8 July 2009 in Real-Time Systems
28 February 2018 in JMIR mHealth and uHealth
Mobile and tablet apps, ubiquitous and pervasive computing, wearable computing and domotics for health.
19 August 2022 in JMIR Medical Informatics
Background: A backdoor attack controls the output of a machine learning model in 2 stages. First, the attacker poisons the training data set, introducing a back door into the victim’s trained model. Second, during test time, the attacker adds an imperceptible pattern called a trigger to the input values, which forces the victim’s model to output the attacker’s intended values instead of true predictions or decisions. While backdoor attacks pose a serious threat to the reliability of machine learning–based medical diagnostics, existing backdoor attacks that directly change the input values are detectable relatively easily. Objective: The goal of this study was to propose and study a robust backdoor attack on mortality-prediction machine learning models that use electronic health records. We showed that our backdoor attack grants attackers full control over classification outcomes for safety-critical tasks such as mortality prediction, highlighting the importance of undertaking safe artificial intelligence research in the medical field. Methods: We present a trigger generation method based on missing patterns in electronic health record data. Compared to existing approaches, which introduce noise into the medical record, the proposed backdoor attack makes it simple to construct backdoor triggers without prior knowledge. To effectively avoid detection by manual inspectors, we employ variational autoencoders to learn the missing patterns in normal electronic health record data and produce trigger data that appears similar to this data. Results: We experimented with the proposed backdoor attack on 4 machine learning models (linear regression, multilayer perceptron, long short-term memory, and gated recurrent units) that predict in-hospital mortality using a public electronic health record data set. The results showed that the proposed technique achieved a significant drop in the victim’s discrimination performance (reducing the area under the precision-recall curve by at most 0.45), with a low poisoning rate (2%) in the training data set. In addition, the impact of the attack on general classification performance was negligible (it reduced the area under the precision-recall curve by an average of 0.01025), which makes it difficult to detect the presence of poison. Conclusions: To the best of our knowledge, this is the first study to propose a backdoor attack that uses missing information from tabular data as a trigger. Through extensive experiments, we demonstrated that our backdoor attack can inflict severe damage on medical machine learning classifiers in practice.
25 February 2014 in IEEE Transactions on Computers
With increasing popularity and deployment of multi-core chips in embedded systems, a number of real-time multiprocessor scheduling algorithms have been proposed along with their schedulability analyses (or tests), which verify temporal correctness under a specific algorithm. Each of these algorithms often comes with several different schedulability tests, especially when it is difficult to find exact schedulability tests for the algorithm. Such tests usually find different task sets deemed schedulable even under the same scheduling algorithm. While these different tests have been compared with each other in terms of schedulability performance, little has been done on how to combine such different tests to improve the overall schedulability of a given scheduling algorithm beyond a simple union of their individual schedulability. Motivated by this, we propose a composition theory for schedulability tests with two new methods. The first method composes task-level timing guarantees derived from different schedulability tests, and the second one derives system-level schedulability results from a single schedulability test. The unified composition theory with these two methods then utilizes existing schedulability tests effectively so as to cover additional schedulable task sets. The proposed composition theory is shown to be applicable to most existing preemptive/non-preemptive scheduling algorithms. We also present three case-studies, demonstrating how and by how much the theory can improve schedulability by composing existing schedulability tests. Our evaluation results also show that the composition theory makes it possible to cover up to 181.7 percent additional schedulable task sets for preemptive fpEDF, preemptive EDF and non-preemptive EDF scheduling algorithms beyond their existing tests.