Responsible and Regulatory Conform Machine Learning for Medicine: A Survey of Challenges and Solutions
Authors:
Eike Petersen,
Yannik Potdevin,
Esfandiar Mohammadi,
Stephan Zidowitz,
Sabrina Breyer,
Dirk Nowotka,
Sandra Henn,
Ludwig Pechmann,
Martin Leucker,
Philipp Rostalski,
Christian Herzog
Abstract:
Machine learning is expected to fuel significant improvements in medical care. To ensure that fundamental principles such as beneficence, respect for human autonomy, prevention of harm, justice, privacy, and transparency are respected, medical machine learning systems must be developed responsibly. Many high-level declarations of ethical principles have been put forth for this purpose, but there i…
▽ More
Machine learning is expected to fuel significant improvements in medical care. To ensure that fundamental principles such as beneficence, respect for human autonomy, prevention of harm, justice, privacy, and transparency are respected, medical machine learning systems must be developed responsibly. Many high-level declarations of ethical principles have been put forth for this purpose, but there is a severe lack of technical guidelines explicating the practical consequences for medical machine learning. Similarly, there is currently considerable uncertainty regarding the exact regulatory requirements placed upon medical machine learning systems. This survey provides an overview of the technical and procedural challenges involved in creating medical machine learning systems responsibly and in conformity with existing regulations, as well as possible solutions to address these challenges. First, a brief review of existing regulations affecting medical machine learning is provided, showing that properties such as safety, robustness, reliability, privacy, security, transparency, explainability, and nondiscrimination are all demanded already by existing law and regulations - albeit, in many cases, to an uncertain degree. Next, the key technical obstacles to achieving these desirable properties are discussed, as well as important techniques to overcome these obstacles in the medical context. We notice that distribution shift, spurious correlations, model underspecification, uncertainty quantification, and data scarcity represent severe challenges in the medical context. Promising solution approaches include the use of large and representative datasets and federated learning as a means to that end, the careful exploitation of domain knowledge, the use of inherently transparent models, comprehensive out-of-distribution model testing and verification, as well as algorithmic impact assessments.
△ Less
Submitted 9 June, 2022; v1 submitted 20 July, 2021;
originally announced July 2021.
Evasive Windows Malware: Impact on Antiviruses and Possible Countermeasures
Authors:
Cédric Herzog,
Valérie Viet Triem Tong,
Pierre Wilke,
Arnaud van Straaten,
Jean-Louis Lanet
Abstract:
The perpetual opposition between antiviruses and malware leads both parties to evolve continuously. On the one hand, antiviruses put in place solutions that are more and more sophisticated and propose more complex detection techniques in addition to the classic signature analysis. This sophistication leads antiviruses to leave more traces of their presence on the machine they protect. To remain un…
▽ More
The perpetual opposition between antiviruses and malware leads both parties to evolve continuously. On the one hand, antiviruses put in place solutions that are more and more sophisticated and propose more complex detection techniques in addition to the classic signature analysis. This sophistication leads antiviruses to leave more traces of their presence on the machine they protect. To remain undetected as long as possible, malware can avoid executing within such environments by hunting down the modifications left by the antiviruses. This paper aims at determining the possibilities for malware to detect the antiviruses and then evaluating the efficiency of these techniques on a panel of antiviruses that are the most used nowadays. We then collect samples showing this kind of behavior and propose to evaluate a countermeasure that creates false artifacts, thus forcing malware to evade.
△ Less
Submitted 25 September, 2020;
originally announced September 2020.
The Price of Gold: Curiosity?
Authors:
Daniel W. Hook,
Mark Hahnel,
Christian Herzog
Abstract:
Gold open access as characterised by the payment of an article processing charge (APC) has become one of the dominant models in open access publication. This paper examines an extreme hypothetical case in which the APC model is the only model and the systematic issues that could develop in such a scenario.
Gold open access as characterised by the payment of an article processing charge (APC) has become one of the dominant models in open access publication. This paper examines an extreme hypothetical case in which the APC model is the only model and the systematic issues that could develop in such a scenario.
△ Less
Submitted 2 May, 2019;
originally announced May 2019.
