Published in arxiv, 2024
We propose a novel multimodal VAE that enables multimodal capabilities by only soft-sharing of information between modalities.
Recommended citation: TM Sutter, Y Meng, A Agostini, D Chopard, N Fortin, JE Vogt, B Shahbaba, S Mandt (2024) "Unity by Diversity: Improved Representation Learning in Multimodal VAEs." arxiv. https://arxiv.org/abs/2403.05300
Published in arxiv, 2024
We propose a novel approach to anomaly detection using contrastive approaches to learning representations of normal samples.
Recommended citation: A Ryser, TM Sutter, A Marx, JE Vogt (2024) "Anomaly Detection by Context Contrasting." arxiv. https://arxiv.org/abs/2405.18848
Published in NeurIPS, 2023
We propose a differentiable two-stage formulation for random partition models that allows gradient-based optimization of the distribution parameters.
Recommended citation: TM Sutter, A Ryser, J Liebeskind, JE Vogt. (2023) "Differentiable Random Partition Models." NeurIPS 2023. https://arxiv.org/abs/2305.16841
Published in GCPR, 2023
Artificial (M)otion-mode based predicting of left ventricular ejection fraction to diagnose cardiomyopathy.
Recommended citation: E Ozkan, TM Sutter, Y Hu, S Balzer, JE Vogt. (2023) "M(otion)-mode Based Prediction of Ejection Fraction using Echocardiograms." GCPR 2023. https://www.dagm-gcpr.de/fileadmin/dagm-gcpr/pictures/2023_Heidelberg/Paper_MainTrack/053.pdf
Published in ICLR 2023 (Spotlight Presentation), 2023
We propose a differentiable formulation for the multivariate hypergeometric distribution.
Recommended citation: TM Sutter, L Manduchi, A Ryser, JE Vogt. (2023). "Learning Group Importance using the Differentiable Hypergeometric Distribution" ICLR 2023. https://arxiv.org/abs/2203.01629
Published in Medical Imaging meets NeurIPS Workshop 2022, 2022
Artificial (M)otion-mode based predicting of left ventricular ejection fraction to diagnose cardiomyopathy.
Recommended citation: TM Sutter, S Balzer, E Ozkan, JE Vogt. (2022) "M(otion)-mode Based Prediction of Cardiac Function on Echocardiograms." Medical Imaging meets NeurIPS Workshop 2022. https://www.research-collection.ethz.ch/handle/20.500.11850/588779
Published in ICLR 2022, 2022
We show limitations of current scalable, multimodal VAEs.
Recommended citation: I Daunhawer, TM Suttter, K Chin-Cheong, E Palumbo, JE Vogt. (2022). "On the limitations of multimodal VAEs." ICLR 2022. https://arxiv.org/abs/2110.04121
Published in Neurips 2021 Workshop on Bayesian Deep Learning, 2021
Explicitely modelling and learning the relation between multiple data types.
Recommended citation: TM Sutter, JE Vogt. (2021). "Multimodal Relational VAE." Neurips 2021 Workshop on Bayesian Deep Learning. https://www.research-collection.ethz.ch/bitstream/handle/20.500.11850/520271/1/bdl_draft_cameraready_20211203.pdf
Published in MIDL 2021 Short Paper, 2021
Evaluating MoPoE-VAE on the Mimic-Cxr database.
Recommended citation: HJ Klug, TM Sutter, JE Vogt. (2021). "Multimodal Generative Learning on the Mimic-CXR database." MIDL 2021 Short Paper. https://openreview.net/pdf?id=ZVqjoKVbYMl
Published in ICLR 2021, 2021
Generalizing previous formulations of scalable multimodal V
Recommended citation: TM Sutter, I Daunhawer, JE Vogt. (2021). "Generalized multimodal ELBO." ICLR 2021. https://arxiv.org/abs/2110.04121
Published in Frontiers in Cellular and Infection Microbiology, 2021
Identifying optimal antigen-cytokine combinations using novel Mycobacterium tuberculosis antigens and cytokine read-outs.
Recommended citation: NR Meier, TM Sutter, M Jacobsen, THM Ottenhoff, JE Vogt, N Ritz. (2021) "Machine Learning Algorithms Evaluate Immune Response to Novel Mycobacterium tuberculosis Antigens for Diagnosis of Tuberculosis." Front. Cell. Infect. Microbiol.. 10(2020). https://www.frontiersin.org/articles/10.3389/fcimb.2020.594030/full
Published in Journal of the American Medical Informatics Association, 2020
Comparing different models with respect to their predictive performance for the detection of unplanned early readmissions.
Recommended citation: TM Sutter, JA Roth, K Chin-Cheong, BL Hug, JE Vogt. (2020). "A comparison of general and disease-specific machine learning models for the prediction of unplanned hospital readmissions." Journal of the American Medical Informatics Association, Volume 28, Issue 4, April 2021. https://academic.oup.com/jamia/article-abstract/28/4/868/6041742
Published in Neurips 2020, 2020
Utilizing the Jensen-Shannon divergence in a VAE setting to learn from multiple data types.
Recommended citation: TM Sutter, I Daunhawer, JE Vogt. (2020). "Multimodal Generative Learning Utilizing the Jensen-Shannon Divergence." Neurips 2020. https://proceedings.neurips.cc/paper/2020/hash/43bb733c1b62a5e374c63cb22fa457b4-Abstract.html
Published in GCPR 2020, 2020
Disentangling shared and modality-specific latent factors in multimodal data.
Recommended citation: I Daunhawer, TM Sutter, R Marcinkevics, JE Vogt. (2020). "Self-supervised Disentanglement of Modality-Specific and Shared Factors Improves Multimodal Generative Models." GCPR 2020. https://link.springer.com/chapter/10.1007/978-3-030-71278-5_33
Published in Arxiv, 2020
Generating differentially private medical records.
Recommended citation: K Chin-Cheong, TM Sutter, JE Vogt. (2020). "Generation of Differentially Private Heterogeneous Electronic Health Records." arxiv preprint. https://arxiv.org/abs/2006.03423