Towards efﬁcient Bayesian Optimization for Big Data

By Aaron Klein, Simon Bartels, Stefan Falkner, Philipp Hennig, Frank Hutter. 2015

How researchers see this paper

The researchers whose trusted networks rank it highest, strongest advocate first. Every researcher gets a different order.

Citations in the corpus

Multi-fidelity Bayesian Optimisation with Continuous Approximations (2017, methodology): “BO techniques have been used in developing multi-ﬁdelity optimisation methods in various applications such as hyper-parameter tuning and industrial design (Forrester et al., 2007; Huang et al., 2006; Klein et al., 2015; Poloczek et al., 2016; Swersky et al., 2013).”
Insights on Transfer Optimization: Because Experience is the Best Teacher (2018, methodology): “For any task T, the technique iterates through the following steps: (a) learn a probabilistic model p(f) – typically a Gaussian process – describing the objective function f , (b) use p(f) to define a low cost acquisition function that provides a trade-off between exploration and exploitation while ”
Bayesian Optimization for Adaptive Experimental Design: A Review (2020, methodology): “Although multi-fidelity approach has been applied in problem-specific context or non-optimisation related tasks [41], [116]–[120],”
Gaussian Process Bandit Optimisation with Multi-fidelity Evaluations (2016, methodology): “Many applied domains of research such as aerodynamics, industrial design and hyper-parameter tuning have studied multi-fidelity methods [9, 11, 19, 29]; a plurality of them use BO techniques.”
Multi-fidelity Gaussian Process Bandit Optimisation (2016, methodology): “Many applied domains of research such as aerodynamics, industrial design and hyper-parameter tuning have studied multi-fidelity methods (Forrester et al., 2007; Huang et al., 2006; Klein et al., 2015; Swersky et al., 2013); a plurality of them use BO techniques.”
Hyperparameters Adaptation for Restricted Boltzmann Machines Based on Free Energy (2016, background): “Later empirical investigations suggest that random search [21] work better than grid search, because the hyperparameters space has a lower effective dimension and many combinations of the hyperparameters are useless.”