?url_ver=Z39.88-2004&rft_val_fmt=info%3Aofi%2Ffmt%3Akev%3Amtx%3Adc&rft.title=Distributional+Gaussian+Process+Layers+for+Outlier+Detection+in+Image+Segmentation&rft.creator=Popescu%2C+SG&rft.creator=Sharp%2C+DJ&rft.creator=Cole%2C+JH&rft.creator=Kamnitsas%2C+K&rft.creator=Glocker%2C+B&rft.description=We+propose+a+parameter+efficient+Bayesian+layer+for+hierarchical+convolutional+Gaussian+Processes+that+incorporates+Gaussian+Processes+operating+in+Wasserstein-2+space+to+reliably+propagate+uncertainty.+This+directly+replaces+convolving+Gaussian+Processes+with+a+distance-preserving+affine+operator+on+distributions.+Our+experiments+on+brain+tissue-segmentation+show+that+the+resulting+architecture+approaches+the+performance+of+well-established+deterministic+segmentation+algorithms+(U-Net)%2C+which+has+never+been+achieved+with+previous+hierarchical+Gaussian+Processes.+Moreover%2C+by+applying+the+same+segmentation+model+to+out-of-distribution+data+(i.e.%2C+images+with+pathology+such+as+brain+tumors)%2C+we+show+that+our+uncertainty+estimates+result+in+out-of-distribution+detection+that+outperforms+the+capabilities+of+previous+Bayesian+networks+and+reconstruction-based+approaches+that+learn+normative+distributions.&rft.publisher=Springer&rft.date=2021-01-01&rft.type=Proceedings+paper&rft.language=eng&rft.source=+++++In%3A++Information+Processing+in+Medical+Imaging.++(pp.+pp.+415-427).++Springer%3A+Cham%2C+Switzerland.+(2021)+++++&rft.format=text&rft.identifier=https%3A%2F%2Fdiscovery.ucl.ac.uk%2Fid%2Feprint%2F10134583%2F1%2FDistributional_Gaussian_Process_Layers_for_Outlier_Detection_in_Image_Segmentation____IPMI_version.pdf&rft.identifier=https%3A%2F%2Fdiscovery.ucl.ac.uk%2Fid%2Feprint%2F10134583%2F&rft.rights=open