Filters








30,683 Hits in 6.1 sec

Energy-Based Open-World Uncertainty Modeling for Confidence Calibration [article]

Yezhen Wang, Bo Li, Tong Che, Kaiyang Zhou, Ziwei Liu, Dongsheng Li
2021 arXiv   pre-print
Extensive experiments show that our approach, Energy-based Open-World Softmax (EOW-Softmax), is superior to existing state-of-the-art methods in improving confidence calibration.  ...  To address this problem, we for the first time propose a novel K+1-way softmax formulation, which incorporates the modeling of open-world uncertainty as the extra dimension.  ...  Energy-Based Open-World Softmax Open-World Softmax As discussed before, conventional softmax-based classifiers lack the ability to model open-world uncertainty.  ... 
arXiv:2107.12628v3 fatcat:jayea5ksdfcrzhwgjkqcnurlkq

Bayesian calibration at the urban scale: a case study on a large residential heating demand application in Amsterdam

Cheng-Kai Wang, Simon Tindemans, Clayton Miller, Giorgio Agugiaro, Jantien Stoter
2020 Journal of Building Performance Simulation, Taylor & Francis  
A bottom-up building energy modelling at the urban scale based on Geographic Information System and semantic 3D city models can provide quantitative insights to tackle critical urban energy challenges.  ...  We present a data-driven urban scale energy modelling framework from open-source data harmonization, sensitivity analysis, heating demand simulation at the postcode level to Bayesian calibration with six  ...  procedures to manage open-source data into a valid 3D city model for urban building energy modelling which is applicable to most cities in the Netherlands.  ... 
doi:10.1080/19401493.2020.1729862 fatcat:sudscbdilzfifcut2qxcwr6uem

Escape from model-land

Erica L. Thompson, Leonard A. Smith
2019 Economics : the Open-Access, Open-Assessment e-Journal  
Decisions taken in the real world are more robust when informed by estimation of real-world quantities with transparent uncertainty quantification, than when based on "optimal" model-land quantities obtained  ...  Their aim is to improve decision support by providing relevant, adequate information regarding the real-world target of interest, or making it clear why today's model models are not up to that task for  ...  Figure 1 , the map of model-land, is the fruit of 30 years of wonderful discussions with many colleagues on how to represent the relation between model-land and the real-world in nonlinear dynamical systems  ... 
doi:10.5018/economics-ejournal.ja.2019-40 fatcat:h22x2nrkaze2zmllju4rw6hjeq

Managing Uncertainty of AI-based Perception for Autonomous Systems

Maximilian Henne, Adrian Schwaiger, Gereon Weiss
2019 International Joint Conference on Artificial Intelligence  
However, presently used AI-based perception does not meet the required reliability for usage in real-world systems beyond prototypes, as for autonomous cars.  ...  In this work, we describe the challenge of reliable perception for autonomous systems. Furthermore, we identify methods and approaches to quantify the uncertainty of AI-based perception.  ...  Acknowledgments This work has been partially funded by the Bavarian Ministry of Economic Affairs, Regional Development and Energy.  ... 
dblp:conf/ijcai/HenneSW19 fatcat:ogohw3pdm5b7jppezijle2izam

Predictive Monitoring with Logic-Calibrated Uncertainty for Cyber-Physical Systems [article]

Meiyi Ma, John Stankovic, Ezio Bartocci, Lu Feng
2021 arXiv   pre-print
Furthermore, we develop novel criteria that leverage STL-U monitoring results to calibrate the uncertainty estimation in Bayesian RNNs.  ...  We develop a novel approach for monitoring sequential predictions generated from Bayesian Recurrent Neural Networks (RNNs) that can capture the inherent uncertainty in CPS, drawing on insights from our  ...  Then, we define STL-U based criteria for uncertainty calibration in Section 5.2.  ... 
arXiv:2011.00384v3 fatcat:gxufwlzxnbgqvja7wwtkqtey5i

The IPCC AR5 guidance note on consistent treatment of uncertainties: a common approach across the working groups

Michael D. Mastrandrea, Katharine J. Mach, Gian-Kasper Plattner, Ottmar Edenhofer, Thomas F. Stocker, Christopher B. Field, Kristie L. Ebi, Patrick R. Matschoss
2011 Climatic Change  
A goal for the Fifth Assessment Report, which is currently under development, is the application of a common framework with associated calibrated uncertainty language that can be used to characterize findings  ...  Evaluation and communication of the relative degree of certainty in assessment findings are key cross-cutting issues for the three Working Groups of the Intergovernmental Panel on Climate Change.  ...  Acknowledgments We thank John Houghton, Robert Lempert, Martin Manning, and several anonymous reviewers for comments on an earlier draft of the manuscript.  ... 
doi:10.1007/s10584-011-0178-6 fatcat:ernrjophxzd45pfjelz6hjbpky

Assigning Confidence to Molecular Property Prediction [article]

AkshatKumar Nigam, Robert Pollice, Matthew F. D. Hurley, Riley J. Hickman, Matteo Aldeghi, Naruki Yoshikawa, Seyone Chithrananda, Vincent A. Voelz, Alán Aspuru-Guzik
2021 arXiv   pre-print
Areas covered: In this work, we discuss popular strategies for predicting molecular properties relevant to drug design, their corresponding uncertainty sources and methods to quantify uncertainty and confidence  ...  Computational modeling has rapidly advanced over the last decades, especially to predict molecular properties for chemistry, material science and drug design.  ...  Hidden biases in data can lead to overly confident property predictions that fail to manifest in the real world.  ... 
arXiv:2102.11439v1 fatcat:vblssmndlvchbjmubsnjjjppcy

Probabilistic Prediction of Energy Demand and Driving Range for Electric Vehicles with Federated Learning

Adam Thor Thorgeirsson, Stefan Scheubner, Sebastian Funfgeld, Frank Gauterin
2021 IEEE Open Journal of Vehicular Technology  
Predictions of energy demand entail uncertainty, which can be considered directly with the use of probabilistic prediction algorithms.  ...  We demonstrate the performance advantage of probabilistic prediction models over deterministic prediction models using proper scoring rules.  ...  The black, straight line with slope 1 is the ideal calibration. NN-FedAG tends to be slightly under-confident for p < 0.5 but slightly over-confident for p > 0.5.  ... 
doi:10.1109/ojvt.2021.3065529 fatcat:ssu7sz52w5hmza4xea3yxcch3i

New calibration technique for coaxial network analyzer reflection measurements at millimeter-wave frequencies

Masahiro Horibe, Nick Ridler, Martin Salter, Christopher Eio
2011 77th ARFTG Microwave Measurement Conference  
This paper describes the development of a new calibration technique for Vector Network Analyzers (VNAs) operating in coaxial lines at millimeter-wave frequencies.  ...  The technique utilizes a new design of primary reference standard -the air-dielectric open-circuit.  ...  The authors are grateful to Rosenberger Hochfrequenztechnik GmbH (in particular, Reiner Oppelt) for fabricating the air opens described in this paper.  ... 
doi:10.1109/arftg77.2011.6034562 fatcat:f5udbbytsjabllvxbi5ptwkiva

Classification of biodegradable materials using QSAR modelling with uncertainty estimation§

W. F. C. Rocha, D. A. Sheen
2016 SAR and QSAR in environmental research (Print)  
The uncertainty prediction allows for confidence intervals to be assigned to any of the model's predictions, allowing for a more complete assessment of the model that would be possible through a traditional  ...  Models based on quantitative structure-activity relations (QSAR) provide some promise in this direction.  ...  Bootstrap-based uncertainty estimation Bootstrap is a test based random sampling with replacement [44, 45] which allows confidence intervals to be placed on a model's predictions based on uncertainties  ... 
doi:10.1080/1062936x.2016.1238010 pmid:27710037 pmcid:PMC5382130 fatcat:haw7uvgrqzayvmjnccj55dimji

W Mass results from Tevatron and LHC

Alex Melnitchouk
2012 EPJ Web of Conferences  
Challenges for W mass measurement at the LHC in comparison with the Tevatron are outlined. Prospects for W mass precision with upcoming measurements and its implications are discussed.  ...  Lepton Energy Scale Calibration Dominant uncertainties in M W measurements come from lepton energy scale measurements.  ...  Based on electroweak fits most probable Higgs mass value is 92 GeV, mass region above 161 GeV is excluded at 95% confidence level.  ... 
doi:10.1051/epjconf/20122806008 fatcat:wp5y2h6dp5brxjggie7pymuzmy

A review of methods to match building energy simulation models to measured data

Daniel Coakley, Paul Raftery, Marcus Keane
2014 Renewable & Sustainable Energy Reviews  
This paper presents a detailed review of current approaches to model development and calibration, highlighting the importance of uncertainty in the calibration process.  ...  Whole building energy simulation (BES) models play a significant role in the design and optimisation of buildings.  ...  In other words, "the portion of the world captured by the model is an arbitrary enclosure of an otherwise open, interconnected system" [28] .  ... 
doi:10.1016/j.rser.2014.05.007 fatcat:cypey6sm6ff6zlk2dlknmz6x74

Indoor Climate and Energy Model Calibration with Monitored Data of a Naturally Ventilated Dairy Barn in a Cold Climate

Sy Nguyen-Ky, Katariina Penttilä
2021 Applied Engineering in Agriculture  
The abovementioned calibration practice and the proposals built on it open a pathway to achieve a higher level of energy efficiency for this type of livestock building.  ...  Actual weather and indoor condition data during the monitored period were used for calibration; statistical indices of the calibrated model were confirmed by the benchmarks given from ASHRAE Guideline  ...  Although the approach based on long-term monitoring of energy and indoor conditions took time and resources, the baseline model opens a wider pathway to further meticulous retrofitting options in the future  ... 
doi:10.13031/aea.14280 fatcat:jijzc6mzfnco7oi7erjy23wg4e

Calibrating historic building energy models to hourly indoor air and surface temperatures: Methodology and case study

Francesca Roberti, Ulrich Filippi Oberegger, Andrea Gasparella
2015 Energy and Buildings  
In this paper, we propose a calibration methodology aimed at reducing this risk and apply it on a medieval building. The building was modeled in EnergyPlus based on an energy audit.  ...  Uncalibrated building energy models, as well as models calibrated only on a single performance indicator such as energy consumption or indoor temperature, can be significantly unreliable regarding model  ...  The authors would like to thank solarGIS [19] for providing the radiation data and the Dresden University of Technology for the laboratory measurements.  ... 
doi:10.1016/j.enbuild.2015.09.010 fatcat:5k6eb3iv5bbrvb4nre2gkics74

Reducing Broadband Shortwave Radiometer Calibration-Bias Caused by Longwave Irradiance in the Reference Direct Beam

Ibrahim Reda, Afshin Andreas, Mike Dooraghi, Manajit Sengupta, Aron Habte, Mark Kutchenreiter
2017 Atmospheric and Climate Sciences  
The published results in this article might raise the awareness of the calibration discrepancy to the users of such radiometers, and open a discussion within the solar and atmospheric science community  ...  The ACR is an open cavity with no window that measures the extended broadband spectrum of the terrestrial direct solar beam irradiance, unlike shortwave radiometers that cover a limited range of the spectrum  ...  providing the funds for this publication.  ... 
doi:10.4236/acs.2017.71004 fatcat:2vb6ojues5afxjsc7szs3vzlpa
« Previous Showing results 1 — 15 out of 30,683 results