arXiv:2605.21094v1 Announce Type: new Abstract: We investigate unpaired image inverse problems, a challenging setting where only independent, non-paired sets of noisy measurements and clean target signals are available for training. We propose a novel inverse problem solver based on Unbalanced Optimal Transport, called Unbalanced Optimal Transport Map for Inverse Problems (UOTIP). Our method formulates the reconstruction task, predicting clean target signals from noisy measurements, as learning a UOT Map from noisy measurement distribution to clean signal distribution by incorporating a likeli
The increasing availability of large, unpaired datasets and advances in optimal transport theory are enabling new approaches to challenging inverse problems in AI.
This research provides a novel method for AI to reconstruct clean signals from diverse noisy measurements without paired training data, which is common in many real-world applications.
The ability to learn effectively from unpaired data sets could significantly broaden the applicability and robustness of AI in fields like medical imaging, environmental sensing, and security.
- · AI researchers
- · Medical imaging industry
- · Security and surveillance sectors
- · Environmental monitoring
- · Methods heavily reliant on perfectly paired datasets
- · AI solutions with high data labeling costs
Improved performance of AI systems in tasks where paired data is scarce or impossible to obtain.
Reduced data collection and labeling burdens for various AI applications, accelerating deployment in new domains.
Enhanced AI capabilities for real-time anomaly detection and prediction from noisy, heterogenous data streams without human supervision.
This signal links to a primary source. Continuum Brief monitors and indexes it as part of the live intelligence stream — we do not republish source content.
Read at arXiv cs.LG