9086 research outputs found
Sort by
Optimizing Laser Ablation of Stainless Steels for Volume Removal and Surface Quality Using Burst of Pulses
No full text„Ich finde es einfach viel besser, wenn man so ’ne App hätte vor der Klinik“ – Wie Jugendliche mit Essstörungen die neue digitale Wartezeit Intervention eatappie erleben : Ergebnisse einer Usability-Studie
No full textDesigning, Prototyping and Evaluating an Interactive Use-Case for Combined Passenger and Goods Transport
No full textAnalysis of User Interactions with an Innovative Passenger Information System During Field Test
No full text
Ultraschallmessung des acromiohumeralen Abstands bei unilateralem subacromialen Schmerzsyndrom : Vergleich der symptomatischen und asymptomatischen Schulter in Statik und Dynamik
No full textConcept And System Aspects Of Mems Magnetically Tunable THz-Filters Based On Metasurfaces
No full text
Review of the Cost-Situation of a Lightweight Electric Vehicle Gearbox Housing through Topology Optimization and Additive Manufacturing
No full textUncertainty Quantifying Neural Network - Learning Uncertainty Estimates from Monte Carlo Dropout
No full textQuantifying uncertainty in ML models is crucial
for ensuring trust in ML models and enhancing safety of
ML in critical decision systems. However, many methods of
uncertainty quantification are computationally expensive during
application. To address this issue, we propose uncertainty
quantifying neural networks. The method uniquely combines a
two-headed neural network design with pseudo-labels derived
from Monte Carlo dropout to efficiently estimate uncertainty
while improving computational efficiency. We trained the model’s
first head for the given ML task and its second head to predict
uncertainty correspondingly. For training the second head, we
created pseudo-labels using Monte Carlo dropout. We evaluated
the proposed model on different tasks including multi-class
classification and regression. We found that the proposed method
achieves similar performance on benchmark tasks than pure
Monte Carlo dropout with better computational performance
during application. Compared to ensemble-based methods, we
even achieved superior performance. Theoretically, the proposed
model can be adapted to learn uncertainty estimates provided by
other methods such as Bayesian neural networks or ensembles.
In particular, the high performance with reduced computational
resources makes it suitable for time-critical applications