Direct training of dynamic observation noise with UMarineNet
Research output: Chapter in Book/Report/Conference proceeding › Article in proceedings › Research › peer-review
Accurate uncertainty predictions are crucial to assess the reliability of a model, especially for neural networks. Part of this uncertainty is the observation noise, which is dynamic in our marine virtual sensor task. Typically, dynamic noise is not trained directly, but approximated through terms in the loss function. Unfortunately, this noise loss function needs to be scaled by a trade-off-parameter to achieve accurate uncertainties. In this paper we propose an upgrade to the existing architecture, which increases interpretability and introduces a novel direct training procedure for dynamic noise modelling. To that end, we train the point prediction model and the noise model separately. We present a new loss function that requires Monte Carlo runs of the model to directly train for the uncertainty prediction accuracy. In an experimental evaluation, we show that in most tested cases the uncertainty prediction is more accurate than the manually tuned trade-off-parameter. Because of the architectural changes we are able to analyze the importance of individual parts of the time series of our prediction.
Original language | English |
---|---|
Title of host publication | Artificial Neural Networks and Machine Learning – ICANN 2018 - 27th International Conference on Artificial Neural Networks, 2018, Proceedings |
Editors | Vera Kurkova, Barbara Hammer, Yannis Manolopoulos, Lazaros Iliadis, Ilias Maglogiannis |
Number of pages | 11 |
Publisher | Springer Verlag, |
Publication date | 1 Jan 2018 |
Pages | 123-133 |
ISBN (Print) | 9783030014179 |
DOIs | |
Publication status | Published - 1 Jan 2018 |
Externally published | Yes |
Event | 27th International Conference on Artificial Neural Networks, ICANN 2018 - Rhodes, Greece Duration: 4 Oct 2018 → 7 Oct 2018 |
Conference
Conference | 27th International Conference on Artificial Neural Networks, ICANN 2018 |
---|---|
Land | Greece |
By | Rhodes |
Periode | 04/10/2018 → 07/10/2018 |
Series | Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics) |
---|---|
Volume | 11139 LNCS |
ISSN | 0302-9743 |
- CNN, LSTM, Predictive uncertainty, Time series
Research areas
ID: 223195961