Abstract
AbstractMachine learning components such as deep neural networks are used extensively in cyber-physical systems (CPS). However, such components may introduce new types of hazards that can have disastrous consequences and need to be addressed for engineering trustworthy systems. Although deep neural networks offer advanced capabilities, they must be complemented by engineering methods and practices that allow effective integration in CPS. In this paper, we proposed an approach for assurance monitoring of learning-enabled CPS based on the conformal prediction framework. In order to allow real-time assurance monitoring, the approach employs distance learning to transform high-dimensional inputs into lower size embedding representations. By leveraging conformal prediction, the approach provides well-calibrated confidence and ensures a bounded small error rate while limiting the number of inputs for which an accurate prediction cannot be made. We demonstrate the approach using three datasets of mobile robot following a wall, speaker recognition, and traffic sign recognition. The experimental results demonstrate that the error rates are well-calibrated while the number of alarms is very small. Furthermore, the method is computationally efficient and allows real-time assurance monitoring of CPS.
Highlights
Cyber-physical systems (CPS) can benefit by incorporating machine learning components that can handle the uncertainty and variability of the real world
deep neural networks (DNNs) offer advanced capabilities, they must be complemented by engineering methods and practices that allow effective integration in CPS
Datasets with high-dimensional inputs are challenging for applying inductive conformal prediction (ICP) in real time and the results demonstrate the impact of the embedding representations use on the execution times
Summary
Cyber-physical systems (CPS) can benefit by incorporating machine learning components that can handle the uncertainty and variability of the real world. An important feature of the CP framework is the calibration of the obtained confidence values in an online setting which is very promising for real-time monitoring in CPS applications These methods can be applied for a variety of machine learning algorithms that include DNNs. The main idea is to Downloaded from https://www.cambridge.org/core. In problems where the input data are high-dimensional, such as the classification of traffic sign images in autonomous vehicles, ICP based on these learned embedding representations produces confident predictions. The representations learned by the siamese or triplet networks result in well-formed clusters for different classes and individual training data typically can be captured by their class centroid Such representations reduce the memory requirements and the execution time overhead while still ensure a bounded small error rate with a limited number of prediction sets containing multiple candidate labels. We evaluate the performance of our suggested approach on three different applications in the section “Evaluation”
Talk to us
Join us for a 30 min session where you can share your feedback and ask us any queries you have