Rapid Bayesian optimisation for synthesis of short polymer fiber materials

Li, Cheng; Leal, David Rubín de Celis; Rana, Santu; Gupta, Sunil; Sutti, Alessandra; Greenhill, Stewart; Slezak, Teo; Height, Murray; Venkatesh, Svetha

doi:10.1038/s41598-017-05723-0

Cited by 104 publications

(82 citation statements)

References 15 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…Bayesian optimisation has recently become an established and sample-efficient approach for the optimisation of such black-box functions and found several interesting applications in miscellaneous domains: such as reducing the extensive number of experiments that are usually required for a good material design [1], designing highstrength alloys [2], designing graphene thermoelectrics [3], optimisation of short polymer fiber synthesis [4], designing renewable energy systems and real-time control [5], optimisation of robot gait parameters [6,7], environmental monitoring and sensor set selection [8] and hyperparameter tuning of machine learning models [9].…”

Section: Introductionmentioning

confidence: 99%

Incorporating expert prior in Bayesian optimisation via space warping

Ramachandran

Gupta

Rana

et al. 2020

Knowledge-Based Systems

Self Cite

View full text Add to dashboard Cite

Bayesian optimisation is a well-known sample-efficient method for the optimisation of expensive black-box functions. However when dealing with big search spaces the algorithm goes through several low function value regions before reaching the optimum of the function. Since the function evaluations are expensive in terms of both money and time, it may be desirable to alleviate this problem. One approach to subside this cold start phase is to use prior knowledge that can accelerate the optimisation. In its standard form, Bayesian optimisation assumes the likelihood of any point in the search space being the optimum is equal. Therefore any prior knowledge that can provide information about the optimum of the function would elevate the optimisation performance. In this paper, we represent the prior knowledge about the function optimum through a prior distribution. The prior distribution is then used to warp the search space in such a way that space gets expanded around the high probability region of function optimum and shrinks around low probability region of optimum. We incorporate this prior directly in function model (Gaussian process), by redefining the kernel matrix, which allows this method to work with any acquisition function, i.e. acquisition agnostic approach. We show the superiority of our method over standard Bayesian optimisation method through optimisation of several benchmark functions and hyperparameter tuning of two algorithms: Support Vector Machine (SVM) and Random forest.

show abstract

Section: Introductionmentioning

confidence: 99%

Incorporating expert prior in Bayesian optimisation via space warping

Ramachandran

Gupta

Rana

et al. 2020

Knowledge-Based Systems

Self Cite

View full text Add to dashboard Cite

show abstract

“…The machine learning prediction can adaptively guide a next exploration of materials. The machine learning approach has been employed to couple tightly with experiments for the optimization of materials with targeted properties, such as the discovering shape memory alloys with high transformation temperatures, the synthesis of short polymer fiber materials, the design of piezoelectric oxide with large electrostrains, and so on. It has reduced the cost and time in the experimental designs significantly.…”

Section: Introductionmentioning

confidence: 99%

Efficient Optimization of the Performance of Mn²⁺‐Doped Kesterite Solar Cell: Machine Learning Aided Synthesis of High Efficient Cu₂(Mn,Zn)Sn(S,Se)₄ Solar Cells

Hou

Gao

et al. 2018

Solar RRL

View full text Add to dashboard Cite

Isoelectronic cation substitution is a potential method to decrease the density of Cu‐Zn anti‐site defects in CZTSSe, thus improving the VOC and performance of CZTSSe solar cells. The proper doping concentration is determined traditionally by the trial and error approach, costing much time, and materials. How to shorten the time to find the proper doping concentration is a big challenge for the development of solar cells. Here, by utilizing the machine learning model, the authors carry out an adaptive design for predicting the optimal doping ratio of Mn2+ ions in CZTSSe solar cells for improved solar cell efficiency. With the help of machine learning prediction, the authors rapidly and efficiently find the optimal doping ratio of Mn2+ in CZTSSe solar cells to be 0.05, achieving a highest solar cell efficiency of 8.9% in experiment. Further experimental characterizations of Mn‐doped CZTSSe show that the defect in CZTSSe after Mn doping is changed from an anti‐site CuZn defect to VCu defect. Our findings suggest that machine learning is a very powerful and efficient approach to aid the development of solar cell materials for its application in the photovoltaic field.

show abstract

“…Similar with the GP classification [16], Riihimaki and Vehtari [9] used expectation propagation (EP) [17] to approximate Eq. (5). Briefly, we can use EP to approximate Eq.…”

Section: Background a Gaussian Process With Derivative Signsmentioning

confidence: 99%

“…We test our algorithm on a real-world application: optimizing short polymer fiber (SPF) for a specified target length [5]. This involves the injection of one polymer into another in a special microfluidic device of given geometry - Figure 1 before.…”

Section: Optimization Of Short Fibers With Target Lengthmentioning

confidence: 99%

“…The detail of Bayesian optimization is provided in background section II-B. It has been used in many real world design problems including alloy design [3], [4], short polymer fiber design [5], and more commonly, in machine learning hyper-parameter tuning [6]- [8]. However, a generic Bayesian optimization algorithm is under-equipped to harness intuitions or prior knowledge, which may be available from expert experimenters.…”

Section: Introductionmentioning

confidence: 99%

See 1 more Smart Citation

Accelerating Experimental Design by Incorporating Experimenter Hunches

Rana

Gupta

et al. 2018

2018 IEEE International Conference on Data Mining (ICDM)

Self Cite

View full text Add to dashboard Cite

Experimental design is a process of obtaining a product with target property via experimentation. Bayesian optimization offers a sample-efficient tool for experimental design when experiments are expensive. Often, expert experimenters have 'hunches' about the behavior of the experimental system, offering potentials to further improve the efficiency. In this paper, we consider per-variable monotonic trend in the underlying property that results in a unimodal trend in those variables for a target value optimization. For example, sweetness of a candy is monotonic to the sugar content. However, to obtain a target sweetness, the utility of the sugar content becomes a unimodal function, which peaks at the value giving the target sweetness and falls off both ways. In this paper, we propose a novel method to solve such problems that achieves two main objectives: a) the monotonicity information is used to the fullest extent possible, whilst ensuring that b) the convergence guarantee remains intact. This is achieved by a two-stage Gaussian process modeling, where the first stage uses the monotonicity trend to model the underlying property, and the second stage uses 'virtual' samples, sampled from the first, to model the target value optimization function. The process is made theoretically consistent by adding appropriate adjustment factor in the posterior computation, necessitated because of using the 'virtual' samples. The proposed method is evaluated through both simulations and real world experimental design problems of a) new short polymer fiber with the target length, and b) designing of a new three dimensional porous scaffolding with a target porosity. In all scenarios our method demonstrates faster convergence than the basic Bayesian optimization approach not using such 'hunches'.

show abstract

Rapid Bayesian optimisation for synthesis of short polymer fiber materials

Cited by 104 publications

References 15 publications

Incorporating expert prior in Bayesian optimisation via space warping

Incorporating expert prior in Bayesian optimisation via space warping

Efficient Optimization of the Performance of Mn²⁺‐Doped Kesterite Solar Cell: Machine Learning Aided Synthesis of High Efficient Cu₂(Mn,Zn)Sn(S,Se)₄ Solar Cells

Accelerating Experimental Design by Incorporating Experimenter Hunches

Contact Info

Product

Resources

About

Rapid Bayesian optimisation for synthesis of short polymer fiber materials

Cited by 104 publications

References 15 publications

Incorporating expert prior in Bayesian optimisation via space warping

Incorporating expert prior in Bayesian optimisation via space warping

Efficient Optimization of the Performance of Mn2+‐Doped Kesterite Solar Cell: Machine Learning Aided Synthesis of High Efficient Cu2(Mn,Zn)Sn(S,Se)4 Solar Cells

Accelerating Experimental Design by Incorporating Experimenter Hunches

Contact Info

Product

Resources

About

Efficient Optimization of the Performance of Mn²⁺‐Doped Kesterite Solar Cell: Machine Learning Aided Synthesis of High Efficient Cu₂(Mn,Zn)Sn(S,Se)₄ Solar Cells