%0 Generic %1 zimmerer2019unsupervised %A Zimmerer, David %A Isensee, Fabian %A Petersen, Jens %A Kohl, Simon %A Maier-Hein, Klaus %D 2019 %K cs.CV cs.LG stat.ML %T Unsupervised Anomaly Localization using Variational Auto-Encoders %U http://arxiv.org/abs/1907.02796 %X An assumption-free automatic check of medical images for potentially overseen anomalies would be a valuable assistance for a radiologist. Deep learning and especially Variational Auto-Encoders (VAEs) have shown great potential in the unsupervised learning of data distributions. In principle, this allows for such a check and even the localization of parts in the image that are most suspicious. Currently, however, the reconstruction-based localization by design requires adjusting the model architecture to the specific problem looked at during evaluation. This contradicts the principle of building assumption-free models. We propose complementing the localization part with a term derived from the Kullback-Leibler (KL)-divergence. For validation, we perform a series of experiments on FashionMNIST as well as on a medical task including >1000 healthy and >250 brain tumor patients. Results show that the proposed formalism outperforms the state of the art VAE-based localization of anomalies across many hyperparameter settings and also shows a competitive max performance.

@misc{zimmerer2019unsupervised, abstract = {An assumption-free automatic check of medical images for potentially overseen anomalies would be a valuable assistance for a radiologist. Deep learning and especially Variational Auto-Encoders (VAEs) have shown great potential in the unsupervised learning of data distributions. In principle, this allows for such a check and even the localization of parts in the image that are most suspicious. Currently, however, the reconstruction-based localization by design requires adjusting the model architecture to the specific problem looked at during evaluation. This contradicts the principle of building assumption-free models. We propose complementing the localization part with a term derived from the Kullback-Leibler (KL)-divergence. For validation, we perform a series of experiments on FashionMNIST as well as on a medical task including >1000 healthy and >250 brain tumor patients. Results show that the proposed formalism outperforms the state of the art VAE-based localization of anomalies across many hyperparameter settings and also shows a competitive max performance.}, added-at = {2020-12-19T12:16:16.000+0100}, author = {Zimmerer, David and Isensee, Fabian and Petersen, Jens and Kohl, Simon and Maier-Hein, Klaus}, biburl = {https://www.bibsonomy.org/bibtex/225fdb6063c472635406703288367fdad/aerover}, description = {Unsupervised Anomaly Localization using Variational Auto-Encoders}, interhash = {f58be860120aac5f9d330a06fbfc4c3f}, intrahash = {25fdb6063c472635406703288367fdad}, keywords = {cs.CV cs.LG stat.ML}, note = {cite arxiv:1907.02796}, timestamp = {2021-01-17T01:33:27.000+0100}, title = {Unsupervised Anomaly Localization using Variational Auto-Encoders}, url = {http://arxiv.org/abs/1907.02796}, year = 2019 }