Manejo de datos OOD en la generación de imágenes sintéticas: revisión del estado del arte

Arbolaez Espinosa  Ariadna; Maria Matilde Garcia Lorenzo; Yusely Ruíz González; Rafael E. Bello Pérez

Authors

Arbolaez Espinosa Ariadna Universidad Central "Martha Abreu" de Las Villas
Maria Matilde Garcia Lorenzo Universidad Central "Martha Abreu" de Las Villas
Yusely Ruíz González Universidad Central "Martha Abreu" de Las Villas
Rafael E. Bello Pérez Universidad Central "Martha Abreu" de Las Villas

Keywords:

anomaly detection, generative models, model evaluation, synthetic image generation, out-of-

Abstract

Synthetic image generation through generative models (GANs, diffusion models) has seen remarkable progress, but the identification and handling of out-of-distribution (OOD) data remains a critical challenge for the reliable deployment of these models. This review examines the state of the art in OOD detection within the context of synthetic image generation. Widely used generative architectures such as Wasserstein GAN with gradient penalty (WGAN-GP) and StyleGAN2-ADA are analyzed, along with evaluation metrics like Fréchet Inception Distance (FID). Furthermore, recent strategies that enhance OOD discrimination without requiring OOD data during training are reviewed, such as SeTAR (which selectively adjusts internal representations) and ABET (which combines energy and learned temperature scaling). The literature synthesis indicates that integrating OOD detection techniques into generative models not only increases the models' ability to distinguish between in-distribution and out-of-distribution examples but also promotes greater visual fidelity and stability in synthetic images, especially in scenarios with limited or critical real data. The review concludes by identifying gaps in knowledge and proposing future research directions to strengthen the applicability of these models in sensitive domains.

References

Amodei, D. (2016). Concrete Problems in AI Safety. https://doi.org/https://doi.org/10.48550/arXiv.1606.06565

Barratt, S., & Sharma, R. (2018). A note on the inception score. ArXiv Preprint ArXiv:1801.01973.

Benny, Y., Galanti, T., Benaim, S., & Wolf, L. (2021). Evaluation Metrics for Conditional Image Generation. International Journal of Computer Vision, 129(5). https://doi.org/10.1007/s11263-020-01424-w

Chong, M. J., & Forsyth, D. (2020). Effectively unbiased fid and inception score and where to find them. Proceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition, 6070–6079.

Duy, P. T., Tien, L. K., Khoa, N. H., Hien, D. T. T., Nguyen, A. G. T., & Pham, V. H. (2021). DIGFuPAS: Deceive IDS with GAN and function-preserving on adversarial samples in SDN-enabled networks. Computers and Security, 109. https://doi.org/10.1016/j.cose.2021.102367

Frid-Adar, M., Diamant, I., Klang, E., Amitai, M., Goldberger, J., & Greenspan, H. (2018). GAN-based synthetic medical image augmentation for increased CNN performance in liver lesion classification. Neurocomputing, 321, 321–331.

Goodfellow, I. J., Pouget-Abadie, J., Mirza, M., Xu, B., Warde-Farley, D., Ozair, S., … Bengio, Y. (2014). Advances in Neural Information Processing Systems: Generative Adversarial Nets. Advances in Neural Information Processing Systems, 27.

Goodfellow, I., Pouget-Abadie, J., Mirza, M., Xu, B., Warde-Farley, D., Ozair, S., … Bengio, Y. (2020). Generative adversarial networks. Communications of the ACM, 63(11). https://doi.org/10.1145/3422622

Gulrajani, I., Ahmed, F., Arjovsky, M., Dumoulin, V., & Courville, A. (2017). Improved training of wasserstein GANs. Advances in Neural Information Processing Systems, 2017-December.

Heusel, M., Ramsauer, H., Unterthiner, T., Nessler, B., & Hochreiter, S. (2017). GANs trained by a two time-scale update rule converge to a local Nash equilibrium. Advances in Neural Information Processing Systems, 2017-December. https://doi.org/10.18034/ajase.v8i1.9

Ho, J., Jain, A., & Abbeel, P. (2020). Denoising diffusion probabilistic models. Advances in Neural Information Processing Systems, 2020-December.

Karras, T., Aittala, M., Hellsten, J., Laine, S., Lehtinen, J., & Aila, T. (2020). Training generative adversarial networks with limited data. Advances in Neural Information Processing Systems, 2020-December.

Kirichenko, P., Izmailov, P., & Wilson, A. G. (2020). Why normalizing flows fail to detect out-of-distribution data. Advances in Neural Information Processing Systems, 2020-December.

LeVine, W., Pikus, B., Phillips, J., Norman, B., Gil, F. A., & Hendryx, S. (2024). Out-of-Distribution Detection & Applications With Ablated Learned Temperature Energy. ArXiv Preprint ArXiv:2401.12129.

Li, Y., Chen, G., Chen, Y., & Xiong, B. (2024). SeTAR: Out-of-Distribution Detection with Selective Low-Rank Approximation. NeurIPS. https://doi.org/https://doi.org/10.48550/arXiv.2406.12629

Liu, W., Wang, X., Owens, J. D., & Li, Y. (2020). Energy-based out-of-distribution detection. Advances in Neural Information Processing Systems, 2020-December.

Mahapatra, D., Bozorgtabar, B., & Garnavi, R. (2019). Image super-resolution using progressive generative adversarial networks for medical image analysis. Computerized Medical Imaging and Graphics, 71. https://doi.org/10.1016/j.compmedimag.2018.10.005

Nalisnick, E., Matsukawa, A., Teh, Y. W., Gorur, D., & Lakshminarayanan, B. (2019). Do deep generative models know what they don’t know? 7th International Conference on Learning Representations, ICLR 2019.

Nitsch, J., Itkina, M., Senanayake, R., Nieto, J., Schmidt, M., Siegwart, R., … Cadena, C. (2021). Out-of-distribution detection for automotive perception. 2021 IEEE International Intelligent Transportation Systems Conference (ITSC), 2938–2943.

Rayavarapu, S. M., & Sasibhushana Rao, G. (2025). Exploración de modelos generativos profundos para una mejor generación de datos en la miocardiopatía hipertrófica. Ingenius Revista de Ciencia y Tecnología, (34).

Rombach, R., Blattmann, A., Lorenz, D., Esser, P., & Ommer, B. (2022). High-Resolution Image Synthesis with Latent Diffusion Models. Proceedings of the IEEE Computer Society Conference on Computer Vision and Pattern Recognition, 2022-June. https://doi.org/10.1109/CVPR52688.2022.01042

Sharma, P., Kumar, M., Sharma, H. K., & Biju, S. M. (2024). Generative adversarial networks (GANs): Introduction, Taxonomy, Variants, Limitations, and Applications. Multimedia Tools and Applications, 83(41). https://doi.org/10.1007/s11042-024-18767-y

Shin, H.-C., Tenenholtz, N. A., Rogers, J. K., Schwarz, C. G., Senjem, M. L., Gunter, J. L., … Michalski, M. (2018). Medical image synthesis for data augmentation and anonymization using generative adversarial networks. International Workshop on Simulation and Synthesis in Medical Imaging, 1–11.

Wang, Z., Ribeiro, F., & Mandt, S. (2024). On the Paradox of Learning to Reason from Data: Can We Learn Out-of-Distribution? Transactions on Machine Learning Research. https://doi.org/https://doi.org/10.48550/arXiv.2205.11502

Wibowo, M. C., Setyawan, I., Setiawan, A., & Sembiring, I. (2025). Deep Learning-Based Visualization of Network Threat Patterns Using GAN-Generated Infographic. Jurnal RESTI, 9(4). https://doi.org/10.29207/resti.v9i4.6717

Yang, J., Zhou, K., Li, Y., & Liu, Z. (2024). Generalized Out-of-Distribution Detection: A Survey. International Journal of Computer Vision, 132(12). https://doi.org/10.1007/s11263-024-02117-4

Yoon, J. S., Oh, K., Shin, Y., Mazurowski, M. A., & Suk, H.-I. (2024). Domain generalization for medical image analysis: A review. Proceedings of the IEEE.