction across noise rates, supports diagonal-noise detection with a parameter-free threshold, and lets detected samples be reused as ordinary-label supervision. On synthetic and human-annotated benchmarks, the authors report near-perfect separation of diagonal noise where transferred OT performs worse than random; with a fraction of clean validation data, CAVA improves every evaluated CLL objective.

## Source excerpt

arXiv:2610.26734v1 Announce Type: new Abstract: Complementary-label learning (CLL) trains a classifier from labels that name a class an instance does not belong to. Such labels are cheap to collect from crowdworkers or vision-language models (VLMs), making them attractive for large-scale annotation. However, both sources can mistakenly assign the true class as the complementary label. Such annotation errors are particularly harmful in CLL because the model is explicitly trained to suppress the correct answer. Optimal transport (OT) data valuation identifies harmful samples by comparing training data with a clean validation set, but transferring it directly to CLL creates a semantic mismatch: label $c$ means not class $c$ in the training but class $c$ in the validation. We show that this makes the OT label groups noisy mixtures whose behavior depends on the noise rate. The transferred score is informative only at sufficiently high noise, becomes uninformative at the crossover rate $1/M$ for $M$ classes, and reverses below it, where practical CLL annotation typically operates. We propose CAVA, which introduces a complementary label cost and a label margin that isolates the contribution of each complementary label to the transport potential. The margin keeps a consistent direction across noise rates, detects diagonal noise with a parameter-free threshold of closed-form precision, and allows detected samples to be recycled as ordinary-label supervision. On synthetic and human-annotated benchmarks, CAVA separates diagonal noise almost perfectly where directly transferred OT valuation performs worse than random. With only a fraction of the clean validation data, it also improves every CLL objective we evaluate, showing that OT-based valuation becomes reliable once the semantics of complementary supervision are modeled explicitly.