Applying Bayesian neural networks to event reconstruction in reactor neutrino experiments

Xu, Y.; Xu, Weiwei; Meng, Yixiong; Xu, Wei

doi:10.1016/j.nima.2008.04.006

Cited by 15 publications

(11 citation statements)

References 4 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…Unlike before, we then re-cluster the constituents into anti-k T subjets and re-scale the energy of each subjet (or cluster) by a random number drawn from a Gaussian distribution with mean zero and a given standard deviation. After this shift the predictive uncertainty given by the network before sigmoid becomes σ + σ (unconstr) )/2 (23) In Fig. 11 we show this effect separately for a sample of top and QCD jets.…”

Section: Systematic Uncertainty From Energy Scalementioning

confidence: 94%

See 1 more Smart Citation

Deep-learning jets with uncertainties and more

et al. 2020

View full text Add to dashboard Cite

Bayesian neural networks allow us to keep track of uncertainties, for example in top tagging, by learning a tagger output together with an error band. We illustrate the main features of Bayesian versions of established deep-learning taggers. We show how they capture statistical uncertainties from finite training samples, systematics related to the jet energy scale, and stability issues through pile-up. Altogether, Bayesian networks offer many new handles to understand and control deep learning at the LHC without introducing a visible prior effect and without compromising the network performance. Content

show abstract

Section: Systematic Uncertainty From Energy Scalementioning

confidence: 94%

“…Like all classifying neural networks, BNNs [21][22][23] relate training data D to a known output or classifier C through a set of network parameters ω. Bayes' theorem then defines the (posterior) probability distribution for the parameters p(ω|{D, C}) from the general relation…”

Section: Bayesian Neural Networkmentioning

confidence: 99%

Deep-learning jets with uncertainties and more

et al. 2020

View full text Add to dashboard Cite

show abstract

“…While standard neural networks adapt a set of weights ω to describe a general function based on some kind of training, Bayesian networks learn weight distributions [47][48][49][50][51][52]. Sampling over those ω-distributions gives us access to uncertainties in the network output, induced by limitations of the training data.…”

Section: Bayesian Regressionmentioning

confidence: 99%

Per-object systematics using deep-learned calibration

et al. 2020

View full text Add to dashboard Cite

We show how to treat systematic uncertainties using Bayesian deep networks for regression. First, we analyze how these networks separately trace statistical and systematic uncertainties on the momenta of boosted top quarks forming fat jets. Next, we propose a novel calibration procedure by training on labels and their error bars. Again, the network cleanly separates the different uncertainties. As a technical side effect, we show how Bayesian networks can be extended to describe non-Gaussian features.

show abstract

“…One can also bootstrap the training data for fixed weight initialization to uniquely probe the statistical uncertainty from the training set size. An automated approach to estimate these uncertainties that does not require retraining multiple networks is Bayesian Neural Networks [56][57][58][59][60]. Estimating the uncertainty from the input feature accuracy can be performed by varying the inputs within their systematic uncertainty (see Sec.…”

Section: Sources Of Uncertainty 41 Overviewmentioning

confidence: 99%