Privacy Preserving and Trustworthy Machine Learning
Despite the growing availability of high performance algorithmic tools for advanced statistical modelling and machine learning, solutions to many of the world’s most important problems are constrained by a lack of available scientific data. The causes for this disparity can be found within legal or ethical restrictions to data utilisation in fields such as medicine and within the competitive advantage provided by such data, discouraging data owners from sharing it. These constraints, although justifiable and necessary, often lead to questionable outcomes. For instance, many of the artificial intelligence (AI) algorithms approved by the Food and Drug Administration have been trained on datasets barely exceeding one thousand records.
In both aforementioned cases, the dilemma between data utilisation and data protection can be reconciled by reformulating its fundamental assumption: it is not access to data which is required, but access to the conclusions which can be drawn from this data.
Privacy Technologies such as Differential Privacy, Encrypted Computation and Distributed Learning can bridge this gap by providing the ability to train algorithms on data which is not locally available while maintaining governance and ownership, as well as allowing the provision of information-theoretic guarantees of privacy and confidentiality.
The Munich Data Science Institute is proclaiming the inaugural workshop on Privacy-Preserving and Trustworthy Machine Learning (PPTML).
[read more]
Coded Distributed Computation with Application to Machine Learning PI: A. Wachter-Zeh; DFG Individual Research Grants
COALESCENCE: Coded Computation for Large Scale Machine Learning with Privacy Guarantees PIs: A. Wachter-Zeh (TUM), D.Gunduz (ICL); TUM-ICL Joint Academy of Doctoral Studies (JADS)
Bitar, Rawad; Xing, Yuxuan; Keshtkarjahromi, Yasaman; Dasari, Venkat; El Rouayheb, Salim; Seferoglu, Hulya: Private and rateless adaptive coded matrix-vector multiplication. EURASIP Journal on Wireless Communications and Networking 2021 (1), 2021 more…
Full text (
DOI
)
Harder, Felix N.; Jungmann, Friederike; Kaissis, Georgios A.; Lohöfer, Fabian K.; Ziegelmayer, Sebastian; Havel, Daniel; Quante, Michael; Reichert, Maximillian; Schmid, Roland M.; Demir, Ihsan Ekin; Friess, Helmut; Wildgruber, Moritz; Siveke, Jens; Muckenhuber, Alexander; Steiger, Katja; Weichert, Wilko; Rauscher, Isabel; Eiber, Matthias; Makowski, Marcus R.; Braren, Rickmer F.: [18F]FDG PET/MRI enables early chemotherapy response prediction in pancreatic ductal adenocarcinoma. EJNMMI Research 11 (1), 2021 more…
Full text (
DOI
)
Hofmeister, Christoph; Bitar, Rawad; Xhemrishi, Marvin; Wachter-Zeh, Antonia: Secure Private and Adaptive Matrix Multiplication Beyond the Singleton Bound. 2021 more…
Hou, Benjamin; Kaissis, Georgios; Summers, Ronald; Kainz, Bernhard: RATCHET: Medical Transformer for Chest X-ray Diagnosis and Reporting. 2021 more…
Kaissis, Georgios; Ziller, Alexander; Passerat-Palmbach, Jonathan; Ryffel, Théo; Usynin, Dmitrii; Trask, Andrew; Lima, Ionésio, Jr; Mancuso, Jason; Jungmann, Friederike; Steinborn, Marc-Matthias; Saleh, Andreas; Makowski, Marcus; Rueckert, Daniel; Braren, Rickmer: End-to-end privacy preserving deep learning on multi-institutional medical imaging. Nature Machine Intelligence 3 (6), 2021, 473-484 more…
Full text (
DOI
)
Knolle, Moritz; Kaissis, Georgios; Jungmann, Friederike; Ziegelmayer, Sebastian; Sasse, Daniel; Makowski, Marcus; Rueckert, Daniel; Braren, Rickmer: Efficient, high-performance semantic segmentation using multi-scale feature extraction. PLOS ONE 16 (8), 2021, e0255397 more…
Full text (
DOI
)
Knolle, Moritz; Ziller, Alexander; Usynin, Dmitrii; Braren, Rickmer; Makowski, Marcus R.; Rueckert, Daniel; Kaissis, Georgios: Differentially private training of neural networks with Langevin dynamics for calibrated predictive uncertainty. 2021 more…
Meissen, Felix; Kaissis, Georgios; Rueckert, Daniel: Challenging Current Semi-Supervised Anomaly Segmentation Methods for Brain MRI. 2021 more…
Paetzold, Johannes C.; McGinnis, Julian; Shit, Suprosanna; Ezhov, Ivan; Büschl, Paul; Prabhakar, Chinmay; Todorov, Mihail I.; Sekuboyina, Anjany; Kaissis, Georgios; Ertürk, Ali; Günnemann, Stephan; Menze, Bjoern H.: Whole Brain Vessel Graphs: A Dataset and Benchmark for Graph Learning and Neuroscience (VesselGraph). 2021 more…
Thalheim, Jörg; Unnibhavi, Harshavardhan; Priebe, Christian; Bhatotia, Pramod; Pietzuch, Peter: rkt-io: a direct I/O stack for shielded execution. Proceedings of the Sixteenth European Conference on Computer Systems, ACM, 2021 more…
Full text (
DOI
)
Usynin, Dmitrii; Ziller, Alexander; Makowski, Marcus; Braren, Rickmer; Rueckert, Daniel; Glocker, Ben; Kaissis, Georgios; Passerat-Palmbach, Jonathan: Adversarial interference and its mitigations in privacy-preserving collaborative machine learning. Nature Machine Intelligence 3 (9), 2021, 749-758 more…
Full text (
DOI
)
Ziller, Alexander; Usynin, Dmitrii; Braren, Rickmer; Makowski, Marcus; Rueckert, Daniel; Kaissis, Georgios: Medical imaging deep learning with differential privacy. Scientific Reports 11 (1), 2021 more…
Full text (
DOI
)
Ziller, Alexander; Usynin, Dmitrii; Remerscheid, Nicolas; Knolle, Moritz; Makowski, Marcus; Braren, Rickmer; Rueckert, Daniel; Kaissis, Georgios: Differentially private federated deep learning for multi-site medical image segmentation. 2021 more…
2020
Bitar, Rawad; Parag, Parimal; El Rouayheb, Salim: Minimizing Latency for Secure Coded Computing Using Secret Sharing via Staircase Codes. IEEE Transactions on Communications 68 (8), 2020, 4609-4619 more…
Full text (
DOI
)
Quoc, Do Le; Gregor, Franz; Arnautov, Sergei; Kunkel, Roland; Bhatotia, Pramod; Fetzer, Christof: secureTF: A Secure TensorFlow Framework. Proceedings of the 21st International Middleware Conference, ACM, 2020 more…
Full text (
DOI
)