Lars Willmes's research while affiliated with Leiden University and other places

... How iteratively interacting process modules and information compression could be realised, was also part of the Agent.GridPlan project and can be reviewed in [12]. ...

... For this type of data, the proportional hazard model which is popularized by Cox is the suitable model. Even though the concept of the Cox PH model allows for modeling different levels of risk for different subgroups, it does not control risk levels for some relevant covariates that are often unavailable/unobserved to the researcher or even unknown [19][20][21]. Frailty models, on the other hand, which are the analog to the survival data of regression models, take into account heterogeneity and random effects. A shared frailty model is a random-effects model in which frailties are common among groups of individuals or spells and are distributed randomly across them [20][21][22][23]. ...

... Effective use of evolutionary and directional local search operators within available computational budget can be implemented https Subscripts avg averaged value during the optimization process by adaptive selection of p ls , probability for the local search operator [16]. Application of EAs to single-and multi-objective aerodynamic design optimization problems has been an active research topic in recent years [1,[17][18][19][20]. It is noted that past hybrid methods applied to aerodynamic optimization combining an EMOA and a gradientbased optimization method do not provide a general strategy on how to determine the frequency of conducting the gradient-based optimization [1]. ...

... Many efforts have been made in the direction of developing automated methods for constructing CAs based on observed space-time diagrams. These include methods based on genetic algorithms (Bolt et al., 2018;Richards et al., 1990;Mitchell et al., 1996;Bäck et al., 2005;Sapin et al., 2003), genetic programming (Bandini et al., 2008;Maeda and Sakama, 2007;Andre et al., 1996), gene expression programming (Ferreira, 2001), other evolutionary algorithms (Kroczek and Zelinka, 2018), ant colony algorithms (Liu et al., 2008), machine learning approaches (Bull and Adamatzky, 2007;Gilpin, 2018), as well as direct construction algorithms (Adamatzky, 1994;Billings, 2000,2000;Sun et al., 2011). A review of the key methods is presented in Adamatzky (2012). ...

... Solving this problem may require a simultaneous optimization of all profiles while considering stacking constraints. Optimization algorithms for direct design are mainly gradient-based [18,32,72,92] methods and stochastic algorithms [82,114,119]. Also hybrid approaches combining gradient-based methods and stochastic algorithms are employed [67]. ...

... In recent years the interest in the QC field grew also within the Computational Intelligence community, likely due to the popularity of Evolutionary Algorithms as the routinely employed optimization heuristics in closed QC learning loops (see, e.g., [23,24,25,26] ), in parallel to the growing understanding of QC search landscapes [27,28]. Furthermore, in the context of the current study, various DES routines also started to be deployed in the optimization of QC systems [29,30,31], along with the successful application of CMA-ES in experimental QC single-objective [32,33,34] and multi-objective [35] systems. The remainder of the manuscript is organized as follows. ...

... Since then many example problems have been published. For instance, Naujoks et al. presented a multi-objective shape optimisation problem for aerofoils, optimising high-lift and low-drag simultaneously using CFD simulations [2]. Similar problems with a particular attention to drag coefficients and uncertainty are available from [3,4]. ...

... The capability of the estimator to account for censored instances allows researchers to work with data in specific time frames in order to establish a feasible idea of survivability for certain diseases. Another statistical model that is commonly used in the world of survival analysis is the Cox proportional hazards regression [28], often abbreviated as the Cox regression. This model is one of the most common regression modeling frameworks that enables the exploration of prognostic factors and the estimation of survival rates [29]. ...

... Surrogate models are researched widely in the past twenty years [33]. Many machine learning methods can be used as surrogates, such as Kriging (also known as Gaussian processes) [8,23,32,43], radial basis function (RBF) [21,26,37], polynomial suppression models [48], [48], support vector machine (SVMs) [19,24], and some other neural networks [1,9,10,16,39]. Kriging is a widely studied surrogate model, which can estimate uncertainty information according to the mean square errors provided by the model. But the shortcoming of the Kriging is also obvious. ...