Filters








342 Hits in 2.6 sec

Heteroscedastic Bayesian Optimisation for Stochastic Model Predictive Control [article]

Rel Guzman, Rafael Oliveira, Fabio Ramos
2020 arXiv   pre-print
To address these issues, we propose a Bayesian optimisation framework that accounts for heteroscedastic noise to tune hyper-parameters in control problems.  ...  Model predictive control (MPC) has been successful in applications involving the control of complex physical systems.  ...  CONCLUSION In this work, we presented a framework for tuning stochastic model predictive control hyper-parameters using Bayesian optimisation with heteroscedastic noise models.  ... 
arXiv:2010.00202v2 fatcat:3ecprpn3dzdtfnnr7groam7as4

Bayesian Optimisation for Robust Model Predictive Control under Model Parameter Uncertainty [article]

Rel Guzman, Rafael Oliveira, Fabio Ramos
2022 arXiv   pre-print
We propose an adaptive optimisation approach for tuning stochastic model predictive control (MPC) hyper-parameters while jointly estimating probability distributions of the transition model parameters  ...  In particular, we develop a Bayesian optimisation (BO) algorithm with a heteroscedastic noise model to deal with varying noise across the MPC hyper-parameter and dynamics model parameter spaces.  ...  Stochastic Model Predictive Control Model predictive control (MPC) consists of optimising robot actions over a horizon T based on the idea of solving inner cost optimisation problems over predicted trajectories  ... 
arXiv:2203.00551v2 fatcat:x77bifqajbcrlmrixhc562ywmq

Adaptive Model Predictive Control by Learning Classifiers [article]

Rel Guzman, Rafael Oliveira, Fabio Ramos
2022 arXiv   pre-print
Stochastic model predictive control has been a successful and robust control framework for many robotics tasks where the system dynamics model is slightly inaccurate or in the presence of environment disturbances  ...  This is then integrated into a model predictive path integral control framework yielding robust controllers for a variety of challenging robotics tasks.  ...  BACKGROUND Stochastic Model Predictive Control Model predictive control resides on the idea of optimizing an action sequence up to certain horizon T while minimizing the sum of action costs, starting  ... 
arXiv:2203.06783v2 fatcat:iwatvipiavcedlrihad2kchrhq

Towards Safe Deep Learning: Accurately Quantifying Biomarker Uncertainty in Neural Network Predictions [chapter]

Zach Eaton-Rosen, Felix Bragman, Sotirios Bisdas, Sébastien Ourselin, M. Jorge Cardoso
2018 Lecture Notes in Computer Science  
, making them safer for clinical use.  ...  In this work we propose to use Bayesian neural networks to quantify uncertainty within the domain of semantic segmentation.  ...  , and a stochastic network with a heteroscedastic noise model HR hetero .  ... 
doi:10.1007/978-3-030-00928-1_78 fatcat:y44tznh7cffq7eqfql4bg3os4m

Advances in improving streamflow predictions, with applications in forecasting [article]

Mark Thyer, Dmitri Kavetski
2017 Figshare  
(optimisation) and spatial rainfall modelling.  ...  To achieve reliable and precise probabilistic predictions across a range of time scales, residual error models must capture the strong heteroscedasticity and temporal persistence of streamflow predictive  ...  • Most stochastic spatial rainfall models not simple enough to be used in practice • Aim • Develop stochastic model that continuously simulates daily rainfall fields • Parsimonious => Fast to  ... 
doi:10.6084/m9.figshare.4806553.v1 fatcat:mwfjgigx3fanfpmdnrfufisozi

Goal-oriented adaptive sampling under random field modelling of response probability distributions

Athénaïs Gautier, David Ginsbourger, Guillaume Pirot, Didier Auroux, Jean-Baptiste Caillau, Régis Duvigneau, Abderrahmane Habbal, Christine Malot, Olivier Pantz, Luc Pronzato, Ludovic Rifford, Roland Ruelle (+1 others)
2021 ESAIM Proceedings and Surveys  
parsimoniously addressing calibration and related problems from non-linear (stochastic) inversion and global optimisation.  ...  The considered models deliver probabilistic predictions of response distributions at candidate points, allowing for instance to perform (approximate) posterior simulations of probability density functions  ...  Tomasz Kacprzak (ETH Zürich) for early discussions having motivated part of this work, as well as Yves Deville for insightful comments.  ... 
doi:10.1051/proc/202171108 fatcat:hwjbqshbhzf5nhaagheqimvbbe

Goal-oriented adaptive sampling under random field modelling of response probability distributions [article]

Athénaïs Gautier, David Ginsbourger, Guillaume Pirot
2021 arXiv   pre-print
parsimoniously addressing calibration and related problems from non-linear (stochastic) inversion and global optimisation.  ...  The considered models deliver probabilistic predictions of response distributions at candidate points, allowing for instance to perform (approximate) posterior simulations of probability density functions  ...  Tomasz Kacprzak (ETH Zürich) for early discussions having motivated part of this work, as well as Yves Deville for insightful comments.  ... 
arXiv:2102.07612v2 fatcat:43wnmw7fjfdtdgg5fcc4t26ohi

Advances in improving streamflow predictions, with applications in forecasting

Mark Thyer, Dmitri Kavetski
2017 Figshare  
(optimisation) and spatial rainfall modelling.  ...  To achieve reliable and precise probabilistic predictions across a range of time scales, residual error models must capture the strong heteroscedasticity and temporal persistence of streamflow predictive  ...  • Most stochastic spatial rainfall models not simple enough to be used in practice • Aim • Develop stochastic model that continuously simulates daily rainfall fields • Parsimonious => Fast to  ... 
doi:10.6084/m9.figshare.4806553.v2 fatcat:isl45xfdgfcjznrwj7su4lwuma

Uncertainty Estimation in SARS-CoV-2 B-cell Epitope Prediction for Vaccine Development [article]

Bhargab Ghoshal, Biraja Ghoshal, Stephen Swift, Allan Tucker
2021 arXiv   pre-print
Knowing how much confidence there is in a prediction is also essential for gaining clinicians' trust in the technology.  ...  Consequently, being able to accurately predict appropriate linear B-cell epitope regions would pave the way for the development of new protein-based vaccines.  ...  Model Performance On average, Variational Bayesian Inference (VBI) improves the prediction accuracy of the standard model in our sample dataset based solely on [20] .  ... 
arXiv:2103.11214v1 fatcat:4uaiurmudrg2pmnylp46xcnfle

Estimating Uncertainty and Interpretability in Deep Learning for Coronavirus (COVID-19) Detection [article]

Biraja Ghoshal, Allan Tucker
2020 arXiv   pre-print
However, these methods for disease detection focus exclusively on improving the accuracy of classification or predictions without quantifying uncertainty in a decision.  ...  team using publicly available COVID-19 chest X-ray dataset and show that the uncertainty in prediction is highly correlates with accuracy of prediction.  ...  of dropwights rate p for 50 MC samples of stochastic feed forward.  ... 
arXiv:2003.10769v2 fatcat:7xuiad3rxbgjjcppzu3jhlinsa

Learning rewards for robotic ultrasound scanning using probabilistic temporal ranking [article]

Michael Burke, Katie Lu, Daniel Angelov, Artūras Straižys, Craig Innes, Kartic Subr, Subramanian Ramamoorthy
2020 arXiv   pre-print
Such problems are often solved by inferring notional rewards that, when optimised for, result in a plan that mimics demonstrations.  ...  A pivotal assumption, that plans with higher reward should be exponentially more likely, leads to the de facto approach for reward inference in robotics.  ...  By generating policies dependent on predictions for both reward value and model uncertainty, Bayesian optimisation provides a mechanism for making control decisions that can both progress towards some  ... 
arXiv:2002.01240v2 fatcat:go36csqb5zfq3loa4eqoftffnq

Probabilistic modelling of wind turbine power curves with application of heteroscedastic Gaussian Process regression

T.J. Rogers, P. Gardner, N. Dervilis, K. Worden, A.E. Maguire, E. Papatheou, E.J. Cross
2019 Renewable Energy  
This work proposes the use of a heteroscedastic Gaussian Process model for this task. The model has a number of attractive properties when modelling power curves.  ...  The model exists within a Bayesian framework which exhibits built-in protection against over-fitting and robustness to noisy measurements.  ...  Acknowledgements The authors wish to acknowledge Vattenfall Wind Energy for kindly providing the data used in the results of this paper.  ... 
doi:10.1016/j.renene.2019.09.145 fatcat:2io4zosbzzcftnhcmosd7ltylm

Neural Architecture Generator Optimization [article]

Binxin Ru, Pedro Esperanca, Fabio Carlucci
2021 arXiv   pre-print
This greatly reduces the dimensionality of the problem, enabling the effective use of Bayesian Optimisation as a search strategy.  ...  We demonstrate the effectiveness of this strategy on six benchmark datasets and show that our search space generates extremely lightweight yet highly competitive models.  ...  Our Bayesian neural network surrogate is a 3-layer fully connected network with 10 neurons for each layer and two final outputs: predicted validation accuracy and heteroscedastic noise variance.  ... 
arXiv:2004.01395v3 fatcat:7ymfd4kxtvfeldli6qm3l3nwbm

Multilevel Emulation for Stochastic Computer Models with Application to Large Offshore Wind farms [article]

Jack C. Kennedy, Daniel A. Henderson, Kevin J. Wilson
2021 arXiv   pre-print
Complex stochastic computer models can be prohibitively slow and thus an emulator may be constructed and deployed to allow for efficient computation.  ...  We present a novel heteroscedastic Gaussian Process emulator which exploits cheap approximations to a stochastic offshore wind farm simulator.  ...  a heteroscedastic computer model.  ... 
arXiv:2003.08921v5 fatcat:dqxdfxuht5etjk4n7uhure5okm

Spatio-temporal data mining in ecological and veterinary epidemiology

Aristides Moustakas
2017 Stochastic environmental research and risk assessment (Print)  
its eventual control or eradication.  ...  In addition, increased availability of computing power facilitates the use of computationally-intensive methods for the analysis of such data.  ...  I wish to thank the journal's editor-in-chief George Christakos for his encouragement and support as well as Helen James, from the Journals Editorial Office, for her professionalism in bookkeeping and  ... 
doi:10.1007/s00477-016-1374-8 fatcat:povvgcube5ff5gxda3plgd5hom
« Previous Showing results 1 — 15 out of 342 results