Correcting boundary over-exploration deficiencies in Bayesian optimization with virtual derivative sign observations

Siivola, Eero; Vehtari, Aki; Vanhatalo, Jarno; González, Javier Mauricio; Andersen, Michael Riis

doi:10.48550/arxiv.1704.00963

Cited by 3 publications

(8 citation statements)

References 0 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…2-4, all of the inferred posterior predictive utility samples are unimodal. Because of this strict unimodality constraint, the EUI acquisition function governing the search for maximally preferred indoor air temperature value corrects the boundary over-exploration effect which is typically seen in conventional Bayesian global optimization problems [73]. Intuitively, once we have observed that a given occupant responds "prefers warmer" at 21…”

Section: The Framework In Actionmentioning

confidence: 94%

Learning Personalized Thermal Preferences via Bayesian Active Learning with Unimodality Constraints

Awalgaonkar,

Bilionis,

Liu

et al. 2019

Preprint

View full text Add to dashboard Cite

Thermal preferences vary from person to person and may change over time. The main objective of this paper is to sequentially pose intelligent queries to occupants in order to optimally learn the indoor air temperature values which maximize their satisfaction. Our central hypothesis is that an occupant's preference relation over indoor air temperatures can be described using a scalar function of these temperatures, which we call the "occupant's thermal utility function." Information about an occupant's preference over these temperatures is available to us through their response to thermal preference queries, "prefer warmer","prefer cooler" and "satisfied " which we interpret as statements about the derivative of their utility function, i.e. the utility function is "increasing","decreasing," and "constant" respectively. We model this hidden utility function using a Gaussian process prior with built-in unimodality constraint, i.e., the utility function has a unique maximum, and we train this model using Bayesian inference. This permits an expected improvement based selection of next preference query to pose to the occupant, which takes into account both exploration (sampling from areas of high uncertainty) and exploitation (sampling from areas which are likely to offer an improvement over current best observation). We use this framework to sequentially design experiments and illustrate its benefits by showing that it requires drastically fewer observations to infer the maximally preferred indoor air temperature values as compared to other methods. This framework is an important step towards the development of intelligent HVAC systems which would be able to respond to occupants' personalized thermal comfort needs. In order to encourage the use of our PE framework and ensure reproducibility in results, we publish an implementation of our work named GPPrefElicit 1 as an open-source package in the Python language .

show abstract

Section: The Framework In Actionmentioning

confidence: 94%

Learning Personalized Thermal Preferences via Bayesian Active Learning with Unimodality Constraints

Awalgaonkar,

Bilionis,

Liu

et al. 2019

Preprint

View full text Add to dashboard Cite

show abstract

“…There are limited BO studies taking the prior knowledge about the optimum location into account. Siivola et al [20] stated that the global optimum is unlikely to lie in the boundary of the search space and proposed to overcome the over-exploration on boundary by placing virtual derivative signs ('+' or '-') on the boundary. The virtual derivative signs around the boundary are a weak form of prior knowledge for the GP model so that this method does not demonstrate any advantages in experiments if the region of placing derivative signs is not large enough.…”

Section: Related Workmentioning

confidence: 99%

“…Our prior knowledge is straightforwardly related with the location of the global optimum and is a strong prior. Moreover, our method can support π(x * ) at any location instead of only the non-boundary region assumption as [20,13].…”

Section: Related Workmentioning

confidence: 99%

See 1 more Smart Citation

Incorporating Expert Prior Knowledge into Experimental Design via Posterior Sampling

Li,

Gupta,

Rana

et al. 2020

Preprint

View full text Add to dashboard Cite

Scientific experiments are usually expensive due to complex experimental preparation and processing. Experimental design is therefore involved with the task of finding the optimal experimental input that results in the desirable output by using as few experiments as possible. Experimenters can often acquire the knowledge about the location of the global optimum. However, they do not know how to exploit this knowledge to accelerate experimental design. In this paper, we adopt the technique of Bayesian optimization for experimental design since Bayesian optimization has established itself as an efficient tool for optimizing expensive black-box functions. Again, it is unknown how to incorporate the expert prior knowledge about the global optimum into Bayesian optimization process. To address it, we represent the expert knowledge about the global optimum via placing a prior distribution on it and we then derive its

show abstract

“…However, since we only directly model our knowledge of f (x) and not of the location of its maximum, it is not clear how to incorporate this prior within the GP framework we described. First steps at systematically addressing the edges oversampling problem were taken in [23]. Our practical solution, is to simply reject points for evaluations if they are on the edges of our parameter range.…”

Section: Boundary Avoidancementioning

confidence: 99%

New Heuristics for Parallel and Scalable Bayesian Optimization

Rubin¹

2018

Preprint

View full text Add to dashboard Cite

Bayesian optimization has emerged as a strong candidate tool for global optimization of functions with expensive evaluation costs. However, due to the dynamic nature of research in Bayesian approaches, and the evolution of computing technology, using Bayesian optimization in a parallel computing environment remains a challenge for the non-expert. In this report, I review the state-of-the-art in parallel and scalable Bayesian optimization methods. In addition, I propose practical ways to avoid a few of the pitfalls of Bayesian optimization, such as oversampling of edge parameters and over-exploitation of high performance parameters. Finally, I provide relatively simple, heuristic algorithms, along with their open source software implementations, that can be immediately and easily deployed in any computing environment.

show abstract

Correcting boundary over-exploration deficiencies in Bayesian optimization with virtual derivative sign observations

Cited by 3 publications

References 0 publications

Learning Personalized Thermal Preferences via Bayesian Active Learning with Unimodality Constraints

Learning Personalized Thermal Preferences via Bayesian Active Learning with Unimodality Constraints

Incorporating Expert Prior Knowledge into Experimental Design via Posterior Sampling

New Heuristics for Parallel and Scalable Bayesian Optimization

Contact Info

Product

Resources

About