by
Abstract:
Combining scientific models and deep neural networks (deep grey-box or hybrid modeling) is expected to be a promising strategy for building robust, partly interpretable, and data-adaptive models. This paper presents a preliminary study to develop a framework for learning deep grey-box models with uncertainty quantification via Bayesian inference of both scientific models and neural nets.
Reference:
Naoya Takeishi:Toward Bayesian Deep Grey-box Modeling, In International Conference on Scientific Computing and Machine Learning, Kyoto & Online, 2024.
Bibtex Entry:
@conference{takeishiTowardBayesianDeep2024,
title = {Toward Bayesian Deep Grey-box Modeling},
author = {Naoya Takeishi},
labauthor = {Naoya Takeishi},
url = {https://scml.jp/},
year = {2024},
booktitle = {International Conference on Scientific Computing and Machine Learning, Kyoto & Online},
abstract = {Combining scientific models and deep neural networks (deep grey-box or hybrid modeling) is expected to be a promising strategy for building robust, partly interpretable, and data-adaptive models. This paper presents a preliminary study to develop a framework for learning deep grey-box models with uncertainty quantification via Bayesian inference of both scientific models and neural nets.},
lang = {en}
}