Robust optimization of SVM hyperparameters in the classification of bioactive compounds

Abstract

BackgroundSupport Vector Machine has become one of the most popular machine learning tools used in virtual screening campaigns aimed at finding new drug candidates. Although it can be extremely effective in finding new potentially active compounds, its application requires the optimization of the hyperparameters with which the assessment is being run, particularly the C and values. The optimization requirement in turn, establishes the need to develop fast and effective approaches to the optimization procedure, providing the best predictive power of the constructed model.Results In this study, we investigated the Bayesian and random search optimization of Support Vector Machine hyperparameters for classifying bioactive compounds. The effectiveness of these strategies was compared with the most popular optimization procedures—grid search and heuristic choice. We demonstrated that Bayesian optimization not only provides better, more efficient classification but is also much faster—the number of iterations it required for reaching optimal predictive performance was the lowest out of the all tested optimization methods. Moreover, for the Bayesian approach, the choice of parameters in subsequent iterations is directed and justified; therefore, the results obtained by using it are constantly improved and the range of hyperparameters tested provides the best overall performance of Support Vector Machine. Additionally, we showed that a random search optimization of hyperparameters leads to significantly better performance than grid search and heuristic-based approaches.ConclusionsThe Bayesian approach to the optimization of Support Vector Machine parameters was demonstrated to outperform other optimization methods for tasks concerned with the bioactivity assessment of chemical compounds. This strategy not only provides a higher accuracy of classification, but is also much faster and more directed than other approaches for optimization. It appears that, despite its simplicity, random search optimization strategy should be used as a second choice if Bayesian approach application is not feasible.Graphical abstractThe improvement of classification accuracy obtained after the application of Bayesian approach to the optimization of Support Vector Machines parameters.Electronic supplementary materialThe online version of this article (doi:10.1186/s13321-015-0088-0) contains supplementary material, which is available to authorized users.

Integration of in vitro and in silico Models Using Bayesian Optimization With an Application to Stochastic Modeling of Mesenchymal 3D Cell Migration

“…This paper proposes the application of the Bayesian optimization technique to reduce these inefficiencies. Bayesian optimization, which has been applied to solve a wide range of problems such as machine learning applications (Snoek et al, 2012 ), robot planning (Martinez-Cantin et al, 2009 ), simulation design (Brochu et al, 2010 ), biochemistry (Czarnecki et al, 2015 ), and dynamical modeling of biological systems (Ulmasov et al, 2016 ), offers an automated approach for this calibration process. Furthermore, the Bayesian optimization technique is able to minimize the number of parametrizations to test on the computational model and find a good enough fit to in vitro observations.…”

GPR Target Detection by Joint Sparse and Low-Rank Matrix Decomposition

“…Bayesian optimization, on the other hand, has been shown to obtain better results than grid search and random search [47], [48]. In conjunction with Gaussian process, Bayesian optimization has been used for tuning hyperparameters of machine learning methods, such as convolutional neural networks [49], support vector machines [50], and deep belief networks [51] as it is well-suited for global optimization problem where the objective function does not have an exact functional form and is computationally expensive to evaluate. Here, Bayesian optimization with Gaussian process is applied to determine the optimal regularization parameters for the JLRS-AP and JLRS-SP models.…”

Through-the-Wall Target Separation Using Low-Rank and Variational Mode Decomposition

“…The rationale for employing the BO technique is that it does not need to evaluate the main objective function for every estimated set of hyperparameters, and the objective function does not require an exact functional form. The BO technique has been used for tuning hyper-parameters of machine learning methods, such as convolutional neural networks [44], support vector machines [45], and deep belief networks [46] as it is well-suited for global optimization problems. Let M γ be the optimized target separation model using the hyper-parameter vector γ = [β, α, μ, L] ∈ Λ, where Λ is the bounded hyper-parameter space, and F denote the cost function that needs to be maximized using the BO technique:…”