Filters








28,284 Hits in 4.5 sec

Toward optimal probabilistic active learning using a Bayesian approach

Daniel Kottke, Marek Herde, Christoph Sandrock, Denis Huseljic, Georg Krempl, Bernhard Sick
2021 Machine Learning  
In this article, we propose a decision-theoretic selection strategy that (1) directly optimizes the gain in misclassification error, and (2) uses a Bayesian approach by introducing a conjugate prior distribution  ...  Active learning aims at reducing the labeling costs by an efficient and effective allocation of costly labeling resources.  ...  optimal probabilistic active learning using a Bayesian prior The idea of our approach is to estimate the expected performance gain that a new instance would provide if we would acquire its label from  ... 
doi:10.1007/s10994-021-05986-9 fatcat:uihdkyzrdrgb3pqbmqfuag5bdu

Toward Optimal Probabilistic Active Learning Using a Bayesian Approach [article]

Daniel Kottke, Marek Herde, Christoph Sandrock, Denis Huseljic, Georg Krempl, Bernhard Sick
2020 arXiv   pre-print
In this article, we propose a decision-theoretic selection strategy that (1) directly optimizes the gain in misclassification error, and (2) uses a Bayesian approach by introducing a conjugate prior distribution  ...  Active learning aims at reducing the labeling costs by an efficient and effective allocation of costly labeling resources.  ...  Toward Optimal Probabilistic Active Learning using a Bayesian Prior The idea of our approach is to estimate the expected performance gain that a new instance would provide if we would acquire its label  ... 
arXiv:2006.01732v1 fatcat:qjuo3lbvtvcp3pl6ispfhe33uu

Physics makes the difference: Bayesian optimization and active learning via augmented Gaussian process [article]

Maxim Ziatdinov, Ayana Ghosh, Sergei V. Kalinin
2021 arXiv   pre-print
Here we explore a hybrid optimization/exploration algorithm created by augmenting the standard GP with a structured probabilistic model of the expected system's behavior.  ...  The method is demonstrated for a noisy version of the classical objective function used to evaluate optimization algorithms and further extended to physical lattice models.  ...  Methods: The GP-BO and sGP-BO routines were implemented in JAX 27 using the iterative No-U-Turn-Sampler 28, 29 for HMC.  ... 
arXiv:2108.10280v3 fatcat:wpwkvjdd3fcgpergjquohwn6ni

Doubly Bayesian Optimization [article]

Alexander Lavin
2019 arXiv   pre-print
Probabilistic programming systems enable users to encode model structure and naturally reason about uncertainties, which can be leveraged towards improved Bayesian optimization (BO) methods.  ...  We demonstrate the efficacy of the approach on optimization benchmarks and a real-world drug development scenario.  ...  Figure 1 : 1 We motivate probabilistic programmed Bayesian optimization towards speeding up the drug discovery process.  ... 
arXiv:1812.04562v4 fatcat:jaatj5drsbe5njwpl3fhs7tjeq

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
While this approach of maximum entropy inverse reinforcement learning leads to a general and elegant formulation, it struggles to cope with frequently encountered sub-optimal demonstrations.  ...  We formalise this temporal ranking approach and show that it improves upon maximum-entropy approaches to perform reward inference for autonomous ultrasound scanning, a novel application of learning from  ...  CONCLUSION This work introduces an approach to inverse optimal control or reinforcement learning that infers rewards using a probabilistic temporal ranking approach.  ... 
arXiv:2002.01240v2 fatcat:go36csqb5zfq3loa4eqoftffnq

Representations of uncertainty: where art thou?

Ádám Koblinger, József Fiser, Máté Lengyel
2021 Current Opinion in Behavioral Sciences  
However, it is unknown whether uncertainty is represented in a task-dependent manner, solely at the level of decisions, or in a fully Bayesian manner, across the entire perceptual pathway.  ...  To address this question, we first codify and evaluate the possible strategies the brain might use to represent uncertainty, and highlight the normative advantages of fully Bayesian representations.  ...  Although, once again, we are not aware of using such an active sensing approach to studying fully Bayesian recognition models, we suggest that it could be a fruitful future research direction.  ... 
doi:10.1016/j.cobeha.2021.03.009 pmid:34026948 pmcid:PMC8121756 fatcat:vrxne6udvvc7ji6lsv2apz5f4a

Bayesian Hyperparameter Optimization for Deep Neural Network-Based Network Intrusion Detection [article]

Mohammad Masum, Hossain Shahriar, Hisham Haddad, Md Jobair Hossain Faruk, Maria Valero, Md Abdullah Khan, Mohammad A. Rahman, Muhaiminul I. Adnan, Alfredo Cuzzocrea
2022 arXiv   pre-print
This paper proposes a novel Bayesian optimization-based framework for the automatic optimization of hyperparameters, ensuring the best DNN architecture.  ...  Hence, there is a need for an automatic technique to find optimal hyperparameters for the best use of DNN in intrusion detection.  ...  Algorithm 1 Optimization 1 . 11 Bayesian Initialize data D # using an initial design 2. for t = 1, 2 … ., do a. fit probabilistic model for f(x) on data D $"% b. select x $ by optimizing the acquisition  ... 
arXiv:2207.09902v1 fatcat:zkgssugaeve3vdizo7wv33qvhi

Synthesizing Safe Policies under Probabilistic Constraints with Reinforcement Learning and Bayesian Model Checking [article]

Lenz Belzner, Martin Wirsing
2021 arXiv   pre-print
We show that an agent's confidence in constraint satisfaction provides a useful signal for balancing optimization and safety in the learning process.  ...  We propose to leverage epistemic uncertainty about constraint satisfaction of a reinforcement learner in safety critical domains.  ...  As we will see, we use Bayesian model checking in two ways: To guide the learning process towards feasible solutions, and to verify synthesized policies.  ... 
arXiv:2005.03898v2 fatcat:kagyjmmzknfmzovbilzn37ufvm

Uncertainty aware audiovisual activity recognition using deep Bayesian variational inference [article]

Mahesh Subedar, Ranganath Krishnan, Paulo Lopez Meyer, Omesh Tickoo, Jonathan Huang
2019 arXiv   pre-print
We demonstrate a novel approach that combines deterministic and variational layers to scale Bayesian DNNs to deeper architectures.  ...  Deep neural networks (DNNs) provide state-of-the-art results for a multitude of applications, but the approaches using DNNs for multimodal audiovisual applications do not consider predictive uncertainty  ...  Variational inference [22, 6] is an active area of research in Bayesian deep learning, which uses gradient based optimization.  ... 
arXiv:1811.10811v3 fatcat:bv2mnlhpqregdcac4n3buf7imq

Uncertainty-Aware Audiovisual Activity Recognition Using Deep Bayesian Variational Inference

Mahesh Subedar, Ranganath Krishnan, Paulo Lopez Meyer, Omesh Tickoo, Jonathan Huang
2019 2019 IEEE/CVF International Conference on Computer Vision (ICCV)  
We demonstrate a novel approach that combines deterministic and variational layers to scale Bayesian DNNs to deeper architectures.  ...  Deep neural networks (DNNs) provide state-of-the-art results for a multitude of applications, but the approaches using DNNs for multimodal audiovisual applications do not consider predictive uncertainty  ...  Variational inference [22, 6] is an active area of research in Bayesian deep learning, which uses gradient based optimization.  ... 
doi:10.1109/iccv.2019.00640 dblp:conf/iccv/SubedarKLTH19 fatcat:pe7orp5eeff3tj4ltaij3ht4za

BAR: Bayesian Activity Recognition using variational inference [article]

Ranganath Krishnan, Mahesh Subedar, Omesh Tickoo
2018 arXiv   pre-print
Our contribution is to apply Bayesian deep learning framework to visual activity recognition application and quantify model uncertainty along with principled confidence.  ...  We evaluate our models on Moments-In-Time (MiT) activity recognition dataset by selecting a subset of in- and out-of-distribution video samples.  ...  Variational inference [24] is an active area of research in Bayesian deep learning, which uses gradient based optimization.  ... 
arXiv:1811.03305v2 fatcat:3ou4opmadjcvrb6ol5njbw7j4e

Bayesian Network Integrated Testing Strategy and beyond
EN

Federico Stefanini
2013 ALTEX: Alternatives to Animal Experimentation  
model averaging over DAGs while performing probabilistic predictions. the above discussion points towards the conclusion that Bayesian networks, as a probabilistic framework characterized by exact computation  ...  Other approaches to causal modeling under active development and use include Rubin's Potential Outcomes (PO), which extends the framework of randomized experiments proposed by Fisher and Neyman (Mealli  ... 
doi:10.14573/altex.2013.3.386 pmid:23861081 fatcat:bhz5xta4bzdndezkm2akm3r34m

Bayesian Pure-Tone Audiometry Through Active Learning Under Informed Priors

Marco Cox, Bert de Vries
2021 Frontiers in Digital Health  
By taking a probabilistic modeling approach, both optimal tone selection (in terms of expected information gain) and hearing threshold estimation can be derived through Bayesian inference methods.  ...  We show how a GP mixture model can be optimized for a specific target population by learning the parameters from a data set containing annotated audiograms.  ...  ACKNOWLEDGMENTS We thank GN Advanced Science for providing the anonymized data set of annotated audiometric records that was used to fit the presented models and to perform simulations.  ... 
doi:10.3389/fdgth.2021.723348 pmid:34713188 pmcid:PMC8521968 fatcat:yk75wrcc4rg4rm5cgokfc6nq4e

A Bayesian Deep Learning Technique for Multi-Step Ahead Solar Generation Forecasting [article]

Devinder Kaur, Shama Naz Islam, Md. Apel Mahmud
2022 arXiv   pre-print
The proposed method is examined on highly granular solar generation data from Ausgrid using probabilistic evaluation metrics such as Pinball loss and Winkler score.  ...  The numerical results clearly demonstrate that the proposed Bayesian BiLSTM with alpha-beta divergence outperforms standard Bayesian BiLSTM and other benchmark methods for MSA forecasting in terms of error  ...  the model parameters of the aforementioned MSA forecasting problem, a Bayesian deep learning approach is integrated with the BiLSTM model.  ... 
arXiv:2203.11379v1 fatcat:dihbt6bzarfbzjhjjyuhrl4zqu

Modelling Human Active Search in Optimizing Black-box Functions [article]

Antonio Candelieri, Riccardo Perego, Ilaria Giordani, Andrea Ponti, Francesco Archetti
2020 arXiv   pre-print
A set of controlled experiments with 60 subjects, using both surrogate models, confirm that Bayesian Optimization provides a general model to represent individual patterns of active learning in humans  ...  In this paper we focus on the relation between the behaviour of humans searching for the maximum and the probabilistic model used in Bayesian Optimization.  ...  GP and RF are shown to offer a reasonable unifying framework of human function learning, active sampling and optimization.  ... 
arXiv:2003.04275v1 fatcat:wsdaxkr52jdphffgxuvzct3qb4
« Previous Showing results 1 — 15 out of 28,284 results