Case-Study: A Generative Approach to Anomaly Detection


Our client was an SME renowned for its solutions validating the ingenuity of user-supplied data. They were interested in anomaly detection for various reasons, including the potential to detect data drifting.


We provided an imaging-based anomaly detection solution that combined the capacity to model (“understand” if you wish) the underlying distribution as well as detect discrepancies relative to it. As we embarked on this mission, we extended our usual toolset of variational and adversarial auto-encoders, for the first time with our custom diffusion models.

The resulting 100% custom codebase was developed mostly in PyTorch.


The client was happy with the expansion of their existing toolbox and decided to integrate it into their own toolkit. Their team took over our PoC for further productization.

Image by Freepik
Image by Freepik

Want to dive deeper into the world of artificial intelligence and machine learning? Our blog page is your go-to destination for comprehensive insights, practical guides, and expert perspectives on the latest trends and developments in AI technology. Whether you’re a seasoned professional or just starting out in the field, our blog offers something for everyone. From in-depth tutorials to thought-provoking analysis, we cover a wide range of topics to help you stay informed and ahead of the curve. Join our community of AI enthusiasts and explore the fascinating world of cutting-edge technology. Don’t miss out on the opportunity to expand your knowledge and enhance your skills—visit our blog today!