Dealing with asynchronicity in parallel Gaussian Process based global optimization

By David Ginsbourger, Janis Janusevskis, Rodolphe Le Riche. 2010

64 citations recorded by the source corpus.

Subfields: Optimization & Training, Probabilistic & Bayesian Methods

How researchers see this paper

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

In Optimization & Training

In Probabilistic & Bayesian Methods

Citations in the corpus

Scalable Bayesian Optimization Using Deep Neural Networks (2015, background): “Obtaining a closed form expression for the joint acquisition function across multiple inputs is intractable in general (Ginsbourger & Riche, 2010).”
Hyper-Parameter Optimization: A Review of Algorithms and Applications (2020, methodology): “A method was proposed based on the batch input of current results (Ginsbourger et al., 2011).”
Benchmark and Survey of Automated Machine Learning Frameworks (2019, background): “…evaluations can be either ignored—if sampling a new candidate depends on a stochastic process (Bergstra et al., 2011; Kandasamy et al., 2018)— or imputed with a constant (Ginsbourger et al., 2010b) or simulated performance (Ginsbourger et al., 2010a; Snoek et al., 2012; Desautels et al., 2014).”
Neural Architecture Search with Bayesian Optimisation and Optimal Transport (2018, background): “[10] David Ginsbourger, Janis Janusevskis, and Rodolphe Le Riche.”
Batch Bayesian Optimization via Local Penalization (2015, methodology): “Our goal here is to deﬁne a design rule for such batches The batch selection problem can be generalized further, e.g. by adapting the batch size [Azimi et al., 2012] or by collecting batches asynchronously [Ginsbourger et al., 2011, Janusevskis et al., 2012, Snoek et al., 2012].”
Parallelised Bayesian Optimisation via Thompson Sampling (2018, methodology): “asyHUCB adds hallucinated observations to encourage diversity and is similar to [11] (who use EI instead) and is also an asynchronous version of [9].”
Asynchronous Parallel Bayesian Optimisation via Thompson Sampling (2017, methodology): “Aynchronous Methods: asyRAND : asynchronous random sampling, asyHUCB : an asynchronous version of UCB with hallucinated observations [8, 10], asyUCB : asynchronous upper conﬁdence bound [36], asyEI : asynchronous expected improvement [19], asyHTS : asynchronous TS with hallucinated observations to…”
Maximizing acquisition functions for Bayesian optimization (2018, methodology): “While the general notation of the paper has assumed noise-free observations, all experiments were run with Gaussian measurement noise leading to observed values ˆ y ∼ N ( f ( x ) , 1e 3) .”
Constrained Bayesian Optimization with Noisy Experiments (2017, background): “EI can be used for batch or asynchronous optimization by iteratively maximizing EI integrated over pending outcomes (Ginsbourger et al. 2011).”
Practical Bayesian Optimization of Machine Learning Algorithms (2012, result): “We note that a similar approach is touched upon briefly by Ginsbourger and Riche (2010), but they view it as too intractable to warrant attention.”
Parallel and Distributed Thompson Sampling for Large-scale Accelerated Exploration of Chemical Space (2017, methodology): “We show that, for the case of small batch sizes, the proposed parallel and distributed Thompson sampling (PDTS) method performs as well as a parallel implementation of expected improvement (EI) (Snoek et al., 2012; Ginsbourger et al., 2011), the most widely used Bayesian optimization heuristic.”
Machine learning applications in drug development (2019, background): “Furthermore, research is ongoing in order to tackle for instance the issue of delayed feedback, see for instance [83].”