analiza aiqbench
This paper introduces a novel Dual-Primal Graph VAE architecture designed to infer ground-truth labels from noisy crowdsourced data using GAT-based message passing on both the dataset graph and its dual. By treating ground-truth labels as latent variables, the method achieves unsupervised representation learning that outperforms existing models on benchmark tasks while effectively incorporating side information from pre-trained classifiers. The work advances the theoretical and practical limits of handling label noise in machine learning pipelines without relying on synthetic data or separate classifier training.
Dual-Primal Graph VAEs for Noisy Label Aggregation
arXiv:2608.11473v2 Announce Type: replace Abstract: Inferring the ground-truth from noisy crowdsourced labels is an important theoretical and practical problem. Neural network-based methods offer an alternative to classical Bayesian models which require specifying a family of generative models used for inference. However, current models either still rely on fairly simple generative models for inference or require pseudo-labels or synthetic data to train the aggregate classifier. We propose a graph VAE architecture in which the decoder and encoder use GAT-based message passing on the adjacency graph of a crowdsourced dataset and its dual, respectively. The ground-truth labels are treated as latent variables, enabling unsupervised representation learning without needing to train a separate classifier. We show our model achieves state of the art performance on crowdsourcing benchmarks. We then demonstrate the generality of our approach by showing how the original crowdsourcing graph can be augmented to incorporate side information such as representations from neural network classifiers trained on the noisy labels to substantially boost their classification performance at test time.
Abstract:Inferring the ground-truth from noisy crowdsourced labels is an important theoretical and practical problem. Neural network-based methods offer an alternative to classical Bayesian models which require specifying a family of generative models used for inference. However, current models either still rely on fairly simple generative models for inference or require pseudo-labels or synthetic data to train the aggregate classifier. We propose a graph VAE architecture in which the decoder and encoder use GAT-based message passing on the adjacency graph of a crowdsourced dataset and its dual, respectively. The ground-truth labels are treated as latent variables, enabling unsupervised representation learning without needing to train a separate classifier. We show our model achieves state of the art performance on crowdsourcing benchmarks. We then demonstrate the generality of our approach by showing how the original crowdsourcing graph can be augmented to incorporate side information such as representations from neural network classifiers trained on the noisy labels to substantially boost their classification performance at test time.
Machine Learning (cs.LG)
arXiv:2608.11473 [cs.LG]
(or arXiv:2608.11473v2 [cs.LG] for this version)
https://doi.org/10.48550/arXiv.2608.11473 Focus to learn more arXiv-issued DOI via DataCite
arXivLabs is a framework that allows collaborators to develop and share new arXiv features directly on our website.
Both individuals and organizations that work with arXivLabs have embraced and accepted our values of openness, community, excellence, and user data privacy. arXiv is committed to these values and only works with partners that adhere to them.
Have an idea for a project that will add value for arXiv's community? Learn more about arXivLabs.
Text integral sub licența deschisă a publicației · 254 cuvinte
Subiecte conexe
Build a Physical AI model factory with NVIDIA Cosmos 3 on SageMaker HyperPod
AWS · 4 sept. 2026
Equivariant Neural Primal-Dual Assignment for Maximum Common Edge Subgraphs
arXiv · 30 sept. 2026
AHEAD: Advancing Multi-Class Label Aggregation with Interpretable Cross-Annotator Modeling
arXiv · 22 iul. 2026
PaSta: Noisy Node Classification with Partial Label Learning
arXiv · 27 aug. 2026