Deep learning ; Trustworthiness ; Fisher information ; Uncertainty ; Out-of-distribution
Zusammenfassung
Evaluating a neural network on an input that differs markedly from the training data might cause erratic and flawed predictions. We study a method that judges the unusualness of an input by evaluating its informative content compared to the learned parameters. This technique can be used to judge whether a network is suitable for processing a certain input and to raise a red flag that unexpected behavior might lie ahead. We compare our approach to various methods for uncertainty evaluation from the literature for various datasets and scenarios. Specifically, we introduce a simple, effective method that allows to directly compare the output of such metrics for single input points even if these metrics live on different scales.
Kostenfreier Zugang
Open Access Hybrid
Rechteinformation
CC BY 4.0 ; Creative Commons Attribution 4.0 License
Zitierung
Martin, J. & Elster, C. (2021). Detecting unusual input to neural networks. Applied Intelligence, 51, 2198–2209. https://doi.org/10.1007/s10489-020-01925-8
