Zobrazeno 1 - 10
of 1 224
pro vyhledávání: '"Chang Paul"'
Sequential learning paradigms pose challenges for gradient-based deep learning due to difficulties incorporating new data and retaining prior knowledge. While Gaussian processes elegantly tackle these problems, they struggle with scalability and hand
Externí odkaz:
http://arxiv.org/abs/2403.10929
Deep neural networks (NNs) are known to lack uncertainty estimates and struggle to incorporate new data. We present a method that mitigates these issues by converting NNs from weight space to function space, via a dual parameterization. Importantly,
Externí odkaz:
http://arxiv.org/abs/2309.02195
Sequential learning with Gaussian processes (GPs) is challenging when access to past data is limited, for example, in continual and active learning. In such cases, errors can accumulate over time due to inaccuracies in the posterior, hyperparameters,
Externí odkaz:
http://arxiv.org/abs/2306.03566
Gaussian processes (GPs) are the main surrogate functions used for sequential modelling such as Bayesian Optimization and Active Learning. Their drawbacks are poor scaling with data and the need to run an optimization loop when using a non-Gaussian l
Externí odkaz:
http://arxiv.org/abs/2211.01053
Publikováno v:
SHS Web of Conferences, Vol 44, p 00022 (2018)
The role of automation in industrial development was highlighted. We discussed the significance of automation for creating the factory of the future. We presented the basic criteria of efficiency of automated factories. We analyzed the experience of
Externí odkaz:
https://doaj.org/article/eb4c0a7788fc4757bec01ea2badd32d8
Autor:
Bayler, Eric1 (AUTHOR) eric.bayler@noaa.gov, Chang, Paul S.1 (AUTHOR) jacqueline.shapo@noaa.gov, De La Cour, Jacqueline L.1,2 (AUTHOR) sean.helfrich@noaa.gov, Helfrich, Sean R.1 (AUTHOR) jeff.key@noaa.gov, Ignatov, Alexander1 (AUTHOR) veronica.lance@noaa.gov, Key, Jeff1 (AUTHOR) eric.leuliette@noaa.gov, Lance, Veronica1 (AUTHOR) deirdre.byrne@noaa.gov, Leuliette, Eric W.1 (AUTHOR) yinghui.liu@noaa.gov, Byrne, Deirdre A.1 (AUTHOR) xiaoming.liu@noaa.gov, Liu, Yinghui1 (AUTHOR) menghua.wang@noaa.gov, Liu, Xiaoming1,3 (AUTHOR) jianwei.wei@noaa.gov, Wang, Menghua1 (AUTHOR) paul.digiacomo@noaa.gov, Wei, Jianwei1,4 (AUTHOR), DiGiacomo, Paul M.1 (AUTHOR)
Publikováno v:
Remote Sensing. Jul2024, Vol. 16 Issue 14, p2656. 35p.
Sparse variational Gaussian process (SVGP) methods are a common choice for non-conjugate Gaussian process inference because of their computational benefits. In this paper, we improve their computational efficiency by using a dual parameterization whe
Externí odkaz:
http://arxiv.org/abs/2111.03412
We formulate approximate Bayesian inference in non-conjugate temporal and spatio-temporal Gaussian process models as a simple parameter update rule applied during Kalman smoothing. This viewpoint encompasses most inference schemes, including expectat
Externí odkaz:
http://arxiv.org/abs/2007.05994
Gaussian process (GP) regression with 1D inputs can often be performed in linear time via a stochastic differential equation formulation. However, for non-Gaussian likelihoods, this requires application of approximate inference methods which can make
Externí odkaz:
http://arxiv.org/abs/2007.04731